Z Os

Front cover
IBM z/OS Parallel Sysplex

Operational Scenarios
Understanding Parallel Sysplex
Handbook for sysplex

management
Operations best practices
Frank Kyne
Peter Cottrell
Christian Deligny
Gavin Foster
Robert Hain
Roger Lowe
Charles MacNiven
Feroni Suhood
ibm.com/redbooks
International Technical Support Organization

IBM z/OS Parallel Sysplex Operational Scenarios
May 2009
SG24-2079-01
Note: Before using this information and the product it supports, read the information in Notices on
page xiii.
Second Edition (May 2009)

This edition applies to Version 1, Release 7 of z/OS (product number 5647-A01) and above.
Copyright International Business Machines Corporation 2009. All rights reserved.

Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule
Contract with IBM Corp.
Contents
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
The team that wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Become a published author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Chapter 1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Introduction to the sysplex environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 What is a sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Functions needed for a shared-everything environment. . . . . . . . . . . . . . . . . . . . . 4
1.2.2 What is a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Sysplex types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Parallel Sysplex test configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Chapter 2. Parallel Sysplex operator commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1 Overview of Parallel Sysplex operator commands . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 XCF and CF commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Determining how many systems are in a Parallel Sysplex . . . . . . . . . . . . . . . . . .
2.2.2 Determining whether systems are active . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.3 Determining what the CFs are called . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.4 Obtaining more information about CF paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.5 Obtaining information about structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.6 Determining which structures are in the CF . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.7 Determining which Couple Data Sets are in use. . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.8 Determining which XCF signalling paths are defined and available . . . . . . . . . . .
2.2.9 Determining whether Automatic Restart Manager is active . . . . . . . . . . . . . . . . .
2.3 JES2 commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1 Determining JES2 checkpoint definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2 Releasing a locked JES2 checkpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.3 JES2 checkpoint reconfiguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4 Controlling consoles in a sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.1 Determining how many consoles are defined in a sysplex . . . . . . . . . . . . . . . . . .
2.4.2 Managing console messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5 GRS commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.1 Determining which systems are in a GRS complex . . . . . . . . . . . . . . . . . . . . . . .
2.5.2 Determining whether any jobs are reserving a device . . . . . . . . . . . . . . . . . . . . .
2.5.3 Determining whether there is resource contention in a sysplex . . . . . . . . . . . . . .
2.5.4 Obtaining contention information about a specific data set. . . . . . . . . . . . . . . . . .
2.6 Commands associated with External Timer References. . . . . . . . . . . . . . . . . . . . . . . .
2.6.1 Obtaining Sysplex Timer status information . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7 Miscellaneous commands and displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7.1 Determining the command prefixes in your sysplex . . . . . . . . . . . . . . . . . . . . . . .
2.7.2 Determining when the last IPL occurred . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7.3 Determining which IODF data set is being used . . . . . . . . . . . . . . . . . . . . . . . . . .
2.8 Routing commands through the sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.9 System symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.10 Monitoring the sysplex through TSO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Contents
13
14
14
14
14
15
16
19
22
23
24
25
25
25
26
27
27
27
28
28
28
29
30
30
31
31
32
33
33
34
34
36
36
iii
iv
Chapter 3. IPLing systems in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1 Introduction to IPLing systems in a Parallel Sysplex. . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 IPL overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1 IPL scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 IPLing the first system image (the last one out) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3.1 IPL procedure for the first system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 IPLing the first system image (not the last one out) . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.1 IPL procedure for the first system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5 IPLing any system after any type of shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5.1 IPL procedure for any additional system in a Parallel Sysplex . . . . . . . . . . . . . . .
3.6 IPL problems in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.1 Maximum number of systems reached . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.2 COUPLExx parmlib member syntax errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.3 No CDS specified . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.4 Wrong CDS names specified . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.5 Mismatching timer references . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.6 Unable to establish XCF connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.7 IPLing the same system name . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.8 Sysplex name mismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6.9 IPL wrong GRS options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
40
40
41
41
41
48
49
50
50
52
52
54
54
55
55
56
57
57
58
Chapter 4. Shutting down z/OS systems in a Parallel Sysplex . . . . . . . . . . . . . . . . . . .

4.1 Introduction to z/OS system shutdown in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . .
4.2 Shutdown overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Overview of Sysplex Failure Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Removing a z/OS system from a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.1 Procedure for a planned shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.2 Procedure for an abnormal stop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4 Running a stand-alone dump on a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.1 SAD required during planned removal of a system. . . . . . . . . . . . . . . . . . . . . . . .
4.4.2 SAD required during unplanned removal of a system with SFM active . . . . . . . .
59
60
60
61
62
63
68
71
71
72
Chapter 5. Sysplex Failure Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1 Introduction to Sysplex Failure Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Status Update Missing condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 XCF signalling failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4 Loss of connectivity to a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5 PR/SM reconfiguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.6 Sympathy sickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7 SFM configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7.1 COUPLExx parameters used by SFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7.2 SFM policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.7.3 Access to the SFM CDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8 Controlling SFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.1 Displaying the SFM couple datasets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.2 Determining whether SFM is active . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.3 Starting and stopping the SFM policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.4 Replacing the primary SFM CDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.8.5 Shutting down systems when SFM is active . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
73
74
74
75
75
76
76
77
77
77
78
79
79
79
81
81
81
Chapter 6. Automatic Restart Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.1 Introduction to Automatic Restart Manager. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 ARM components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Displaying ARM status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83
84
85
88
6.4 ARM policy management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

6.4.1 Starting or changing the ARM policy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.4.2 Displaying the ARM policy status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.5 Defining SDSF as a new ARM element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.5.1 Defining an ARM policy with SDSF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.5.2 Starting SDSF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.5.3 Cancelling SDSF,ARMRESTART with no active ARM policy . . . . . . . . . . . . . . . . 94
6.5.4 Cancelling SDSF,ARMRESTART with active ARM policy . . . . . . . . . . . . . . . . . . 94
6.5.5 ARM restart_attempts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.6 ARM and ARMWRAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.7 Operating with ARM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.7.1 Same system restarts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.7.2 Cross-system restarts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Chapter 7. Coupling Facility considerations in a Parallel Sysplex. . . . . . . . . . . . . . .
7.1 Introduction to the Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2 Overview of the Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Displaying a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3.1 Displaying the logical view of a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . .
7.3.2 Displaying the physical view of a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . .
7.3.3 Displaying Coupling Facility structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3.4 Displaying information about a specific structure . . . . . . . . . . . . . . . . . . . . . . . .
7.3.5 Structure and connection disposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3.6 Displaying connection attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4 Structure duplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4.1 System-managed Coupling Facility (CF) structure duplexing . . . . . . . . . . . . . . .
7.4.2 Rebuild support history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4.3 Difference between user-managed and system-managed rebuild . . . . . . . . . . .
7.4.4 Enabling system-managed CF structure duplexing . . . . . . . . . . . . . . . . . . . . . .
7.4.5 Identifying which structures are duplexed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.5 Structure full monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6 Managing a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6.1 Adding a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6.2 Removing a Coupling Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6.3 Restoring the Coupling Facility to the sysplex . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7 Coupling Facility Control Code (CFCC) commands . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7.1 CFCC display commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7.2 CFCC control commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7.3 CFCC Help commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8 Managing CF structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8.1 Rebuilding structures that support rebuild. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8.2 Stopping structure rebuild . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8.3 Structure rebuild failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8.4 Deleting persistent structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
101
102
102
103
103
104
105
107
109
110
112
113
114
114
115
116
119
119
120
125
136
137
137
143
146
147
147
161
161
163
Chapter 8. Couple Data Set management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8.1 Introduction to Couple Data Set management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 The seven Couple Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.3 Couple Data Set configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.4 How the system knows which CDS to use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5 Managing CDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.1 Displaying CDSs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.5.2 Displaying whether a policy is active. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
165
166
166
167
168
169
169
170
Contents
8.5.3
8.5.4
8.5.5
8.5.6
8.5.7
vi
Starting and stopping a policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Changing the primary CDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IPLing a system with the wrong CDS definition . . . . . . . . . . . . . . . . . . . . . . . . .
Recovering from a CDS failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Concurrent CDS and system failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
171
172
176
177
180
Chapter 9. XCF management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9.1 Introduction to XCF management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 XCF signalling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.1 XCF signalling using CTCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.2 XCF signalling using structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.3 Displaying XCF PATHIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.4 Displaying XCF PATHOUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.5 Displaying XCF PATHIN - CTCs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.6 Displaying XCF PATHOUT - CTCs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.7 Displaying XCF PATHIN - structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.8 Displaying XCF PATHOUT - structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.9 Starting and stopping signalling paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.10 Transport classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2.11 Signalling problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.3 XCF groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.3.1 XCF stalled member detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.4 XCF system monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
183
184
184
185
186
186
187
187
188
188
189
190
192
193
194
197
198
Chapter 10. Managing JES2 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10.1 Introduction to managing JES2 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . .
10.2 JES2 multi-access spool support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3 JES2 checkpoint management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.1 JES2 checkpoint reconfiguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.2 JES2 loss of CF checkpoint reconfiguration . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.3 JES2 checkpoint parmlib mismatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4 JES2 restart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4.1 JES2 cold start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4.2 JES2 warm start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4.3 JES2 hot start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.5 JES2 subsystem shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.5.1 Clean shutdown on any JES2 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . .
10.5.2 Clean shutdown of the last JES2 in a Parallel Sysplex. . . . . . . . . . . . . . . . . . .
10.5.3 Abend shutdown on any JES2 in a Parallel Sysplex MAS . . . . . . . . . . . . . . . .
10.6 JES2 batch management in a MAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.7 JES2 and Workload Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.7.1 WLM batch initiators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.7.2 Displaying batch initiators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.7.3 Controlling WLM batch initiators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.8 JES2 monitor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
201
202
202
204
205
208
213
213
214
216
219
220
220
221
222
223
225
225
225
226
227
Chapter 11. System Display and Search Facility and OPERLOG . . . . . . . . . . . . . . . .

11.1 Introduction to System Display and Search Facility . . . . . . . . . . . . . . . . . . . . . . . . .
11.2 Using the LOG command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2.1 Example of the SYSLOG panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2.2 Example of the OPERLOG panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.3 Using the ULOG command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.3.1 Example of the ULOG panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.4 Using the DISPLAY ACTIVE (DA) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
231
232
233
233
235
235
235
237
11.4.1 Example of the DISPLAY ACTIVE panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11.5 Printing and saving output in SDSF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.1 Print menu. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.2 Print command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5.3 XDC command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.6 Using the STATUS (ST) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.6.1 Using the I action on STATUS panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.7 Resource monitor (RM) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.8 SDSF and MAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.9 Multi-Access Spool (MAS) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.10 Using the JOB CLASS (JC) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.11 Using the SCHEDULING ENVIRONMENT (SE) command . . . . . . . . . . . . . . . . . .
11.12 Using the RESOURCE (RES) command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.13 SDSF and ARM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.14 SDSF and the system IBM Health Checker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.15 Enclaves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.16 SDSF and REXX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
238
239
240
240
242
244
244
246
246
247
248
248
251
252
252
253
255
Chapter 12. IBM z/OS Health Checker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12.1 Introduction to z/OS Health Checker. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.2 Invoking z/OS Health Checker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.3 Checks available for z/OS Health Checker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.4 Working with check output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.5 Useful commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
257
258
258
259
261
267
Chapter 13. Managing JES3 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13.1 Introduction to JES3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.2 JES3 job flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.3 JES3 in a sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.4 Global-only JES3 configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.5 Global-local JES3 single CEC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.6 Global-Local JES3 multiple CEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.7 z/OS system failure actions for JES3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.8 Dynamic system interchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.9 Starting JES3 on the global processor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.10 Starting JES3 on a local processor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.11 JES3 networking with TCP/IP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.11.1 JES3 TCP/IP NJE commands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.12 Useful JES3 operator commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
271
272
272
273
274
275
275
276
276
276
277
277
279
281
Chapter 14. Managing consoles in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . .

14.1 Introduction to managing consoles in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . .
14.2 Console configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.1 Sysplex master console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.2 Extended MCS consoles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.3 SNA MCS consoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.4 Console naming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.5 MSCOPE implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2.6 Console groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.3 Removing a console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.4 Operating z/OS from the HMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.5 Console buffer shortages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.6 Entering z/OS commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.6.1 CMDSYS parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
283
284
285
286
286
287
288
289
290
291
291
295
298
298
Contents
vii
viii
14.6.2 Using the ROUTE command. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14.6.3 Command prefixes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.7 Message Flood Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.8 Removing consoles using IEARELCN or IEARELEC . . . . . . . . . . . . . . . . . . . . . . . .
14.9 z/OS Management Console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
299
300
301
304
304
Chapter 15. z/OS system logger considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15.1 Introduction to z/OS system logger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.1.1 Where system logger stores its data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.2 Starting and stopping the system logger address space . . . . . . . . . . . . . . . . . . . . .
15.3 Displaying system logger status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.4 Listing logstream information using IXCMIAPU . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.5 System logger offload monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.6 System logger ENQ serialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.7 Handling a shortage of system logger directory extents . . . . . . . . . . . . . . . . . . . . . .
15.8 System logger structure rebuilds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.8.1 Operator request . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.8.2 Reaction to failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.9 LOGREC logstream management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.9.1 Displaying LOGREC status. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.9.2 Changing the LOGREC recording medium. . . . . . . . . . . . . . . . . . . . . . . . . . . .
307
308
309
309
310
313
316
317
317
319
319
320
320
320
321
Chapter 16. Network considerations in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . .

16.1 Introduction to network considerations in Parallel Sysplex . . . . . . . . . . . . . . . . . . . .
16.2 Overview of VTAM and Generic Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.2.1 VTAM start options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.2.2 Commands to display information about VTAM GR . . . . . . . . . . . . . . . . . . . . .
16.3 Managing Generic Resources. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.3.1 Determine the status of Generic Resources . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.3.2 Managing CICS Generic Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.3.3 Managing TSO Generic Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.4 Introduction to TCP/IP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.4.1 Useful TCP/IP commands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5 Sysplex Distributor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5.1 Static VIPA and dynamic VIPA overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.6 Load Balancing Advisor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.7 IMS Connect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
323
324
324
327
328
330
330
333
334
336
338
339
340
341
342
Chapter 17. CICS operational considerations in a Parallel Sysplex. . . . . . . . . . . . . .

17.1 Introduction to CICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.2 CICS and Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.3 Multiregion operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4 CICS log and journal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.1 DFHLOG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.2 DFHSHUNT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.3 USRJRNL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.4 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.5 Initiating use of the DFHLOG structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.6 Deallocating the DFHLOG structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.7 Modifying the size of DFHLOG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.8 Moving the DFHLOG structure to another Coupling Facility . . . . . . . . . . . . . . .
17.4.9 Recovering from a Coupling Facility failure. . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.10 Recovering from a system failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5 CICS shared temporary storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
345
346
346
347
348
349
349
350
350
350
350
350
351
352
352
352
17.5.1 Initiating use of a shared TS structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

17.5.2 Deallocating a shared TS structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5.3 Modifying the size of a shared TS structure . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5.4 Moving the shared TS structure to another CF . . . . . . . . . . . . . . . . . . . . . . . . .
17.5.5 Recovery from a CF failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5.6 Recovery from a system failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6 CICS CF data tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6.1 Initiating use of the CFDT structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6.2 Deallocating the CFDT structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6.3 Modifying the size of the CFDT structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6.4 Moving the CFDT structure to another CF . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.6.5 Recovering CFDT after CF failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7 CICS named counter server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7.1 Initiating use of the NCS structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7.2 Deallocating the NCS structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7.3 Modifying the size of the NCS structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7.4 Moving the NCS structure to another CF . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7.5 Recovering NCS after a CF failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.8 CICS and ARM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.9 CICSPlex System Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.10 What is CICSPlex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.10.1 CPSM components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.10.2 Coupling Facility structures for CPSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
352
353
353
354
355
355
355
356
357
357
358
358
359
359
360
360
360
361
362
362
362
363
364
365
366
Chapter 18. DB2 operational considerations in a Parallel Sysplex . . . . . . . . . . . . . .

18.1 Introduction to DB2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.1.1 DB2 and data sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.2 DB2 structure concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3 GBP structure management and recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3.1 Stopping the use of GBP structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3.2 Deallocate all GBP structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4 DB2 GBP user-managed duplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4.1 Preparing for user-managed duplexing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4.2 Initiating user-managed duplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4.3 Checking for successful completion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.5 Stopping DB2 GBP duplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.6 Modifying the GBP structure size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.6.1 Changing the size of a DB2 GBP structure. . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.6.2 Moving GBP structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.6.3 GBP simplex structure recovery after a CF failure . . . . . . . . . . . . . . . . . . . . . .
18.6.4 GBP duplex structure recovery from a CF failure . . . . . . . . . . . . . . . . . . . . . . .
18.7 SCA structure management and recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.1 SCA list structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.2 Allocating the SCA structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.3 Removing the SCA structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.4 Altering the size of a DB2 SCA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.5 Moving the SCA structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.6 SCA over threshold condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.7 SCA recovery from a CF failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.7.8 SCA recovery from a system failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.8 How DB2 and IRLM use the CF for locking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
367
368
368
369
369
370
371
371
372
374
376
380
383
384
385
386
387
388
388
388
388
389
389
389
390
390
390
Contents
ix
18.9 Using DB2 lock structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18.9.1 Deallocating DB2 lock structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.9.2 Altering the size of a DB2 lock structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.9.3 Moving DB2 lock structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.9.4 DB2 lock structures and a CF failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.9.5 Recovering from a system failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.9.6 DB2 restart with Restart Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.10 Automatic Restart Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.11 Entering DB2 commands in a sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
391
391
391
392
392
393
393
394
394
Chapter 19. IMS operational considerations in a Parallel Sysplex . . . . . . . . . . . . . . .

19.1 Introduction to Information Management System . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.1.1 IMS Database Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.1.2 IMS Transaction Manager. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.1.3 Common IMS configurations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.1.4 Support of IMS systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.1.5 IMS database sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.2 IMS system components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.2.1 Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.3 Introduction to IMS in a sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.3.1 Local IMS data sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.3.2 Global IMS data sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.3.3 Global IMS data sharing with shared queues . . . . . . . . . . . . . . . . . . . . . . . . . .
19.4 IMS communication components of an IMSplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.4.1 IMS Connect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.4.2 VTAM Generic Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.4.3 Rapid Network Reconnect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.5 IMS naming conventions used for this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.6 IMS structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.6.1 IMS structure duplexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.6.2 Displaying structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.6.3 Handling Coupling Facility failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.6.4 Rebuilding structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7 IMS use of Automatic Restart Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7.1 Defining ARM policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7.2 ARM and the IMS address spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7.3 ARM and IMS Connect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7.4 ARM in this test example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.7.5 Using the ARM policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.8 IMS operational issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.8.1 IMS commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.8.2 CQS commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.8.3 IRLM commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9 IMS recovery procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.1 Single IMS abend without ARM and without FDR. . . . . . . . . . . . . . . . . . . . . . .
19.9.2 Single IMS abend with ARM but without FDR. . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.3 Single IMS abend with ARM and FDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.4 Single system abend without ARM and without FDR . . . . . . . . . . . . . . . . . . . .
19.9.5 Single system abend with ARM but without FDR . . . . . . . . . . . . . . . . . . . . . . .
19.9.6 Single system abend with ARM and FDR . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.7 Single Coupling Facility failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.8 Dual Coupling Facility failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.9.9 Complete processor failures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
397
398
398
398
398
401
401
402
404
406
406
407
409
411
411
411
412
412
413
415
416
417
420
421
421
421
425
425
425
426
427
428
429
431
431
431
432
433
433
433
434
438
443
19.9.10 Recovering from an IRLM failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

19.10 IMS startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.1 SCI startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.2 RM startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.3 OM startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.4 IRLM startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.5 IMSCTL startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.6 DLISAS startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.7 DBRC startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.8 CQS startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.9 FDBR startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.10.10 IMS Connect startup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11 IMS shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11.1 SCI/RM/OM shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11.2 IRLM shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11.3 IMSCTL shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11.4 CQS shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.11.5 IMS Connect shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19.12 Additional information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
444
446
447
447
447
447
448
449
449
450
451
451
451
452
452
452
452
452
453
Chapter 20. WebSphere MQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20.1 Introduction to WebSphere MQ. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.2 Sysplex considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.3 WebSphere MQ online monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.4 MQ ISPF panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.4.1 WebSphere MQ commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.5 WebSphere MQ structure management and recovery . . . . . . . . . . . . . . . . . . . . . . .
20.5.1 Changing the size of an MQ structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.5.2 Moving a structure from one CF to another . . . . . . . . . . . . . . . . . . . . . . . . . . .
20.5.3 Recovering MQ structures from a CF failure. . . . . . . . . . . . . . . . . . . . . . . . . . .
20.5.4 Recovering from the failure of a connected system . . . . . . . . . . . . . . . . . . . . .
20.6 WebSphere MQ and Automatic Restart Manager. . . . . . . . . . . . . . . . . . . . . . . . . . .
20.6.1 Verifying the successful registry at startup . . . . . . . . . . . . . . . . . . . . . . . . . . . .
455
456
460
461
461
462
464
464
464
465
465
466
466
Chapter 21. Resource Recovery Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21.1 Introduction to Resource Recovery Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.1.1 Functional overview of RRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.2 RRS exploiters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.2.1 Data managers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.2.2 Communication managers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.2.3 Work managers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.3 RRS logstream types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.4 Starting RRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.5 Stopping RRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.6 Displaying the status of RRS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.7 Display RRS logstream status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.8 Display RRS structure name summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.9 Display RRS structure name detail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.10 RRS ISPF panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.11 Staging data sets, duplexing, and volatility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.12 RRS Health Checker definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.13 RRS troubleshooting using batch jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21.14 Defining RRS to Automatic Restart Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
467
468
468
469
469
469
469
469
470
472
472
473
474
475
476
478
479
480
481
Contents
xi
Chapter 22. z/OS UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.2 z/OS UNIX file system structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.2.1 Hierarchical File System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.2.2 Temporary File System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.2.3 Network File System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.2.4 System z File System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.3 z/OS UNIX files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.3.1 Root file system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.3.2 Shared environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22.4 zFS administration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
483
484
484
485
485
485
486
487
487
488
489
Appendix A. Operator commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491

A.1 Operator commands table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
Appendix B. List of structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
B.1 Structures table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500
Appendix C. Stand-alone dump on a Parallel Sysplex example . . . . . . . . . . . . . . . . .
C.1 Reducing SADUMP capture time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.2 Allocating the SADUMP output data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.3 Identifying a DASD output device for SAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.4 Identifying a tape output device for SAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.5 Performing a hardware stop on the z/OS image . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.6 IPLing the SAD program. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.7 Sysplex partitioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.8 Sending a null line on Operating System Messages task . . . . . . . . . . . . . . . . . . . . .
C.9 Specifying the SAD output address . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.10 Confirming the output data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.11 Entering the SAD title . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.12 Dumping real storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.13 Entering additional parameters (if prompted) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.14 Dump complete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
C.15 Information APAR for SAD in a sysplex environment. . . . . . . . . . . . . . . . . . . . . . . .
503
504
504
504
504
505
508
509
510
510
511
512
512
512
513
513
Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Other publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
How to get Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Help from IBM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
517
517
517
519
519
519
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
xii
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area. Any
reference to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in any
manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the
materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.
Copyright IBM Corp. 2009. All rights reserved.
xiii
Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business Machines
Corporation in the United States, other countries, or both. These and other IBM trademarked terms are
marked on their first occurrence in this information with the appropriate symbol ( or ), indicating US
registered or common law trademarks owned by IBM at the time this information was published. Such
trademarks may also be registered or common law trademarks in other countries. A current list of IBM
trademarks is available on the Web at http://www.ibm.com/legal/copytrade.shtml
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
AIX
AS/400
CICSPlex
CICS
DB2
IBM
IMS/ESA
Language Environment
NetView
OMEGAMON
OS/390
Parallel Sysplex
PR/SM
RACF
Redbooks
Redbooks (logo)
Sysplex Timer
System z10
System z
Tivoli
VTAM
WebSphere
z/OS
z/VM
zSeries
The following terms are trademarks of other companies:

Java, RSM, ZFS, and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United
States, other countries, or both.
Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States, other
countries, or both.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel
SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its
subsidiaries in the United States and other countries.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
xiv
Preface
This IBM Redbooks publication is a major update to the Parallel Sysplex Operational
Scenarios book, originally published in 1997.
The book is intended for operators and system programmers, and is intended to provide an
understanding of Parallel Sysplex operations. This understanding, together with the
examples provided in this book, will help you effectively manage a Parallel Sysplex and
maximize its availability and effectiveness.
The book has been updated to reflect the latest sysplex technologies and current
recommendations, based on the experiences of many sysplex customers over the last 10
years.
It is our hope that readers will find this to be a useful handbook for day-to-day sysplex
operation, providing you with the understanding and confidence to expand your exploitation
of the many capabilities of a Parallel Sysplex.
Knowledge of single-system z/OS operations is assumed. This book does not go into
detailed recovery scenarios for IBM subsystem components, such as CICS Transaction
Server, DB2 or IMS. These are covered in great depth in other Redbooks publications.
The team that wrote this book

This book was produced by a team of specialists from around the world working at the
International Technical Support Organization Poughkeepsie Center and the Australian
Development Lab, Gold Coast Center.
Frank Kyne is a Senior Consulting IT Specialist at the International Technical Support
Organization (ITSO), Poughkeepsie, NY. He is responsible for ITSO projects related to
Parallel Sysplex and High Availability. Frank joined IBM in 1985 as an MVS Systems
Programmer in the IBM software lab in Ireland. Since joining the ITSO in 1998, he has been
responsible for IBM Redbooks projects and workshops related to Parallel Sysplex, High
Availability, and Performance.
Peter Cottrell is a Senior z/OS Technical Specialist in IBM Australia. He has more than 20
years of experience in mainframe operating systems. His areas of expertise include the
implementation and configuration of the z/OS operating system, Parallel Sysplex, z/OS
storage, and z/OS security. Peter holds a Masters degree in Information Technology from the
University of Canberra.
Christian Deligny is a Senior Systems Operator at the IBM data center in Sydney, Australia,
supporting both IBM Asia Pacific and external clients. He has more than 25 years of
experience in operations on a variety of platforms, including OS/390 and z/OS for the last
10 years. Chris specializes in change control, operational procedures, and operations
documentation.
Gavin Foster is a z/OS Technical Consultant in IBM Australia. He has 22 years of experience
in the mainframe operating systems field. His areas of expertise include systems
programming and consulting on system design, upgrade strategies, platform deployment and
Parallel Sysplex. Gavin coauthored the IBM Redbooks publication Merging Systems into a
Sysplex, SG24-6818.
xv
Robert Hain is an IMS Systems Programmer in IBM Australia, based in Melbourne. He has
23 years of experience in the mainframe operating systems field, specializing for the past 20
in IMS. His areas of expertise include the implementation, configuration, management, and
support of IMS systems. He is also a member of the IMS worldwide advocate team, part of
the IMS development labs in San Jose, California. Robert coauthored a number of IBM
Redbooks publications about IMS, as well as the IBM Press publication An Introduction to
IMS.
Roger Lowe is a Senior Technical Consultant in the Professional Services division of
Independent Systems Integrators, an IBM Large Systems Business Partner in Australia. He
has 23 years of experience in the operating systems and mainframe field. His areas of
expertise include the implementation and configuration of the z/OS operating system and
Parallel Sysplex. Roger coauthored the IBM Redbooks publication Merging Systems into a
Sysplex, SG24-6818.
Charles MacNiven is a z/OS System Programmer in IBM Australia. Charles has more than
21 years of experience with working with customers in large mainframe environments in
Europe and Australia. His areas of expertise include the implementation, configuration, and
support of the z/OS operating system, DB2, and CICS.
Feroni Suhood is a Senior Performance Analyst in IBM Australia. He has 25 years of
experience in the mainframe operating systems field. His areas of expertise include Parallel
Sysplex, performance, and hardware evaluation. Feroni coauthored the IBM Redbooks
publication Merging Systems into a Sysplex, SG24-6818.
Thanks also to those responsible for the original version of this book:
David Clitherow
IBM UK
Fatima Cavichione
IBM Brazil
Howard Charter
IBM UK
Jim Ground
IBM US
Brad Habbershaw
IBM Canada
Thomas Hauge
DMData, Denmark
Simon Kemp
IBM UK
Marcos Roberto de Lara
IBM Portugal
Wee Heong Ng
IBM Singapore
Vicente Ranieri Junior
IBM Brazil
xvi
Thanks to the following people for their invaluable contributions and support to this project:
Bob Haimowitz
International Technical Support Organization, Poughkeepsie Center
Carol Woodhouse
Australian Development Lab, Gold Coast Center
Become a published author

Join us for a two- to six-week residency program! Help write a book dealing with specific
products or solutions, while getting hands-on experience with leading-edge technologies. You
will have the opportunity to team with IBM technical professionals, Business Partners, and
Clients.
We want our books to be as helpful as possible. Please send us your comments about this or
other books in one of the following ways:
Use the electronic evaluation form found on the Redbooks Web sites:
For Internet users
http://www.redbooks.ibm.com/
For IBM intranet users
http://w3.itso.ibm.com/
Send us a note at the following address:
[email protected]
Comments welcome
Your comments are important to us!
We want our books to be as helpful as possible. Send us your comments about this book or
other IBM Redbooks publications in one of the following ways:
Use the online Contact us review Redbooks form found at:
ibm.com/redbooks
Send your comments in an e-mail to:
[email protected]
Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
Preface
xvii
xviii
Chapter 1.
Introduction
This chapter explains the structure of this book and introduces the concepts and principles of
a sysplex environment. It highlights the main components in a Parallel Sysplex environment
and touches on the following topics:
The difference between a base sysplex and a Parallel Sysplex
The functions of the hardware and software components that operators encounter in a
sysplex environment
The test Parallel Sysplex used for the examples in this document
1.1 Introduction to the sysplex environment

This book gives operators and system programmers a better understanding of what a sysplex
is, how it works, and the operational considerations that are unique to a sysplex environment.
All the products that run in a sysplex environment also work in a non-sysplex environment.
However, there are additional functions, or changed behaviors, that are specific to sysplex.
This book helps you to exploit those functions to achieve better availability and easier system
management in a sysplex environment.
In addition to discussing how to operate a sysplex, the book provides you with background
and positioning information. For example, to understand the importance of Sysplex Failure
Management and how to control it, you first must understand why it is especially important to
react quickly when a member of a sysplex fails.
The book begins by describing, at a high level, what constitutes a sysplex. It gives an
overview of the major components that play an important role in a sysplex environment. Then
it briefly describes some of the more common sysplex-related commands. These commands
can help you to build a picture of your sysplex.
Next, the book explains how to IPL a system into a sysplex and how to remove a system from
a sysplex, discussing considerations that only apply to a sysplex. The remainder of the book
provides more detail about the major components and subsystems that you will be interacting
with in a sysplex, and discusses the additional functions, messages, and commands that only
apply to a sysplex environment.
1.2 What is a sysplex

A sysplex (or SYStems comPLEX) consists of 1 to 32 z/OS systems integrated into one
multisystem environment (somewhat like a cluster in the UNIX world). To be a member of a
sysplex, all the participating systems must share a common time source and a common set of
data sets (called Couple Data Sets). They must also be able to communicate with each other
over a set of links called cross-system coupling facility (XCF) signalling paths.
The individual z/OS systems communicate and cooperate through a set of multisystem
software and hardware components to process work as a single entity. When individual z/OS
systems are integrated into one sysplex, it allows for greater application availability, easier
system management, and improved scalability.
Base sysplex versus Parallel Sysplex
A base sysplex is a group of z/OS systems integrated into a multisystem environment.
A Parallel Sysplex is a base sysplex, with the addition of a specialized component
called a Coupling Facility. The Coupling Facility enables many functions in a Parallel
Sysplex that are not available in a base sysplex.
This book concentrates on operations in a Parallel Sysplex environment, so any reference
to sysplex is referring to a Parallel Sysplex.
Of the many challenges imposed on IT departments today, the business requirement for
applications to be always available is probably the most common and perhaps the most
challenging. This requirement ignores the need to shut down systems and subsystems from
time to time for changes or scheduled maintenance. So how do you perform the impossible:
keeping your applications available while at the same time maintaining your systems?
The only way to do this is to have at least two copies of all the components that deliver the
application servicethat is, two z/OS systems, two database manager instances (both being
able to update the same database), two sets of CICS regions that run the same applications,
and so on. Parallel Sysplex provides the infrastructure to deliver this capability by letting you
share databases across systems, and enabling you to automatically route work to the most
appropriate system. Figure 1-1 shows the major components of a sysplex that contains two
systems.
CF2
XCF2
CF1
XCF1
VTAM/TCP
CICS TOR
DB2 IMS
CICS AOR
CICS AOR
CICS AOR
CICS AOR
VTAM/TCP
CICS TOR
DB2 IMS
Cons
XES
XCF
WLM
Logger
SFM
ARM
z/OS
Chan
Ext.
Switch
Primary Alternate
Sysplex CFRM
CDS
CDS
Network
Sysplex
Timer
Sysplex Timer
XCF
CICS AOR
CICS AOR
CICS AOR
CICS AOR
Network
Cons
XES
XCF
WLM
Logger
SFM
ARM
z/OS
DWDM
Switch
Primary Alternate
CFRM Sysplex
CDS
CDS
Figure 1-1 Components of a Parallel Sysplex
Having multiple copies (known as clones) of your production environment allows your
applications to continue to run on other systems if you should experience a planned or
unplanned outage of one of the systems, thereby masking the outage from the application
users. Also, you have the ability to restart the impacted subsystems on another system in the
sysplex, pending the recovery of the failed system. When this failure and restart management
is called for it can be initiated automatically, based on policies you define for the sysplex.
Being able to run multiple instances of a subsystem using the same data across multiple
z/OS systems also makes it possible to process more transactions than would be possible
with a single-system approach (except, of course, in the unlikely case where all instances
need to update exactly the same records at the same time). The transaction programs do not
need to be rewritten, because it is the database managers that transparently provide the data
sharing capability.
Chapter 1. Introduction
There are also value-for-money advantages that you can realize from exploiting the sysplex
capabilities. Imagine you have two processors, and one has 75 MIPS of unused capacity and
the other has 50 MIPS. Also imagine that you want to add a new application that requires
100 MIPS.
If the application supports data sharing, you can divide it up and run some transactions on
one system and some on the other, thereby fully exploiting the unused capacity. On the other
hand, if the workload does not support data sharing, you must run all 100 MIPS of work in the
same system, meaning that you must purchase an upgrade for one of the two processors.
Additionally, if your work can run on any system in the sysplex, and you need more capacity,
you have the flexibility to add capacity to any of the current processors, or even to add
another processor to the sysplex, whichever is the most cost-effective option.
It may also be possible to break up large database queries into smaller parts and run those
parts in parallel across the members of the sysplex, resulting in significantly reduced elapsed
times for these transactions.
1.2.1 Functions needed for a shared-everything environment

Imagine you are given the job of designing a completely new operating system, and are given
the following design points:
The system must provide the capability to deliver near-continuous application availability.
This effectively means that you must have multiple cooperating instances in order to
remove single points of failure.
The system must provide the ability to share databases at the record level across multiple
instances of the database manager.
It should be possible to manage and administer the system (or systems) with minimal
duplication of effort.
The system must accomplish all this as efficiently as possible.
Given these challenging requirements, what functions would you need to code into your
operating system?
Common time
The first thing you will need is an ability to have every system use exactly the same time. Why
is this needed? Consider what happens when a database manager updates a database. For
every update, a log record is written containing a copy of the record before the update (so
failed updates can be backed out) and a copy of the record after the update (so updates can
be reapplied if the database needs to be recovered from a backup).
If there is only a single database manager updating the database, all the log records will be
created in the correct sequence, and the time stamps in the log records will be consistent with
each other. So, if you need to recover a database, you would restore the backup, then apply
all the updates using the log records from the time of the backup through to the time of the
failure.
But what happens if two or more database managers are updating the database? If you need
to recover the database, you would again restore it from the backup, then merge the log files
(in time sequence) and apply the log records again. Because the log records contain the after
image for each update, it is vital that the updates are applied in the correct sequence. This
means that both database managers must have their clocks synchronized, to ensure that the
time stamps in each log record are consistent, regardless of which database manager
instance created them.
4
In a sysplex environment, the need to have a consistent time across all the members of the
sysplex is addressed by attaching all the processors in the sysplex to a Sysplex Timer, or by
its replacement, Server Time Protocol (STP). Note that the objective of having a common
time source is not to have a more accurate time, but rather to have the same time across all
members of the sysplex. For more information about Sysplex Timers and STP, see 2.6,
Commands associated with External Timer References on page 31.
Buffer coherency
Probably the most common way to improve the performance of a database manager is to
give it more buffers. Having more buffers means that it can keep copies of more data records
in processor storage, thus avoiding the delay associated with having to read the data from
disk.
In a data sharing environment you will have multiple database managers, each with its own
set of buffers. It is likely that some data records will be contained in the buffers of more than
one database manager instance. This does not cause any issues as long as all the database
managers are only reading the data. But, what happens if database manager DB2A updates
data that is currently in the buffers of database manager DB2B? If there is no mechanism for
telling DB2B that its copy is outdated, then that old record could be passed to a transaction
which treats that data as current.
Therefore, when you have multiple database manager instances, all with update access to a
shared database, you need some mechanism that the database managers can use to
determine whether a record in their buffer has been updated elsewhere. One way to address
this would be for every instance to tell all the other instances every time it adds or removes a
record to its buffers. But this would generate tremendous overhead, especially as the number
of instances in the sysplex increases.
The solution that is implemented in a Parallel Sysplex is for each database manager to tell
the Coupling Facility (CF) every time it adds a record to its local buffer. The CF then knows
which instances have a copy of any given piece of data. Each instance also tells the CF every
time it updates one of those records. Because the CF knows who has a copy of each record,
it also knows who it has to tell when a given record is updated. This process is called
Cross-Invalidation, and it is handled automatically by the database managers and the CF.
Serialization
Because you can have multiple database manager instances all able to update any data in
the database, you may be wondering how to avoid having two instances make two different
updates to the same piece of data at the same time.
Again, one way to achieve this could be for every instance to talk to all the other instances to
ensure that no one else is updating a piece of data that it is about to update. However, this
would be quite inefficient, especially if there are many instances with access to the shared
database.
In a Parallel Sysplex, this requirement for serializing data access is achieved by using a lock
structure in the CF. Basically, every time a database manager instance wants to work with a
record (either to read it or to update it), it sends a lock request to the CF, identifying the record
in question and the type of access requested. Because the CF has knowledge of all the lock
requests, it knows what types of accesses are in progress for that record.
If the request is for shared access, and no one else has exclusive access, the CF grants the
request. Or if the request is for exclusive access, and no one else is accessing the record at
this time, the request is granted.
But if the type of serialized access needed by this request is not compatible with an instance
that is already accessing the record, the CF denies the request and identifies the current
owner of that data (who is doing the exclusive access).1 When the current update completes,
access will be granted to the next database manager in the queue, allowing it to make its
update.
Monitoring
If you are going to be able to run your work on any of the systems in the sysplex, then you will
probably want some way for products that provide a service to be aware of the status of their
peers on other systems. Of course, you could do this by having all the peers constantly
talking to each other to ensure they are still alive. But this would waste a lot of resource, with
all these programs talking back and forth to each other all the time, and only discovering a
failure a tiny fraction of the time.
A more efficient alternative is for the products to register their presence with the system, and
ask the system to inform them if one of the peer instances disappears. Because the system is
aware any time an address space is started or ends, it automatically knows if any of the peers
stop. As a result, it is much more efficient to have the system monitor for the availability of the
peer members, and to inform the remaining address spaces should one of them go away.
The system component that provides this service is called Cross-System Coupling Facility
(XCF).
Building on top of this concept, you also have the ability to monitor complete systems. Every
few seconds, every system updates a data set called the Sysplex Couple Data Set with its
current time stamp. At the same time, it checks the time stamp of all the other members of the
sysplex. If it finds that a system has not updated its time stamp in a certain interval (known as
the Failure Detection Interval), it can inform the operator that the system in question appears
to have failed. It can even automatically remove that failed system from the sysplex using a
function known as Sysplex Failure Management.
Communication within the sysplex

If you have peer programs providing services within the sysplex, it is probable that the
programs will want to communicate with each other, perhaps to share work or exchange
status information.
One way to achieve this would be to have the programs use services such as VTAM or
TCP. However, this would mean that if either of these services were unavailable for some
reason, then all the programs would be unable to communicate with each other. Another
option would be for the programs to communicate directly, using dedicated devices such as
Channel-to-Channel adapters (CTCs). This would eliminate the dependency on TCP or
VTAM, but it involves complex programming. The other option is for the operating system to
provide an easy-to-use service to communicate between programs within the same sysplex.
This service, called XCF Signalling Services, is provided by XCF and is used by many IBM
and non-IBM system components.
Workload distribution
We have now discussed how you have the ability to run work (transactions and batch jobs) on
more than one system in the sysplex, all accessing the same data. And the programs that
provide services are able to communicate with each other using XCF. This means that any
work you want to run can potentially run anywhere in the sysplex. And if one of the systems is
unavailable for some reason, the work can be processed on one of the other systems.
This description is not strictly accurate, but it is sufficient in the context of this discussion.
However, to derive the maximum benefit from this, you need two other things:
The ability to present a single system image to the user, so that if one system is down, the
user can still log on in the normal way, completely unaware of the fact that one of the
systems is down.
The ability to send incoming work requests to whichever system is best able to service
that work. The decision about which is the best system might be based on the response
times being delivered by the different systems in the sysplex, or on which system has the
most spare capacity.
Both of these capabilities are provided in a sysplex. Both VTAM and TCP provide the ability
for multiple transaction manager instances (CICS or IMS, for example) to use the same name
and have work routed to one of those instances. For example, you might have four CICS
Terminal Owning Regions that call themselves CICSPROD. When the users want to use this
service, they would logon to CICSPROD. Even if three of the four regions were down, the
user would still be able to logon to the fourth region, unaware that the other three regions are
currently down.
This capability to have multiple work managers use the same name can then be combined
with support in a system component called the Workload Manager (WLM). WLM is
responsible for assigning sysplex resources to work items in accordance with
installation-specified objectives. WLM works together with VTAM and TCP and the
transaction managers to decide which is the most appropriate system for each piece of work
to run on. This achieves the objectives of helping the work achieve its performance targets,
masking planned or unplanned outages from users, and also making full use of capacity
wherever it might be available in the sysplex.
Other useful services

In addition to all the core services listed, there are a number of other services that are
included as part of the operating system, but whose use is optional.
System Logger
If you can run work anywhere in the sysplex, what other services would be useful? Many
system services create logs; syslog is probably the one you are most familiar with. z/OS
contains a system component called System Logger that provides a single syslog which
combines the information from all the systems. This avoids you having to look at multiple logs
and merge the information yourself. The exploiters you are probably most familiar with are
OPERLOG (for sysplex-wide syslog) and LOGREC (for sysplex-wide error information).
Other users of System Logger are CICS (for its log files), IMS (when using Shared Message
Queue), RRS, z/OS Health Checker, and others.
Automatic Restart Manager

Given that you have the ability to run anything anywhere in the sysplex, it would be useful if
there was a way to quickly restart critical subsystems after a system failure. And it would be
even more useful if that mechanism could take into account how much spare capacity is
available in the remaining systems. After all, if you have a DB2 subsystem that needs
200 MIPS, it would be better to restart that on a system that has 300 MIPS available, rather
than one that only has 50 MIPS.
z/OS includes a component called Automatic Restart Manager (ARM) that has the ability to
not only restart failed address spaces (on the same or a different system), but also to work
with the z/OS Workload Manager to determine which is the most suitable system to restart a
given address space on. The installation can decide whether or not to use this function. If it
does so, it can control various aspects of what is restarted and how. This information is stored
in the ARM policy in the ARM Couple Data Set.
The Automatic Restart Manager is discussed in more detail in Chapter 6, Automatic Restart
Manager on page 83.
CICSPlex
There are a number of aspects to a CICS service. For example, there is the function that
manages interactions with the users terminal. There is the part that runs the actual
application code (such as reading a database and obtaining account balances). And there
may be other specialized functions, like providing access to in-storage databases. In the past,
it was possible that an error in application code could crash a CICS region, impacting all the
users logged on to that region.
To provide a more resilient environment, CICS provides the ability to run each of these
functions in a different region. For example, the code that manages interactions with the
users terminal tends to be very stable, so setting up a CICS region that only provides this
function (known as a Terminal Owning Region, or TOR) results in a very reliable service. And
by providing many regions that run the application code (called an Application Owning
Region, or AOR), if one region abends (or is stopped to make a change), other regions are
still available to process subsequent transactions. Running your CICS regions like this is
known as Multi Region Option (MRO). MRO can be used in both a single system environment
or in a sysplex.
When used in a sysplex, MRO is often combined with a CICS component called CICSPlex
System Manager. CICSPlex System Manager provides a single point of control for all the
CICS regions in the sysplex. It also provides the ability to control which AOR a given
transaction should be routed to.
Global Resource Serialization

In a multi-tasking, multi-processing environment, resource serialization is the technique used
to coordinate access to resources that are used by more than one program.
Global Resource Serialization (GRS) is the component of z/OS that provides the control
needed to ensure the integrity of resources in a multisystem environment. All the members of
a given sysplex must be in the same GRS complex, so the access to the resources shared by
the members of the sysplex are controlled.
1.2.2 What is a Coupling Facility

A Coupling Facility (CF) can be viewed as very high speed shared storage with an intelligent
front-end. Rather than being concerned about what is where in the CF, exploiters of
Coupling Facility services request it to carry out some function on their behalf. For example, if
program FREDA wants to send a message to its peer FREDB on another system, FREDA
issues a CF command requesting the CF to store the message and inform FREDB that there
is a message awaiting collection.
There are three basic types of services that the CF can provide:
Lock services, used for serializing access to some resource
Exploiters of lock services include DB2, IMS, CICS/VSAM RLS, and GRS.
Cache services, used for keeping track of who has an in-storage copy of what data within
the sysplex; can also be used to provide high performance access to shared data
Exploiters of cache services include DB2, RACF, CICS/VSAM RLS, and IMS.
List services, used for passing information between systems, organizing work queues, or
storing log data
Exploiters of list services include JES2, VTAM, and XCF.
8
Storage in the CF is assigned to entities called structures. The type of services that can be
provided in association with a given structure is dependent on the structure type. In normal
operation, you do not need to know what type a given structure is; this is all handled
automatically by whatever product is using the CF services. However, understanding that the
CF provides different types of services is useful when you are managing a CF, or if there is a
failure of a CF.
A CF has the unique ability to be stopped without impacting the users of its services. For
example, a CF containing a DB2 Lock structure could be shut down, upgraded, and brought
back online without impacting the DB2 subsystems that are using the structure. In fact, the
CF could even fail unexpectedly, and the users of its services could continue operating. This
capability relies on a combination of services provided by the CF and support in the products
that use its services that enable the contents of one CF to be dynamically moved to another
CF. For this reason, we recommend that every sysplex have at least two Coupling Facilities.
CFRM policy
The names and attributes of the structures that can reside in your CFs are described in a file
called the CFRM policy, which is stored in the CFRM Couple Data Set. The CFRM policy
would normally be created and maintained by the Systems Programmer. The contents of the
active CFRM policy can be displayed using a display command from the console. Some
structures have a fixed name (ISGLOCK, the GRS structure, for example). Other structures
(the JES2 checkpoint, for example) have a completely flexible name. Some of the information
that is included in the policy includes:
Information about your Coupling Facilities (LPAR name, serial number, and so on)
The name and sizes (minimum, initial, and maximum amounts) of each structure
Which CF each structure may be allocated in
Whether the system should monitor the use of storage within the structure and the
threshold at which the system should automatically adjust the structures size
The CF runs in an LPAR on any System z or zSeries processor. The code that is executed
in the LPAR is called Coupling Facility Control Code. This code is stored on the Service
Element of the processor and is automatically loaded when the CF LPAR is activated.
Unlike a z/OS system, a CF is not connected to disks or tapes or any of the normal peripheral
devices. Instead, the CF is connected to the z/OS systems that use its services by special
channels called CF Links. The only other device connected to the CF is the HMC, through
which a small set of commands can be issued to the CF. The links used to connect z/OS to
the CF are shown when you display information about the CF on the MVS console.
1.3 Sysplex types

Sysplex provides many capabilities, and it is up to each installation to decide which are the
most appropriate for them. However, to make discussions about sysplexes types easier,
there are three broad categories:
BronzePlex
In a BronzePlex, the minimum possible is shared between the systems in

the sysplex. Generally, such sysplexes are set up more to obtain the
financial benefits of Parallel Sysplex aggregation pricing, than to exploit the
technical benefits of Parallel Sysplex. However, even in a BronzePlex,
there are a number of components that must be shared, including the
Coupling Facility, the XCF signalling infrastructure, the common time
source, the console infrastructure, the WLM policy, the GRS environment,
and others.
GoldPlex
A Goldplex derives more technical benefits from Parallel Sysplex. In a

GoldPlex, many components of the systems infrastructure would be
common across the sysplex (for example, a single security database, a
single SMS policy, a single shared DASD environment, a single logical
system residence drive (SYSRES) even if there are multiple physical
copies, a single tape management system, shared tape libraries, a single
set of automation rules, and a single shared catalog structure). A
configuration like this reduces the effort needed to manage and administer
the sysplex. It also provides the possibility to move work from one system
to another in case of a planned or unplanned outage (although the move
would be disruptive).
PlatinumPlex
In a PlatinumPlex, everything is shared between all members of the

sysplex, and any work can run anywhere in the sysplex. The ability to
access any data from anywhere in the sysplex means that work can be
dynamically routed to whichever member of the sysplex is best able to
deliver the required level of service. It should be possible to maintain
application availability across both planned and unplanned outages in a
PlatinumPlex.
This document uses these terms to refer to the different types of sysplex. For more
information about this topic, refer to the IBM Redbooks publication Merging Systems into a
Sysplex, SG24-6818.
1.4 Parallel Sysplex test configuration

The Parallel Sysplex configuration used for the examples in this book is shown in Figure 1-2.
z/VM
#@$#PLEX
FACIL01
FACIL02
CFLvl 14
CFLvl 14
#@$1
#@$2
#@$3
CICS TS 3.1
DB2 V8
IMS V9
MQ V6
CICS TS 3.1
DB2 V8
IMS V9
MQ V6
CICS TS 3.1
DB2 V8
IMS V9
MQ V6
z/OS 1.7
z/OS 1.8
z/OS 1.8
Prim
Figure 1-2 The Test Parallel Sysplex configuration
10
This is a three-way, data sharing Parallel Sysplex with two Coupling Facilities. Each system
contains DB2, IMS, CICS, and MQ. All are set up to use the Coupling Facility to enable data
sharing, queue sharing, and dynamic workload balancing.
This sysplex is actually based on an offering known as the Parallel Sysplex Training
Environment, which is sold through IBM Technology and Education Services. The offering
consists of a full volume dump of the environment, a set of workloads to generate activity in
the sysplex, and an Exercise Guide. The offering can be installed in native LPARs or under
z/VM. We find that z/VM provides an excellent test environment because nearly everything
works exactly as it would in a native environmentbut you have more control over the scope
of things that can be touched from the test environment. The use of z/VM also makes it very
easy to add more systems, more Coupling Facilities, to add or remove CTCs, and so on.
Note: The unusual sysplex and system names (and subsystem names, as you will see
later in this book) were deliberately selected to minimize the chance of this sysplex having
the same names as any customer environment.
The three-way sysplex allows you to test recovery from a CF link failure, a CF failure, and a
system failure. Having workloads running at the same time makes the reaction of the system
and subsystems to these failure more closely resemble what happens in a production
environment.
11
12
Chapter 2.
Parallel Sysplex operator

commands
This chapter introduces the operator commands that are most commonly used to monitor and
control a Parallel Sysplex.
For more detailed information about specific commands, refer to z/OS MVS System
Commands, SA22-7627 and z/OS JES2 Commands, SA22-7526.
13
2.1 Overview of Parallel Sysplex operator commands

Display commands are not simply convenient to use. They are fundamental to the operator
gaining a working knowledge of your environment, particularly during problem diagnosis.
Some degree of monitoring may be performed by an automation product. However, the
operator not only needs to understand which display commands should be issued, but also
be able to interpret their output.
The following commands cover the most common operational aspects of the sysplex, such as
XCF, Coupling Facilities, GRS, and consoles. You can also refer to Appendix A, Operator
commands on page 491, for more information about this topic.
2.2 XCF and CF commands

This section describes the D XCF and D CF commands that can be used to gather information
about the Parallel Sysplex. As you will see, the D XCF command obtains the requested
information from the Couple Data Sets. The D CF command obtains its information from the
Coupling Facility.
2.2.1 Determining how many systems are in a Parallel Sysplex

To determine how many systems are in the Parallel Sysplex, as well as the system names,
issue the D XCF command as shown in Figure 2-1.
D XCF
IXC334I 18.39.46 DISPLAY XCF 438
SYSPLEX #@$#PLEX:
#@$1
#@$2
#@$3
Figure 2-1 Display XCF command
The output displays the name of your sysplex (in this example, #@$#PLEX).
Note: Be aware that, although system names are shown in this display, it does not
necessarily mean they are currently active. They may be in the process of being
partitioned out of the sysplex, for example.
2.2.2 Determining whether systems are active

As shown in Figure 2-2 on page 15, the D XCF,S,ALL command shows you which systems are
active, where they are running, when they last updated their status, and their timer mode.
Their ACTIVE status means that they have updated the sysplex Couple Data Set within the
failure detection interval.
14
D XCF,S,ALL
IXC335I 18.53.10 DISPLAY
SYSTEM
TYPE SERIAL LPAR
#@$3
2084 6A3A
N/A
#@$2
2084 6A3A
N/A
#@$1
2084 6A3A
N/A
XCF 491
STATUS TIME
06/21/2007 18:53:10
06/21/2007 18:53:06
06/21/2007 18:53:07
SYSTEM STATUS
ACTIVE
ACTIVE
ACTIVE
TM=SIMETR
TM=SIMETR
TM=SIMETR
Figure 2-2 Display all active systems
2.2.3 Determining what the CFs are called

As shown in Figure 2-3, the D XCF,CF command provides you with summary information about
the CFs physically connected to the z/OS system that the command was issued on.
D XCF,CF
CFNAME
FACIL01
FACIL02
COUPLING FACILITY
SIMDEV.IBM.EN.0000000CFCC1
PARTITION: 00 CPCID: 00
SITE
N/A
N/A
Figure 2-3 Display Coupling Facility names
For more information about the contents of each CF, as well as information about which
systems are connected to each CF, use the D XCF,CFNM=ALL command. This is discussed
further in 2.2.6, Determining which structures are in the CF on page 22.
For detailed physical information about each Coupling Facility, issue the D CF command as
shown in Figure 2-4 on page 16. The display is repeated for each CF defined in the CFRM
policy that is currently available to the z/OS system where the command was issued.
For example, some installations define their Disaster Recovery Coupling Facilities in their
CFRM policy. These CFs would be shown in the output from the D XCF,CF command.
However, they would not show in the output from the D CF command because they are not
online to that system.
The information displayed in Figure 2-4 on page 16 contains the following details:
1 The name and physical information about the CF
2 Space utilization and the CF Level and service level
3 CF information (type and status)
4 Subchannel status
5 Information about remote CFs (used for System Managed Duplexing)
CF Level 15 provides additional information:
The number of dedicated and shared PUs in the CF
Whether Dynamic CF Dispatching is enabled on this CF
For more detailed information about the Coupling Facility, refer to Chapter 7, Coupling
Facility considerations in a Parallel Sysplex on page 101.
Chapter 2. Parallel Sysplex operator commands
15
D CF
IXL150I 19.02.15 DISPLAY CF 516
COUPLING FACILITY SIMDEV.IBM.EN.0000000CFCC1
CONTROL UNIT ID: 0309
NAMED FACIL01 1
COUPLING FACILITY SPACE UTILIZATION
ALLOCATED SPACE
DUMP SPACE UTILIZATION
STRUCTURES:
108544 K
STRUCTURE DUMP TABLES:
DUMP SPACE:
2048 K
TABLE COUNT:
FREE SPACE:
612864 K
FREE DUMP SPACE:
TOTAL SPACE:
723456 K
TOTAL DUMP SPACE:
MAX REQUESTED DUMP SPACE:
VOLATILE:
YES
STORAGE INCREMENT SIZE:
CFLEVEL:
14
CFCC RELEASE 14.00, SERVICE LEVEL 00.29
BUILT ON 03/26/2007 AT 17:58:00
COUPLING FACILITY HAS ONLY SHARED PROCESSORS
COUPLING FACILITY SPACE CONFIGURATION 2
IN USE
FREE
CONTROL SPACE:
110592 K
612864 K
NON-CONTROL SPACE:
0 K
0 K
SENDER PATH 3
09
0E
PHYSICAL
ONLINE
ONLINE
LOGICAL
ONLINE
ONLINE
COUPLING FACILITY SUBCHANNEL STATUS 4

TOTAL:
6 IN USE:
6
NOT USING:
0
DEVICE
SUBCHANNEL
STATUS
4030
0004
OPERATIONAL
4031
0005
OPERATIONAL
4032
0006
OPERATIONAL
4033
0007
OPERATIONAL
4034
0008
OPERATIONAL
4035
0009
OPERATIONAL
0
0
2048
2048
0
256
K
K
K
K
K
TOTAL
723456 K
0 K
CHANNEL TYPE
ICP
ICP
NOT USABLE:
REMOTELY CONNECTED COUPLING FACILITIES 5

CFNAME
COUPLING FACILITY
--------------------------------FACIL02
CHPIDS ON FACIL01 CONNECTED TO REMOTE FACILITY
RECEIVER: CHPID
TYPE
F0
ICP
SENDER:
CHPID
TYPE
E0
ICP
NOT OPERATIONAL CHPIDS ON FACIL01
81
Figure 2-4 Display CF details
2.2.4 Obtaining more information about CF paths

When IBM first introduced Coupling Facilities, the links to connect the CF to z/OS were
defined as being either CF receiver (CFR) or sender (CFS) paths (type 0B and 0C in the D
16
M=CHP display). When the zSeries range of processors were announced, an enhanced link
type (peer mode links) was introduced. Because CFR and CFS links are not strategic, this
document only discusses peer mode links.
Previously, on zSeries processors, three types of CF links were supported:
Internal Coupling (IC)
This is used to connect a Coupling Facility to a z/OS

LPAR in the same processor.
Integrated Cluster Bus (ICB)
These are copper links, typically used to connect a

Coupling Facility to z/OS in another processor. The
processors must be within 7 meters of each other.
Inter-System Coupling (ISC)
These are fiber links, with the lowest performance. They

can be used to connect z/OS to a Coupling Facility that is
up to 100 km away when used with a multiplexor.
System z10 introduced a new type of CF link known as Parallel Sysplex over Infiniband
(PSIFB) or Coupling over Infiniband (CIB). These also use fiber connections, and at the time
of writing support a maximum distance of 150 meters between the processors.
There are two ways you can obtain information about the CF links. The first way is to issue a
D M=CHP command; Figure 2-5 on page 18 shows an example of the use of this command.
Results of the display that are irrelevant to this exercise have been omitted and replaced with
an ellipsis (...).
17
D M=CHP
IEE174I 20.02.38 DISPLAY M 635
CHANNEL PATH STATUS
0 1 2 3 4 5 6 7 8 9 A B C D E F
0 + + + + + + + + + + + + + + + +
...
F + + + + + + + + + + + + + + + +
************************ SYMBOL EXPLANATIONS ********************
+ ONLINE
@ PATH NOT VALIDATED - OFFLINE
. DOES NOT EXIST
* MANAGED AND ONLINE
# MANAGED AND OFFLINE
CHANNEL PATH TYPE STATUS
0 1 2 3 4 5 6 7 8 9 A B C D E F
0 11 11 11 11 11 11 11 14 11 23 14 14 11 14 23 23 1
...
A 1B 1D 00 00 00 00 00 00 00 00 00 00 00 00 00 00
B 00 00 00 00 00 00 00 00 00 00 00 00 00 00 22 22 2
C 21 21 21 21 21 21 21 21 17 17 21 21 00 00 00 00
D 00 00 00 00 23 23 23 23 23 23 23 23 23 23 00 00
E 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
F 24 24 24 24 00 00 00 00 00 00 00 00 24 24 24 24
************************ SYMBOL EXPLANATIONS ******
00 UNKNOWN
UNDEF
01 PARALLEL BLOCK MULTIPLEX
BLOCK
02 PARALLEL BYTE MULTIPLEX
BYTE
03 ESCON POINT TO POINT
CNC_P
04 ESCON SWITCHED OR POINT TO POINT CNC_?
05 ESCON SWITCHED POINT TO POINT
CNC_S
06 ESCON PATH TO A BLOCK CONVERTER
CVC
07 NATIVE INTERFACE
NTV
08 CTC POINT TO POINT
CTC_P
09 CTC SWITCHED POINT TO POINT
CTC_S
0A CTC SWITCHED OR POINT TO POINT
CTC_?
0B COUPLING FACILITY SENDER
CFS
0C COUPLING FACILITY RECEIVER
CFR
0D UNKNOWN
UNDEF
0E UNKNOWN
UNDEF
0F ESCON PATH TO A BYTE CONVERTER
CBY
...
1A FICON POINT TO POINT
FC
1B FICON SWITCHED
FC_S
1C FICON TO ESCON BRIDGE
FCV
1D FICON INCOMPLETE
FC_?
1E DIRECT SYSTEM DEVICE
DSD
1F EMULATED I/O
EIO
20 RESERVED
UNDEF
21 INTEGRATED CLUSTER BUS PEER
CBP
22 COUPLING FACILITY PEER
CFP 3
23 INTERNAL COUPLING PEER
ICP 4
24 INTERNAL QUEUED DIRECT COMM
IQD
25 FCP CHANNEL
FCP
NA INFORMATION NOT AVAILABLE
Figure 2-5 Display all CHPs
The information displayed in Figure 2-5 contains the following details:

A type 22 channel is a CFP link 3. It is used by channels BE and BF 2.
A type 23 channel is an ICP link 4. It is used by channels 0E and 0F 1.
18
You can also display specific CHPIDs to learn their details, as shown in Figure 2-6.
D M=CHP(BE)
IEE593I CHANNEL PATH BE HAS NO OWNERS
CHPID BE: TYPE=22, DESC=COUPLING FACILITY PEER, ONLINE
Figure 2-6 Display CFP-type channel
Figure 2-6 indicates that even though the channel is online, it is not in use. (We would not
really expect it to be in this configuration, because all of the systems are on the same CEC.)
Figure 2-7 shows the ICP type channel 0E, which was established previously. This is the type
that we would expect to be in use in this exercise.
D M=CHP(0E)
CHPID 0E: TYPE=23, DESC=INTERNAL COUPLING PEER, ONLINE
NAMED FACIL01
SENDER PATH
0E
PHYSICAL
ONLINE
LOGICAL
ONLINE
COUPLING FACILITY SUBCHANNEL STATUS

TOTAL:
4 IN USE:
4
NOT USING:
0
DEVICE
SUBCHANNEL
STATUS
5030
0004
OPERATIONAL
5031
0005
OPERATIONAL
5032
0006
OPERATIONAL
5033
0007
OPERATIONAL
CHANNEL TYPE
ICP
NOT USABLE:
Figure 2-7 Display ICP-type channel
The second way to obtain information is to issue a D CF command for the CF you want to
know about, as shown in Figure 2-8. The CF names are shown in Figure 2-4 on page 16.
D CF,CFNAME=FACIL01
...
SENDER PATH
PHYSICAL
09
ONLINE
0E
ONLINE
LOGICAL
ONLINE
ONLINE
CHANNEL TYPE
ICP
ICP
Figure 2-8 Display sender paths in CF name detail
2.2.5 Obtaining information about structures

To obtain a list of all structures defined in the active CFRM policy in your Parallel Sysplex,
use the D XCF,STR command as shown in Figure 2-9 on page 20. The structures are listed
alphabetically.
The display shows all defined structures, whether each is allocated or not. If the structure is
allocated, the time and date when it was allocated are displayed. For allocated structures, the
structure type and whether it is duplexed or not are also shown.
19
For a display of only the structures that are currently allocated, use the
D XCF,STR,STAT=ALLOC command.
D XCF,STR
STRNAME
ALLOCATION TIME
CIC_DFHLOG_001 06/21/2007 01:47:54
CIC_DFHSHUNT_001 06/21/2007 01:47:56
CIC_GENERAL_001
--D#$#_GBP1
06/20/2007 04:11:05
D#$#_GBP1
06/20/2007 04:11:01
D#$#_GBP32K1
D#$#_LOCK1
--06/20/2007 03:32:17
D#$#_LOCK1
06/20/2007 03:32:15
D#$#_SCA
06/20/2007 03:32:12
D#$#_SCA
06/20/2007 03:32:10
DFHCFLS_#@$CFDT1
I#$#EMHQ
I#$#LOCK1
IGWCACHE1
IGWLOCK00
IRRXCF00_B001
ISGLOCK
ISTGENERIC
IXC_DEFAULT_1
JES2CKPT_1
LOG_FORWARD_001
LOG_SA390_MISC
. . .
06/21/2007
---06/16/2007
06/18/2007
06/18/2007
06/16/2007
06/18/2007
----
01:47:27
---06:36:16
03:43:29
03:43:12
06:36:26
03:43:00
----
STATUS
TYPE
ALLOCATED
LIST
ALLOCATED
LIST
NOT ALLOCATED
ALLOCATED (NEW)
CACHE
DUPLEXING REBUILD
METHOD: USER-MANAGED
PHASE: DUPLEX ESTABLISHED
ALLOCATED (OLD)
CACHE
DUPLEXING REBUILD
NOT ALLOCATED
ALLOCATED (NEW)
LOCK
DUPLEXING REBUILD
METHOD: SYSTEM-MANAGED
ALLOCATED (OLD)
LOCK
ALLOCATED (NEW)
LIST
DUPLEXING REBUILD
ALLOCATED (OLD)
LIST
DUPLEXING REBUILD
ALLOCATED
LIST
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
ALLOCATED
LOCK
ALLOCATED
CACHE
ALLOCATED
LOCK
ALLOCATED
SLIST
ALLOCATED
LIST
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
Figure 2-9 Displaying all defined structures
20
If you need more detail about a specific structure, for example the ISGLOCK structure, issue
the D XCF,STR,STRNAME=name command as shown in Figure 2-10.
D XCF,STR,STRNAME=ISGLOCK
STRNAME: ISGLOCK
STATUS: ALLOCATED
EVENT MANAGEMENT: POLICY-BASED
TYPE: LOCK
POLICY INFORMATION:
POLICY SIZE
: 8704 K
POLICY INITSIZE: 8704 K
POLICY MINSIZE : 0 K
FULLTHRESHOLD : 80
ALLOWAUTOALT : NO
REBUILD PERCENT: 1
DUPLEX
: DISABLED
ALLOWREALLOCATE: YES
PREFERENCE LIST: FACIL02 FACIL01
ENFORCEORDER : NO
EXCLUSION LIST IS EMPTY
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/18/2007 03:43:12
CFNAME
: FACIL02
COUPLING FACILITY: SIMDEV.IBM.EN.0000000CFCC2
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 8704 K
STORAGE INCREMENT SIZE: 256 K
LOCKS:
TOTAL:
1048576
PHYSICAL VERSION: C0C39A21 7B9444C5
LOGICAL VERSION: C0C39A21 7B9444C5
SYSTEM-MANAGED PROCESS LEVEL: 8
XCF GRPNAME
: IXCLO007
DISPOSITION
: DELETE
1
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3
CONNECTION NAME
---------------ISGLOCK##@$1
ISGLOCK##@$2
ISGLOCK##@$3
ID
-03
02
01
VERSION
-------00030067
00020060
0001008D
SYSNAME
-------#@$1
#@$2
#@$3
JOBNAME
-------GRS
GRS
GRS
ASID
---0007
0007
0007
STATE
-------ACTIVE
ACTIVE
ACTIVE
DIAGNOSTIC INFORMATION: STRNUM: 00000007 STRSEQ: 00000002

MANAGER SYSTEM ID: 00000000
Figure 2-10 Display structure details
The response to this command consists of two sections, as explained here:

Most of the information displayed preceding the ACTIVE STRUCTURE line is retrieved
from the CFRM policy, and is presented regardless of whether the structure is currently
allocated or not. This represents the definition of the structure. It may not match how the
structure is currently allocated.
21
The information displayed following the ACTIVE STRUCTURE line represents the actual
structure. This shows which CF the structure is currently allocated in, the structure size,
which address spaces are connected to it, and so on.
In the command output, the disposition 1 of DELETE is of particular interest. This specifies
that when the final user of this structure shuts down cleanly, the structure will be deleted. The
next time an address space that uses this structure tries to connect to it, the structure will be
allocated again, using information from the CFRM policy in most cases.
The extended version of this command, D XCF,STR,strname,CONNAME=ALL, provides all the
information shown in Figure 2-10 on page 21, as well as information unique to each
connector to the structure. This information can help you determine whether the structure
connectors support functions such as User-Managed Duplexing, System-Managed Rebuild,
System-Managed Duplexing, and so on.
2.2.6 Determining which structures are in the CF

Use the D XCF,CF,CFNAME=name command shown in Figure 2-11 to learn which structures are
currently located in a CF.
D XCF,CF,CFNAME=FACIL01
CFNAME: FACIL01
COUPLING FACILITY
:
SITE
:
N/A
POLICY DUMP SPACE SIZE:
2000 K
ACTUAL DUMP SPACE SIZE:
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
CIC_DFHSHUNT_001
D#$#_LOCK1(OLD)
DFHNCLS_#@$CNCS1
IXC_DEFAULT_2
D#$#_GBP0(NEW)
D#$#_SCA(OLD)
DFHXQLS_#@$STOR1
SYSTEM_OPERLOG
D#$#_GBP1(NEW)
DFHCFLS_#@$CFDT1
IRRXCF00_P001
Figure 2-11 Display CF content information
This command also shows information about which systems are connected to this CF.
Note: In case of a CF failure, the information in the output from this command represents
the CF contents at the time of the failure. Normally, structures will automatically rebuild
from a failed CF to an alternate.
If you issue this command before the failed CF is brought online again, you will see that
some structures are listed as being in both CFs. After the failed CF comes online, it
communicates with z/OS to verify which structures are still in the CF (normally, the CF
would be empty at this point), and this information will be updated at that time.
22
2.2.7 Determining which Couple Data Sets are in use

In addition to the CFs, another critical set of resources in a sysplex environment consists of
Couple Data Sets.
Using the D XCF,COUPLE command, as shown in Figure 2-12 on page 24, you obtain
information about the primary and alternate (if one is currently defined) Couple Data Sets.
This figure only shows the first two Couple Data Sets 4; successive Couple Data Sets appear
in the same display format.
The typical CDS types that may be displayed are ARM, BPXMCDS, CFRM, LOGR, SFM, and
WLM.
The output from the D XCF,COUPLE command displays a large amount of information that will
be frequently referred to in this book, particularly the INTERVAL 1 and CLEANUP 2 values, as
explained here:
The INTERVAL value is used to determine at which point a system is deemed to probably
be dead and ready to be partitioned out of the sysplex. In this example, there is no active
SFM policy. You can tell this by the N/A value in the SSUM ACTION field 3. If SFM was
active, this field would show ISOLATE (automatically perform system partitioning) or
PROMPT (notify operator). More information about SFM is available in Chapter 5,
Sysplex Failure Management on page 73.
The CLEANUP value controls how long a system will wait for address spaces to shut
themselves down following a V XCF,sysname,OFFLINE command before it places itself in a
wait state.
After listing other information about the sysplex, information about the various Couple Data
Sets is provided 4. In addition to the names of the primary and alternate Couple Data Sets,
information about the formatting options used for each data set is provided as well.
23
D XCF,COUPLE
SYSTEM #@$3 DATA
1INTERVAL OPNOTIFY
MAXMSG
85
88
2000
SSUM ACTION
3 N/A
CLEANUP2
15
SSUM INTERVAL
N/A
RETRY
10
CLASSLEN
956
WEIGHT MEMSTALLTIME
N/A
N/A
MAX SUPPORTED CFLEVEL: 14

MAX SUPPORTED SYSTEM-MANAGED PROCESS LEVEL: 14
CF REQUEST TIME ORDERING FUNCTION: NOT-INSTALLED
SYSTEM NODE DESCRIPTOR: 002084.IBM.02.000000026A3A
PARTITION: 19
CPCID: 00
SYSTEM IDENTIFIER: 031B085B 0100029C
COUPLEXX PARMLIB MEMBER USED AT IPL: COUPLE00
SYSPLEX COUPLE DATA SETS
PRIMARY
4DSN: SYS1.XCF.CDS01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM MAXGROUP(PEAK) MAXMEMBER(PEAK)
11/20/2002 16:27:24
4
100
(52)
203 (18)
ALTERNATE DSN: SYS1.XCF.CDS02
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM MAXGROUP
MAXMEMBER
11/20/2002 16:27:28
4
100
203
Figure 2-12 Displaying CDS information
2.2.8 Determining which XCF signalling paths are defined and available
For the XCF function on the different members of the sysplex to be able to communicate with
each other, some method of connecting the systems must be defined. These communication
paths are known as XCF signalling resources.
The D XCF,PATHIN/PATHOUT commands provide information for only those devices and
structures that are defined to the system where the commands are entered (in this example,
#@$3).
To obtain information about the inbound paths, enter D XCF,PI as shown in Figure 2-13.
D XCF,PI
PATHIN FROM SYSNAME: #@$1
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
PATHIN FROM SYSNAME: 1 ???????? - PATHS NOT CONNECTED TO OTHER SYSTEMS
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
Figure 2-13 Display inbound signalling paths
To obtain information about the outbound paths, enter D XCF,PO as shown in Figure 2-14
on page 25.
24
D XCF,PO
PATHOUT TO SYSNAME:
#@$1
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
PATHIN FROM SYSNAME: 1 ???????? - PATHS NOT CONNECTED TO OTHER SYSTEMS
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
Figure 2-14 Display outbound signalling paths
As shown, there is one path not connected to another system. A likely reason for this is that
the target system (#@$2) may not be active at the time the display was done.
For a more detailed display, issue either D XCF,PI,DEV=ALL or D XCF,PO,DEV=ALL.
2.2.9 Determining whether Automatic Restart Manager is active

The Automatic Restart Manager (ARM) is a standard function of z/OS. However, its use is
optional. As its name implies, the Automatic Restart Manager can be used to automatically
restart address spaces after either an address space failure or an entire system failure.
Using the D XCF,ARMSTATUS,DETAIL command shown in Figure 2-15, you obtain summary
information about the status of ARM and detailed information about jobs and started tasks
that are registered as elements of ARM.
D XCF,ARMSTATUS,DETAIL
ARM RESTARTS ARE ENABLED
-------------- ELEMENT STATE SUMMARY -------------- -TOTAL- -MAXSTARTING AVAILABLE FAILED RESTARTING RECOVERING
0
36
0
0
0
36
200
RESTART GROUP:CICS#@$1
PACING :
0
FREECSA:
0
0
ELEMENT NAME :SYSCICS_#@$CCM$1 JOBNAME :#@$CCM$1 STATE
:AVAILABLE
CURR SYS :#@$1
JOBTYPE :STC
ASID
:0024
INIT SYS :#@$1
JESGROUP:XCFJES2A TERMTYPE:ELEMTERM
EVENTEXIT:*NONE*
ELEMTYPE:SYSCICS LEVEL
:
2
TOTAL RESTARTS :
0
INITIAL START:06/21/2007 01:47:25
RESTART THRESH :
0 OF 3
FIRST RESTART:*NONE*
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
Figure 2-15 Display ARM detail
2.3 JES2 commands

This section describes the JES2 commands you would use to handle:
Checkpoint reconfiguration
Checkpoint lock situations
2.3.1 Determining JES2 checkpoint definitions

The $DCKPTDEF command shown in Figure 2-16 on page 26 provides you with information
about the JES2 checkpoint definitions. The checkpoint can reside in a CF structure, on DASD,
or both.
25
$DCKPTDEF
$HASP829 CKPTDEF 489
$HASP829 CKPTDEF CKPT1=(STRNAME=JES2CKPT_1,INUSE=YES,
1
$HASP829
VOLATILE=YES),CKPT2=(DSNAME=SYS1.JES2.CKPT2,2
$HASP829
VOLSER=#@$#M1,INUSE=YES,VOLATILE=NO),
$HASP829
NEWCKPT1=(DSNAME=,VOLSER=),NEWCKPT2=(DSNAME=,
$HASP829
VOLSER=),MODE=DUPLEX,DUPLEX=ON,LOGSIZE=7,
$HASP829
VERSIONS=(STATUS=ACTIVE,NUMBER=2,WARN=80,
$HASP829
MAXFAIL=0,NUMFAIL=0,VERSFREE=2,MAXUSED=1),
$HASP829
RECONFIG=NO,VOLATILE=(ONECKPT=WTOR,
$HASP829
ALLCKPT=WTOR),OPVERIFY=NO
Figure 2-16 Display JES2 checkpoint definitions
The response to this command shows that:

1 The primary JES2 checkpoint (CKPT1) is defined to be in a structure.
2 The alternate checkpoint (CKPT2) is defined to be on DASD.
You will notice that in the definitions shown here, neither a NEWCKPT1 nor a NEWCKPT2 is
defined, meaning that if either CKPT1 or CKPT2 were to fail, the system would not be able to
automatically forward the checkpoint to the recovery location. Instead, operator intervention
would be required.
Tip: In general, IBM recommends placing CHKPT1 in a CF and CKPT2 on DASD,
especially if there are a large number of systems in the JES2 Multi-Access Spool (MAS).
2.3.2 Releasing a locked JES2 checkpoint

If as in Figure 2-17your installation is using a JES2 MAS and one JES2 member tries to
reserve the software lock on the checkpoint data set, but determines that another member
has control of it, message $HASP264 is issued, as in Figure 2-17.
$HASP264 WAITING FOR RELEASE OF JES2 CKPT LOCK BY #@$1
Figure 2-17 Waiting for JES2 checkpoint release
As shown in Figure 2-18, the recovery for this is to first issue a $D MASDEF command. This is
best done with the RO *ALL option, for a display of all members.
$D MASDEF
$HASP843 MASDEF 604
$HASP843 MASDEF OWNMEMB=#@$1,AUTOEMEM=ON1,CKPTLOCK=ACTION,
$HASP843
COLDTIME=(2006.164,19:53:14),COLDVRSN=z/OS 1.4,
$HASP843
DORMANCY=(0,100),HOLD=0,LOCKOUT=1000,
$HASP843
RESTART=YES2,SHARED=CHECK,SYNCTOL=120,
$HASP843
WARMTIME=(2007.192,03:11:03),XCFGRPNM=XCFJES2A,
$HASP843
QREBUILD=0
Figure 2-18 Display JES2 MASDEF
As shown, the AUTOEMEM parm is set to ON 1 and RESTART 2 is set to YES. In this case,
it should auto-recover.
26
If AUTOEMEM were OFF (or if it were set to ON but the RESTART parm was set to NO), then
the operator should issue the command $E CKPTLOCK,HELDBY=sysname, an example of which
is shown in Figure 2-19.
$E CKPTLOCK,HELDBY=#@$1
Figure 2-19 Release JES2 checkpoint lock
This removes the lock on the checkpoint data set held by the identified system, #@$1.
Important:
Do not confuse message:
$HASP263 WAITING FOR ACCESS TO JES2 CHECKPOINT
With this message:
$HASP264 WAITING FOR RELEASE OF JES2 CKPT LOCK BY sysname
2.3.3 JES2 checkpoint reconfiguration

The JES2 checkpoint reconfiguration dialog can be initiated for a variety of reasons:
You want to move the checkpoint from one volume to another.

You want to change from having the checkpoint in a CF structure to being in a data set.
You need to implement new checkpoint data sets.
You need to suspend and resume the use of a checkpoint data set.
The checkpoint reconfiguration process is explained in Chapter 10, Managing JES2 in a

Parallel Sysplex on page 201.
2.4 Controlling consoles in a sysplex

In a single-system environment, or prior to the delivery of sysplex, a console was only
associated with a single system. However, in a sysplex environment, consoles are a sysplex
resource. This means that each console must have a name that is unique within the sysplex;
each console can issue commands to any system in the sysplex; and each console can
receive messages from any system in the sysplex.
For a detailed description about managing consoles in a Parallel Sysplex, refer to Chapter 14,
Managing consoles in a Parallel Sysplex on page 283.
2.4.1 Determining how many consoles are defined in a sysplex

To obtain information about the active consoles in the sysplex, use the D C,A,CA command.
Figure 2-20 on page 28 shows three active consoles. Notice that each console has the same
address. However, each one is on a different system, meaning that there are three physical
consoles, each with a unique name (containing the system name), and each defined to use
the same address on their respective systems.
27
D C,A,CA
IEE890I 03.54.03
NAME
ID
#@$1M01
13
#@$2M01
11
#@$3M01
01
CONSOLE DISPLAY 830

SYSTEM
ADDRESS
#@$1
08E0
#@$2
08E0
#@$3
08E0
STATUS
ACTIVE
ACTIVE
ACTIVE
Figure 2-20 Display active consoles
Note: Starting with z/OS 1.8, it is no longer necessary (or possible) to have a single
sysplex Master console, although you can still have multiple consoles that have master
authority.
2.4.2 Managing console messages

z/OS provides great flexibility regarding which messages will appear on each console. In
general, the decision about which subset of messages a given console will see is determined
by the responsibilities of the group that uses that console. For example, the console in the
tape drive area might be set up to see all tape mount requests from all systems in the
sysplex.
Explaining how to plan and set up your console configuration is beyond the scope of this
book. Here, we simply highlight some of the commands you can use to control the scope of
which systems can send messages to a particular console.
To receive only messages from the image that the console is defined on, use the
V CN(*),MSCOPE=(*) command. To receive messages from all systems in your sysplex, use
the V CN(*),MSCOPE=*ALL command. Note, however, that setting a console up in this manner
is not recommended due to possible console flooding. However, when used with the
ROUTCDE parm to reduce the potential number of messages being routed to the console, it
may be acceptable. You may also use the V CN(*),MSCOPE=(sys1,sys2,....) format to
receive messages from more than one, but less than all the systems in the sysplex.
2.5 GRS commands

This section describes the GRS commands you can use to get information about your Global
Resource Serialization (GRS) Star complex. The old Ring configuration is not addressed in
this book. This is because all Parallel Sysplex members must be in the same GRS complex,
and it is recommended that you use GRS Star for both improved performance and
availability. Note that you cannot have a GRS complex containing a mix of systems in the
sysplex and systems outside the sysplex when using GRS Star.
2.5.1 Determining which systems are in a GRS complex

For summary information about your GRS complex, use the D GRS command shown in
Figure 2-21 on page 29.
28
D GRS
ISG343I 19.40.20 GRS STATUS 015
SYSTEM
STATE
SYSTEM
STATE
#@$1
CONNECTED
#@$2
CONNECTED
#@$3
CONNECTED
GRS STAR MODE INFORMATION 1
LOCK STRUCTURE (ISGLOCK) CONTAINS 1048576 LOCKS.
THE CONTENTION NOTIFYING SYSTEM IS #@$3
SYNCHRES:
YES
Figure 2-21 Display GRS information
This information 1 indicates that GRS is operating in Star mode. The GRS lock structure
(which must be called ISGLOCK) contains some number of lock entries. Note that the number
of lock entries in a GRS structure must always be a power of 2, meaning that if you want to
increase the size of the structure, the size must be doubled each time. This lock information
only appears when GRS is in Star mode.
2.5.2 Determining whether any jobs are reserving a device

One way of being notified that a device has a RESERVE against it is when you see the START
PENDING message, as shown in Figure 2-22.
IOS078I 1D06,5A,XCFAS, I/O TIMEOUT INTERVAL HAS BEEN EXCEEDED
IOS071I 1D06,**,*MASTER*, START PENDING
Figure 2-22 Start Pending message
If the device is reserved by another system, message IOS431I might follow; it identifies the
system holding the reserve.
For information about a specific device enter D GRS,DEV=devno, as shown in Figure 2-23.
Using this information, you can see which job is causing the reserve. You can decide if that
job should be allowed to continue, or if it is experiencing problems and should be cancelled.
D GRS,DEV=1D06
DEVICE:1D06 VOLUME:#@$#X1 RESERVED BY SYSTEM #@$3
S=SYSTEMS MVSRECVY ES3090.RNAME1
SYSNAME
JOBNAME
ASID
TCBADDR
EXC/SHR
#@$3
RESERVE
001A
007E4B58 EXCLUSIVE
STATUS
OWN
Figure 2-23 Display GRS by device
You may also see that a device has a reserve against it if you issue a DEVSERV command for
that device. Figure 2-24 on page 30 shows an example where the DEVSERV command has
been issued for a device that currently has a reserve.
29
DS P,1D00
IEE459I 20.23.09 DEVSERV PATHS 060
UNIT DTYPE M CNT VOLSER CHPID=PATH STATUS
RTYPE
SSID CFW TC DFW PIN DC-STATE CCA DDC
ALT
1D00,33903 ,A,023,#@$#M1,5A=R 5B=R 5C=R 5D=R
2105
8981 Y YY. YY.
N
SIMPLEX
32 32
************************ SYMBOL DEFINITIONS *****************
A = ALLOCATED
R = PATH AVAILABLE AND RES
Figure 2-24 Devserv on paths of reserved device
The reserve also shows up with the D U command shown in Figure 2-25.
D U,,,1D00,1
IEE457I 20.21.59 UNIT STATUS 047
UNIT TYPE STATUS
VOLSER
1D00 3390 A
-R
#@$#M1
VOLSTATE
PRIV/RSDNT
Figure 2-25 D U of a reserved device
Even if a device has a reserve against it, that is not necessarily a problem. Be aware,
however, that no other system will be able to update a data set on a volume that has a
reserve against it, so reserves that impact another job for a long time should be investigated.
2.5.3 Determining whether there is resource contention in a sysplex

Another way that a program can serialize a resource is by issuing an ENQ request. This is
considered preferable to using a reserve, because only the ENQed resource is serialized,
rather than the whole volume.
However, there is still the possibility than a program can hold an ENQ for a long time, locking
out other programs that may want to use that resource. If you find that a program is stopped,
you can use the D GRS,C command to determine if that program is being delayed because of
ENQ or Latch contention, as shown in Figure 2-26.
D GRS,C
S=SYSTEMS SYSDSN
EXAMPLE1.XX
SYSNAME
JOBNAME
ASID
TCBADDR
EXC/SHR
#@$3
SAMPJOB1
001A
007FF290 EXCLUSIVE
#@$3
SAMPJOB2
001F
007FF290
SHARE
NO REQUESTS PENDING FOR ISGLOCK STRUCTURE
NO LATCH CONTENTION EXISTS
STATUS
OWN
WAIT
Figure 2-26 Display GRS contention
2.5.4 Obtaining contention information about a specific data set

You can use the D GRS,RES=(*,dsn) command to obtain information about a specific data set,
as shown in Figure 2-27 on page 31. The output is similar to that provided by the D GRS,C
command.
30
D GRS,RES=(*,EXAMPLE1.XX)
S=SYSTEMS SYSDSN
EXAMPLE1.XX
SYSNAME
JOBNAME
ASID
#@$3
SAMPJOB1
001A
#@$3
SAMPJOB2
001F
1
TCBADDR
EXC/SHR
007FF290 EXCLUSIVE
007FF290
SHARE
2
STATUS
OWN
WAIT
Figure 2-27 Display GRS by resource name
The possible values for 1 EXC/SHR and 2 STATUS are as follows:

EXCLUSIVE
The job requested exclusive use of the resource.
SHARE
The job requested shared use of the resource.
OWN
The job currently owns the resource.
WAIT
The job is waiting for the resource.
Another option is to use a variation of the D GRS,ANALYZE command, which will provide
information about the root cause of any contention you may be encountering. We recommend
that you become familiar with the use of this command, so that you can quickly use it to
diagnose any contention problems that may arise.
2.6 Commands associated with External Timer References

This section describes how to obtain information about the common time source for your
sysplex.
2.6.1 Obtaining Sysplex Timer status information

A common time source is needed in a sysplex to keep the local time on all systems
synchronized. If the members of the sysplex are all on the same CEC (as in the case of our
test sysplex), the common time source can be simulated (this is known as SIMETR mode).
If the members of the sysplex are spread over more than one CEC, the common time source
can come from an IBM 9037 Sysplex Timer, or from the Server Time Protocol (STP). The
parameter that determines the timer mode of the system is coded in the CLOCKxx member of
PARMLIB.
The D ETR command is used to obtain information about the timing mode. The following
displays illustrate the differences in the D ETR output, depending on which timer mode
Simulated ETR mode, ETR mode, or STP mode) is in use.
Simulated ETR mode

When all systems are on the same physical CEC, an operational sysplex timer is not
necessary, and the systems can run in Simulation mode (SIMETR). A Sysplex Timer can still
be attached and used if required. Figure 2-28 shows the response to the D ETR commend
when the systems is in SIMETR mode.
D ETR
IEA282I 23.47.59 TIMING STATUS 231
ETR SIMULATION MODE, SIMETRID = 00
Figure 2-28 Display Sysplex Timer in SIMulation mode
31
ETR mode
The 9037s provide the setting and synchronization for the TOD clocks of the CEC or multiple
CECs. The IPLing system determines the timing mode from the CEC. Each CEC should be
connected to two 9037s, thus providing the ability to continue operating even if one 9037 fails.
Figure 2-29 shows the status of the two ETR ports on the CEC. In this display, the ETR NET
ID of both 9037s is the same; only the port numbers and ETR ID differ. The display shows
which 9037 is currently being used for the time synchronization signals. If that 9037 or the
connection to it were to fail, the CEC will automatically switch to the backup.
D ETR
IEA282I 23.38.48 TIMING STATUS 550
SYNCHRONIZATION MODE = ETR
CPC PORT 0 <== ACTIVE
CPC PORT 1
OPERATIONAL
OPERATIONAL
ENABLED
ENABLED
ETR NET ID=01
ETR NET ID=01
ETR PORT=01
ETR PORT=02
ETR ID=00
ETR ID=01
Figure 2-29 Display Sysplex Timer ETR
STP mode
Server Time Protocol is the logical replacement for 9037s. STP is a message-based protocol
in which timekeeping information is passed over data links between CECs.
STP must run a Coordinated Timing Network (CTN). Like the 9037, the same network ID
must be used by all systems that are to have synchronized times.
This network can be configured as STP-only, where all CECs use only STP, or the network
can be configured as Mixed. A Mixed network uses both STP and 9037s. In a Mixed CTN, the
9037 still controls the time for the whole sysplex.
Figure 2-30 shows the response from a system in STP mode.
D ETR
SYNCHRONIZATION MODE = STP
THIS SERVER IS A STRATUM 1
CTN ID = ISCTEST
THE STRATUM 1 NODE ID = 002084.C24.IBM.02.000000046875
THIS IS THE PREFERRED TIME SERVER
THIS STP NETWORK HAS NO SERVER TO ACT AS ARBITER
Figure 2-30 Display Sysplex Timer with STP
Using D XCF,S,All
Another way to see the timing mode of each system in the sysplex is to issue the D XCF,S,ALL
command. This will show TM=ETR, TM=SIMETR, or TM=STP.
For more information about STP, refer to Server Time Protocol Planning Guide, SG24-7280.
2.7 Miscellaneous commands and displays

This section describes additional commands that are useful in managing a Parallel Sysplex.
32
2.7.1 Determining the command prefixes in your sysplex

Command prefixes allow a subsystem (like JES2 or DB2) to create unique command prefixes
for each copy of the subsystem in the sysplex, and to control which systems can accept the
subsystem commands for processing.
Use the D OPDATA command (or D O), as shown in Figure 2-31, to obtain information about the
command prefixes that are defined in your sysplex. The SCOPE column indicates whether the
prefix has a destination of only the indicated system, or if it applies to commands issued from
any system in the sysplex.
D O
IEE603I 19.41.15 OPDATA DISPLAY 604
PREFIX
OWNER
SYSTEM
SCOPE
$
JES2
#@$3
SYSTEM
$
JES2
#@$2
SYSTEM
$
JES2
#@$1
SYSTEM
%
RACF
#@$3
SYSTEM
%
RACF
#@$2
SYSTEM
%
RACF
#@$1
SYSTEM
#@$1
IEECMDPF
#@$1
SYSPLEX
#@$2
IEECMDPF
#@$2
SYSPLEX
#@$3
IEECMDPF
#@$3
SYSPLEX
REMOVE
NO
NO
NO
NO
NO
NO
YES
YES
YES
FAILDSP
SYSPURGE
SYSPURGE
SYSPURGE
PURGE
PURGE
PURGE
SYSPURGE
SYSPURGE
SYSPURGE
Figure 2-31 Display sysplex prefixes by OPDATA
2.7.2 Determining when the last IPL occurred

To obtain the date, time, and other useful information about the last IPL, use the D IPLINFO
command as shown in Figure 2-32.
D IPLINFO
IEE254I 19.42.55 IPLINFO DISPLAY 609
SYSTEM IPLED AT 21.59.10 ON 06/22/2007
RELEASE z/OS 01.08.00
LICENSE = z/OS
USED LOADSS IN SYS0.IPLPARM ON 1D00
ARCHLVL = 2 MTLSHARE = N
IEASYM LIST = FK
IEASYS LIST = (FK,FK) (OP)
IODF DEVICE 1D00
IPL DEVICE 1D0C VOLUME #@$#R3
1
2
3
4
5
5
6
7
Figure 2-32 Display IPLINFO
This display output shows the following information:

1 The date and time of the IPL (in mm/dd/yyyy format)
2 The z/OS release level of your system
3 LOADxx information used for the IPL
4 64-bit addressing mode and MTL tape device parms
5 The suffixes of the IEASYSxx and IEASYMxx members
6 The IODF Device address
7 The IPL Device address and volser
33
2.7.3 Determining which IODF data set is being used

Use the command D IOS,CONFIG, shown in Figure 2-33, to determine the name of the IODF
data set 1 that contains the active I/O configuration definition.
D IOS,CONFIG
IOS506I 19.58.45 I/O CONFIG DATA 629
ACTIVE IODF DATA SET = IODF.IODF59 1
CONFIGURATION ID = TRAINER
EDT ID = 01
TOKEN: PROCESSOR DATE
TIME
DESCRIPTION
SOURCE: SCZP901 07-06-19 16:37:47 SYS6
IODF07
ACTIVE CSS: 1
SUBCHANNEL SETS IN USE: 0
CHANNEL MEASUREMENT BLOCK FACILITY IS ACTIVE
Figure 2-33 Display IOS config data
In this case, the active IODF data set 1 is IODF.IODF59.
2.8 Routing commands through the sysplex

There are several ways to process a command in one or more systems in your sysplex.
To route a command to one other system in the sysplex, use only its system name from one
of the other systems in the sysplex as shown in Figure 2-34.
RO #@$1,D IPLINFO
IEE254I 21.39.41 IPLINFO DISPLAY 100
SYSTEM IPLED AT 22.48.44 ON 06/22/2007
RELEASE z/OS 01.07.00
LICENSE = z/OS
...
Figure 2-34 Route to one system
To route commands to multiple systems in your sysplex, there are several options available:
Two system names can be enclosed in parentheses ( ), as shown in Figure 2-35 on
page 35. This method was issued in system #@$3, and shows the response from system
#@$1 1 and system #@$2 2.
A group name can be defined by the system programmer in IEEGSYS in SYS1.SAMPLIB
with a combination of any desired system names.
A combination of group and system names can be used, enclosed in parentheses.
34
RO (#@$1,#@$2),D TS,L
IEE421I RO (LIST),D TS,L 807
#@$1 1
RESPONSES --------------------------------------------------IEE114I 21.59.40 2007.175 ACTIVITY 107
JOBS
M/S
TS USERS
SYSAS
INITS
ACTIVE/MAX VTAM
OAS
00004
00017
00002
00032
00016
00002/00030
00004
SMITH
OWT
JONES
OWT
#@$2 2 RESPONSES --------------------------------------------------IEE114I 21.59.40 2007.175 ACTIVITY 623
JOBS
M/S
TS USERS
SYSAS
INITS
ACTIVE/MAX VTAM
OAS
00001
00010
00001
00032
00016
0002/00030
00004
SOUTH
OWT
NORTH
OWT
Figure 2-35 Route to a group of systems
To route commands to all other systems in your sysplex, use the *OTHER parm as shown in
Figure 2-36. This was issued in #@$1, so the response shows systems #@$2 and #@$3.
RO *OTHER,D TS,L
#@$2
JOBS
M/S
TS USERS
SYSAS
INITS
ACTIVE/MAX VTAM
OAS
00001
00011
00001
00032
00016
00002/00030
00004
SOUTH
OWT
NORTH
OWT
#@$3
JOBS
M/S
TS USERS
SYSAS
INITS
ACTIVE/MAX VTAM
OAS
00001
00014
00004
00032
00021
00002/00030
00005
EAST
OWT
WEST
OWT
Figure 2-36 Route to all other systems
To route commands to all systems in your sysplex, see Figure 2-37.

RO *ALL,D U,IPLVOL
IEE421I RO *ALL,D U,IPLVOL
#@$1
RESPONSES ------------------------------IEE457I 21.50.24 UNIT STATUS 123
UNIT TYPE STATUS
VOLSER
VOLSTATE
A843 3390 S
#@$#R1
PRIV/RSDNT
#@$2
UNIT TYPE STATUS
VOLSER
VOLSTATE
1D0C 3390 S
#@$#R3
PRIV/RSDNT
#@$3
UNIT TYPE STATUS
VOLSER
VOLSTATE
1D0C 3390 S
#@$#R3
PRIV/RSDNT
Figure 2-37 Route to all systems
There may be another way, however. If IEECMDPF (an IBM-supplied sample program in
SYS1.SAMPLIB) has run at IPL time, it defines the system name as a command prefix that
substitutes for the ROUTE command on each system.
35
For example, the following commands have the same effect on each system in the sysplex:
ROute #@$1,command
#@$1command
2.9 System symbols

System symbols are a very powerful capability in z/OS that can significantly ease the work
involved in managing and administering a sysplex. Symbols can be used in PARMLIB
members, batch jobs, VTAM definitions, system commands, and others.
The D SYMBOLS command shows the value of the system symbols for the system the
command is issued on. Figure 2-38 shows the response to the command.
D SYMBOLS
IEA007I STATIC SYSTEM
&SYSALVL. =
&SYSCLONE. =
&SYSNAME. =
&SYSPLEX. =
&SYSR1.
=
&BPXPARM. =
&CACHEOPT. =
&CICLVL.
=
&CLOCK.
=
&COMMND.
=
&DBDLVL.
=
. . .
SYMBOL VALUES 248

"2"
"$3"
"#@$3"
"#@$#PLEX"
"#@$#R3"
"FS"
"NOCACHE"
"V31LVL1"
"VM"
"00"
"V8LVL1"
Figure 2-38 Display static system symbol values
2.10 Monitoring the sysplex through TSO

SDSF has panels that can assist in monitoring the sysplex, as highlighted here. This topic is
covered in more depth in Chapter 11, System Display and Search Facility and OPERLOG
on page 231.
Multi-Access Spool
This SDSF Multi-Access Spool (MAS) panel displays the members of the Multi-Access Spool,
as shown in Figure 2-39.
SDSF MAS DISPLAY #@$1 XCFJES2A 79% SPOOL
COMMAND INPUT ===>
PREFIX=* DEST=(ALL) OWNER=* SYSNAME=*
NP NAME Status
SID PrevCkpt
Hold
#@$1 ACTIVE
1
0.58
0
#@$2 ACTIVE
2
0.63
0
#@$3 ACTIVE
3
0.76
0
Figure 2-39 MAS display
36
LINE 1-3 (3)

SCROLL ===> PAGE
ActHold
0
0
0
Dormancy
(0,100)
(0,100)
(0,100)
ActDorm SyncT
100
1
100
1
101
1
Job classes
The SDSF JC command displays the JES-managed and WLM-managed job classes, as
shown in Figure 2-40.
JOB CLASS DISPLAY ALL CLASSES
COMMAND INPUT ===>
PREFIX=* DEST=(ALL) OWNER=* SYSNAME=*
NP CLASS
Status Mode Wait-Cnt Xeq-Cnt
A
NOTHELD JES
B
NOTHELD JES
C
NOTHELD JES
D
NOTHELD WLM
E
NOTHELD WLM
LINE 1-35 (38)

SCROLL ===> PA
Hold-Cnt ODisp
(,)
(,)
(,)
1 (,)
(,)
QHld
NO
NO
NO
NO
NO
Ho
NO
NO
NO
NO
NO
Figure 2-40 Job Classes display
WLM resources
The SDSF RES command displays the WLM-managed resources, as shown in Figure 2-41.
SDSF RESOURCE DISPLAY
COMMAND INPUT ===>
PREFIX=* DEST=(ALL)
NP RESOURCE
CB390ELEM
DB2_PROD
PRIME_SHIFT
MAS SYSTEMS
OWNER=*
#@$1
RESET
RESET
RESET
LINE 1-3 (3)

SCROLL ===> PAGE
SYSNAME=*
#@$2
#@$3
RESET
RESET
RESET
RESET
RESET
RESET
Figure 2-41 WLM-managed resources
37
38
Chapter 3.
IPLing systems in a Parallel

Sysplex
This chapter explains the process of IPLing a z/OS system image so that it can join a Parallel
Sysplex.
The chapter concentrates on three areas:
IPLing the first system image in a Parallel Sysplex
IPLing an additional system image in a Parallel Sysplex
Possible IPL problems in a Parallel Sysplex
39
3.1 Introduction to IPLing systems in a Parallel Sysplex

Having a Parallel Sysplex introduces a number of differences in the IPL process, compared to
base sysplex or non-sysplex environments. This chapter highlights the differences by
showing the operator messages and activities that differ within a Parallel Sysplex.
As mentioned, the focus is on the following three areas:
IPLing the first system image in a Parallel Sysplex
Possible IPL problems in a Parallel Sysplex
Important: Visibility to the logs, actions, and messages shown here may depend on your
installation and your type of console, whether 3x74, 2074, OSA-ICC, or HMC.
3.2 IPL overview

When IPLing an image into a Parallel Sysplex, there are four significant stages for operators:
Performing the load on the processor

z/OS initialization
Subsystem restart
Subsystem workload restart
The descriptions in this chapter follow the system to the point where the z/OS initialization is
completed and the system is ready for the restart of subsystems and their workload. For
details of these stages, refer to this book's chapters on specific subsystems.
Before IPLing:
To avoid receiving messages IXC404I and IXC405D (indicating that other systems are
already active in the sysplex), the first system IPLed back into the sysplex should
preferably also have been the last one removed from the sysplex.
We do not recommend that you IPL additional systems at the same time as the first
system IPL. Wait until GRS initializes, as indicated by message ISG188I (Ring mode)
or message ISG300I (Star mode).
We also recommend that you try to avoid IPLing multiple systems from the same
physical sysres at the same time, to avoid possible sysres contention.
After the load has been performed, z/OS runs through four stages during initialization. These
stages are:
Nucleus Initialization Program (NIP)
Acquiring the Time Of Day (TOD) from the CEC TOD clock, and verifying that the CEC is
in the time mode indicated in the CLOCKxx member
z/OS joining the Parallel Sysplex
XCF initialization
Coupling Facility (CF) connection
Global Resource Serialization (GRS) initialization
Console initialization
40
The z/OS part of the IPL is complete when message IEE389I informs you that z/OS
command processing is available. At this point, the operator can concentrate on the restart of
the subsystems such as JES2, IMS, DB2, and CICS. It is preferable to use an automated
operations package to perform some of this activity.
3.2.1 IPL scenarios

The following sections describe three different IPL scenarios:
IPLing the first system image in a Parallel Sysplex that was the last one out
IPLing the first system image in a Parallel Sysplex that was not the last one out
The subsequent section describing IPL problems details these scenarios:
Maximum number of systems reached
COUPLExx parmlib member syntax errors
No CDS specified
Wrong CDS names specified
Mismatched timer references
Unable to establish XCF connectivity
IPLing the same system name
Sysplex name mismatch
IPL using wrong GRS options
3.3 IPLing the first system image (the last one out)
This section describes how the first image, in our example #@$1, is IPLed into a Parallel
Sysplex called #@$#PLEX. This means that no other system images are active in the Parallel
Sysplex prior to this IPL taking place. Prior to the IPL, all systems were stopped in an orderly
manner, with #@$1 being the last system to be stopped.
.
Note: As previously noted, IPLing the first system in a Parallel Sysplex should not be done
concurrently with other systems.
The description follows the sequence of events from the processor load through to the
completion of z/OS initialization.
3.3.1 IPL procedure for the first system

The activities that the operator performs, from issuing the load to full z/OS initialization, are
usually minimal if the sysplex was shut down cleanly before the IPL. However, the events are
highlighted in this section to allow familiarization with the IPL.
Nucleus Initialization Program processing

The IPL displays the Nucleus Initialization Program (NIP) messages on the console. They are
recorded in the syslog, along with the master catalog selection, z/OS system symbol values,
and page data set allocation.
Chapter 3. IPLing systems in a Parallel Sysplex
41
The messages seen will depend on the second last character of load parameter specified
during IPL. This is the Initial Message Suppression Indicator (IMSI). It can be coded to
suppress most informational messages and to not prompt for system parameters.
Figure 3-1 shows the NIP messages related to system #@$1 being IPLed.
IEA371I SYS0.IPLPARM ON DEVICE 1D00 SELECTED FOR IPL PARAMETERS
IEA246I LOAD
ID SS SELECTED
IEA246I NUCLST ID $$ SELECTED
IEA519I IODF DSN = IODF.IODF59 1
IEA520I CONFIGURATION ID = TRAINER . IODF DEVICE NUMBER = 1D00
IEA528I IPL IODF NAME DOES NOT MATCH IODF NAME IN HARDWARE TOKEN
SYS6.IODF07
IEA091I NUCLEUS 1 SELECTED
IEA093I MODULE IEANUC01 CONTAINS UNRESOLVED WEAK EXTERNAL REFERENCE
IECTATEN
IEA370I MASTER CATALOG SELECTED IS MCAT.V#@$#M1
IST1096I CP-CP SESSIONS WITH USIBMSC.#@$1M ACTIVATED
IEE252I MEMBER IEASYMFK FOUND IN SYS1.PARMLIB
IEA008I SYSTEM PARMS FOLLOW FOR z/OS 01.07.00 HBB7720 013 2
IEASYSFK
IEASYSFK
IEE252I MEMBER IEASYS00 FOUND IN SYS1.PARMLIB
IEE252I MEMBER IEASYSFK FOUND IN SYS1.PARMLIB
IEA007I STATIC SYSTEM SYMBOL VALUES 018 3
&SYSALVL. = "2"
&SYSCLONE. = "$1"
&SYSNAME. = "#@$1"
&SYSPLEX. = "#@$#PLEX"
&SYSR1.
= "#@$#R1"
&BPXPARM. = "FS"
&CICLVL.
= "V31LVL1"
&CLOCK.
= "VM"
&COMMND.
= "00"
&LNKLST.
= "C0,C1"
&LPALST.
= "00,L"
&MQSLVL1. = "V60LVL1"
&OSREL.
= "ZOSR17"
&SMFPARM. = "00"
&SSNPARM. = "00"
&SYSID1.
= "1"
&SYSNAM.
= "#@$1"
&SYSR2.
= "#@$#R2"
&VATLST.
= "00"
&VTAMAP.
= "$1"
IFB086I LOGREC DATA SET IS SYS1.#@$1.LOGREC 045
IEE252I MEMBER GRSCNF00 FOUND IN SYS1.PARMLIB
IEE252I MEMBER GRSRNL02 FOUND IN SYS1.PARMLIB
IEA940I THE FOLLOWING PAGE DATA SETS ARE IN USE:
PLPA ........... - PAGE.#@$1.PLPA
COMMON ......... - PAGE.#@$1.COMMON
LOCAL .......... - PAGE.#@$1.LOCAL1 .....
Figure 3-1 IPL NIP phase display
42
The messages in Figure 3-1 on page 42 include information that is valuable to the operator:
1 The IODF DSN (and below it, the device number)
2 The version of z/OS that is being IPLed
3 The system symbol name and their values
The usual z/OS library loading and concatenation messages then follow; Figure 3-2 shows an
example of the messages displayed during this phase. The usual pause in the IPL message
flow follows at this point, while the LPA is built.
IEE252I MEMBER LPALST00 FOUND IN SYS1.PARMLIB
IEA713I LPALST LIBRARY CONCATENATION
SYS1.LPALIB
Figure 3-2 IPL LPA library concatenation
Time-Of-Day (TOD) clock setting

The next milestone in the IPL process is when the system reaches the point where the
system TOD clock is initialized. Any offset from the system TOD value is applied.
Part of this processing involves checking that this system is attached to the same time source
(Sysplex Timer or Server Time Protocol) as all the other members of the sysplex.
If all systems in the sysplex are run in LPARs on a single CEC, then a simulated timer
(SIMETR) can be used.
Figure 3-3 shows the message for the time zone setting.
.
IEA598I TIME ZONE = W.04.00.00

Figure 3-3 Time zone offset setting
Joining (initializing) the sysplex

At this point in the IPL, XCF checks to see if there are active members in the sysplex being
joined. Depending on how the sysplex was stopped, the sysplex CDS may show that other
members of the sysplex are still active. However, for this example, the systems were shut
down cleanly, and the system being IPLed was the last one shut down, so there are no other
systems that show a status of ACTIVE in the sysplex CDS. As a result, no alert is issued and
the IPL proceeds.
XCF next starts its links to the CF. As each link successfully connects to the CF, message
IXL157I is issued, as shown in Figure 3-4 on page 44.
43
IXL157I PATH 09 IS NOW OPERATIONAL TO CUID: 0309 103

IXL157I PATH 0E IS NOW OPERATIONAL TO CUID: 0309 104
IXL157I PATH 0F IS NOW OPERATIONAL TO CUID: 030F 105
IXL157I PATH 10 IS NOW OPERATIONAL TO CUID: 030F 106
Figure 3-4 Sysplex channel connection
When XCF initialization is complete, message IXC418I is issued, as shown in Figure 3-5.
This indicates that the system is now part of the named sysplex.
IXC418I SYSTEM #@$1 IS NOW ACTIVE IN SYSPLEX #@$#PLEX
Figure 3-5 System active in sysplex
Note: The IXC4181 message is easy to miss because it occurs around the same time as
the PATHIN and PATHOUT activity, which can generate a large number of messages.
PATHIN and PATHOUT activation

To be able to communicate with the other XCFs in the sysplex, XCF now starts the PATHINs
and PATHOUTs that are defined in the COUPLExx member. PATHIN and PATHOUT
activation consists of attempting to start all PATHINs and PATHOUTs using the CF structures
or CTCs.
Connectivity is achieved if:
The associated CTC is online or the CF containing the structure is started and accessible.
The link they support is to an active system.
The system at the other end of the PATHIN or PATHOUT has its corresponding
PATHOUT or PATHIN started.
Because this is the first system in the sysplex, the second and third requirements in this list
will not be met. As a result, the start command will fail and the system issues message
IXC305I 1, as shown in Figure 3-6 on page 45. The system then issues a stop command for
the PATHIN or PATHOUT devices or structures which could not be started. The stop
command messages for a structure are IXC467I 2, followed by IXC307I 3.
The desired structure is then created and allocated. When the start command to a device or
structure succeeds, message IXC306I 4 is issued, as shown in Figure 3-6 on page 45.
Following are examples of each message type. There are no IXC466I messages shown
(which would indicate that connectivity to another system has been established). Connectivity
is not possible to other systems because they are not yet in the sysplex.
44
IXC305I START PATHOUT REQUEST FOR STRUCTURE IXC_DEFAULT_2 103

WAS NOT SUCCESSFUL:
DELAYED UNTIL AN ACTIVE SYSTEM ALLOCATES STRUCTURE
DIAG073:086F0004 00000002 00000008 00071007 00000000
IXC467I STOPPING PATHIN STRUCTURE IXC_DEFAULT_2 104
RSN: START REQUEST FAILED
DIAG073: 086F0004 00000002 00000008 00071007 00000000
IXC307I STOP PATHOUT REQUEST FOR STRUCTURE IXC_DEFAULT_2 106
COMPLETED SUCCESSFULLY: START REQUEST FAILED
...
IXC582I STRUCTURE IXC_DEFAULT_2 ALLOCATED BY SIZE/RATIOS.
...
IXC306I START PATHOUT REQUEST FOR STRUCTURE IXC_DEFAULT_2 148
COMPLETED SUCCESSFULLY: COUPLING FACILITY RESOURCES AVAILABLE
IXC306I START PATHIN REQUEST FOR STRUCTURE IXC_DEFAULT_2 149
COMPLETED SUCCESSFULLY: COUPLING FACILITY RESOURCES AVAILABLE
2
3
Figure 3-6 Starting PATHs to devices and structures
When the other systems are IPLed and start their end of the paths, communication will be
possible with other systems in the Parallel Sysplex.
Couple Data Set addition and structure allocation

The function Couple Data Sets are added progressively during Parallel Sysplex activation.
This is shown by message IXC286I for each CDS. Depending on your installation, it may
include any or all of the CFRM, SFM, BPXMCDS, WLM, LOGR, and ARM Couple Data Sets.
Figure 3-7 shows the messages for the CFRM Couple Data Sets.
IXC286I COUPLE DATA SET
SYS1.XCF.CFRM01,
VOLSER #@$#X2, HAS BEEN
FOR CFRM ON SYSTEM #@$1
IXC286I COUPLE DATA SET
SYS1.XCF.CFRM02,
VOLSER #@$#X1, HAS BEEN
FOR CFRM ON SYSTEM #@$1
128
ADDED AS THE PRIMARY
129
ADDED AS THE ALTERNATE
Figure 3-7 Adding Couple Data Sets
When the SFM CDS is added, if the SFM policy is active, it generates messages IXC602I,
IXC609I, and IXC601I. These indicate which policy is loaded and which attributes are used.
These messages are shown in Figure 3-8.
IXC602I SFM POLICY SFM01 INDICATES FOR SYSTEM #@$1 A STATUS 060
UPDATE MISSING ACTION OF ISOLATE AND AN INTERVAL OF 0 SECONDS.
THE ACTION WAS SPECIFIED FOR THIS SYSTEM.
IXC609I SFM POLICY SFM01 INDICATES FOR SYSTEM #@$1 A SYSTEM WEIGHT OF
80 SPECIFIED BY SPECIFIC POLICY ENTRY
IXC601I SFM POLICY SFM01 HAS BEEN MADE CURRENT ON SYSTEM #@$1
Figure 3-8 Adding SFM policy
45
During the process of starting XCF, XCF detects that you are IPLing this system as the first in
the sysplex, and verifies with each CF that the CF contains the same structures that the
CFRM indicates are present in the CF. This process is known as reconciliation. This activity,
shown in Figure 3-9, generates messages IXC504I, IXC505I, IXC506I and IXC507I, as
appropriate.
This happens regardless of whether the CFs were restarted. As additional systems are IPLed
into the sysplex, they detect that there are active systems and bypass this step.
IXC504I INCONSISTENCIES BETWEEN COUPLING FACILITY NAMED FACIL01 451
AND THE CFRM ACTIVE POLICY WERE FOUND.
THEY HAVE BEEN RESOLVED.
...
IXC505I STRUCTURE JES2CKPT_1 IN 442
PARTITION: 00
CPCID: 00
NOT FOUND IN COUPLING FACILITY. CFRM ACTIVE POLICY CLEARE
...
TRACE THREAD: 00022E92.
IXC507I CLEANUP FOR 452
PARTITION: 00
CPCID: 00
HAS COMPLETED.
TRACE THREAD: 00022E92.
Figure 3-9 CFRM initialization
When the reconciliation process completes, the system is able to use each of the CFs, as
confirmed by message IXC517I shown in Figure 3-10.
IXC517I SYSTEM #@$1 ABLE TO USE 123
PARTITION: 00
CPCID: 00
NAMED FACIL01
Figure 3-10 CF connection confirmation
While the Couple Data Sets are added, any requested allocations of structures will also take
place. This is indicated with the IXL014I and IXL015I messages, as shown in Figure 3-11 on
page 47.
These messages may occur for z/OS components such as XCFAS, ALLOCAS, and
IXGLOGR, depending on which functions your installation is exploiting. The messages tell
you in which CF the structure was allocated and why.
46
IXL014I IXLCONN REQUEST FOR STRUCTURE ISGLOCK 134

WAS SUCCESSFUL. JOBNAME: GRS ASID: 0007
CONNECTOR NAME: ISGLOCK##@$1 CFNAME: FACIL01
IXL015I STRUCTURE ALLOCATION INFORMATION FOR 242
STRUCTURE I#$#OSAM, CONNECTOR NAME IXCLO0180001
CFNAME
ALLOCATION STATUS/FAILURE REASON
---------------------------------------FACIL02
STRUCTURE ALLOCATED CC007800
FACIL01
PREFERRED CF ALREADY SELECTED CC007800
Figure 3-11 Structure allocation
GRS complex initialization

During this stage, the system initializes the GRS complex. The type of GRS configuration that
you will be using (Ring or Star) is coded on the GRS keyword in the IEASYSxx member.
Figure 3-12 shows the message that you will receive when GRS comes up in Star mode.
ISG300I GRS=STAR INITIALIZATION COMPLETE FOR SYSTEM #@$1
Figure 3-12 GRS initialization complete
CONSOLE activation
The CONSOLxx member is processed next. Using the information obtained in the member,
Multiple Console Support (MCS) is activated. The system initializes the system console as an
extended console. See Figure 3-13 for the type of messages you can expect to see during
this processing.
Note: Starting with z/OS 1.8, the sysplex master console is no longer supported or
required.
IEE252I
IEA630I
IEE828E
IEA549I
MEMBER CONSOL00 FOUND IN SYS1.PARMLIB

OPERATOR #@$1
NOW ACTIVE,
SYSTEM=#@$1
SOME MESSAGES NOW SENT TO HARDCOPY ONLY
SYSTEM CONSOLE FUNCTIONS AVAILABLE 204
SYSTEM CONSOLE NAME ASSIGNED #@$1
IEA630I OPERATOR *SYSLG$1 NOW ACTIVE,
SYSTEM=#@$1
IEA630I OPERATOR *ROUTE$1 NOW ACTIVE,
SYSTEM=#@$1
...
, LU=#@$1
, LU=*SYSLG$1
, LU=ROUTEALL
Figure 3-13 Console activation
For more information about consoles, refer to Chapter 14, Managing consoles in a Parallel
Sysplex on page 283.
RACF initialization
The next step in the IPL process occurs as RACF starts using its databases. The system
programmers should have customized the RACF data set name table module to indicate
whether RACF sysplex communication is to be enabled. If sysplex communication is enabled,
RACF will automatically use the RACF structures in the CF to share the database between
the systems that are sharing that RACF database.
47
The RACF data-sharing operation mode in the Parallel Sysplex is indicated by the messages
shown in Figure 3-14. Note that some installations may use a security product other than
RACF.
ICH559I MEMBER #@$1 ENABLED FOR SYSPLEX COMMUNICATIONS
Figure 3-14 RACF initialization
Additional sysplex exploiters

Further messages may be generated relating to Parallel Sysplex components. The specific
messages that you see will depend on which sysplex mode options that have been
implemented at your installation (for example, LOGGER and ARM).
z/OS command processing

The formal MVS IPL process is considered complete when the IEE389I message is issued,
as shown in Figure 3-15. Even though the core MVS IPL process may be complete, the
greater IPL process, which includes initializing subsystems such as JES and VTAM and
TCP/IP, will continue at this point.
IEE389I MVS COMMAND PROCESSING AVAILABLE
Figure 3-15 MVS command processing available
3.4 IPLing the first system image (not the last one out)
This section describes the IPL process when the first system to be IPLed in the sysplex is not
the last one to be stopped when the sysplex was shut down. This means that no other system
images are active in the Parallel Sysplex prior to this IPL taking place.
During initialization, XCF checks the CDS for systems other than the one being IPLed. In this
situation, that condition is met.
For example, if #@$1 was the last to leave the sysplex, and you IPL system #@$2 first, XCF
checks for systems other than the one being IPLed. If it finds #@$1 or #@$3 (which it does,
in this scenario), then it issues IXC404I, which lists the system names in question, and follows
it with IXC405D.
The procedure will depend on what happened before the IPL, and how the system or systems
left the sysplex.
If there was, for instance, an unplanned power outage and all systems failed at the same
time, then upon the first IPL of any system, IXC404I and IXC405D are issued.
Note: As previously mentioned, IPLing the first system in a Parallel Sysplex should not be
done concurrently with other systems. The cleanup of the sysplex CDSs and CF structures
is disruptive to other systems. Only the IPL of additional systems into the sysplex can run
concurrently, and we do not recommend having them in NIP at the same time.
48
3.4.1 IPL procedure for the first system

The overall sequence of events is the same as the previous sections, and is not repeated
here. Only the differences are shown.
Joining the sysplex

At this point in the IPL, z/OS detects a connection to the CFs. If the first system to be IPLed
(in this example, #@$2) is not the last system to have left the sysplex, it results in message
IXC404I, seen in Figure 3-16, identifying the system which XCF believes to be active (#@$1).
An IXC405D write to operator with reply (WTOR) message follows. It offers options I, J or R,
and waits for the operator response, as shown in Figure 3-16. These options are explained
here:
I
Use the I option to request that sysplex initialization continue because none of the
systems identified in message IXC404I are in fact participating in an operating
sysplex; that is, they are all residual systems. This system will perform cleanup of old
sysplex data, initialize the Couple Data Set, and start a new sysplex. If any of the
systems identified in message IXC404I are currently active in the sysplex, they will be
placed into a disabled wait state.
Use the J option to request that this system join the already active sysplex. Choose
this reply if this system belongs in the sysplex with the systems identified in message
IXC404I, despite the fact that some of those systems appear to have out-of-date
system status update times. The initialization of this system will continue.
Use the R option to request that XCF be reinitialized on this system. XCF will stop
using the current Couple Data Sets and issue message IXC207A to prompt the
operator for a new COUPLExx parmlib member.
Choose R also to change the sysplex name and reinitialize XCF to remove any
residual data for this system from the Couple Data Set. The system prompts the
operator for a new COUPLExx parmlib member.
Consult your support staff if necessary. If no other systems are in fact active, you can answer
I to initialize the sysplex. The alternative options (J or R) are only valid with an active sysplex
or to make changes to the XCF parameters.
IXC404I SYSTEM(S) ACTIVE OR IPLING: #@$1
IXC405D REPLY I TO INITIALIZE THE SYSPLEX, J TO JOIN SYSPLEX #@$#PLEX,
OR R TO REINITIALIZE XCF
IEE600I REPLY TO 00 IS;I
Figure 3-16 Initialize the sysplex
If I is replied, the IPL continues and message IXC418I is issued, indicating that this system is
now part of a sysplex; see Figure 3-17.
IXC418I SYSTEM #@$2 IS NOW ACTIVE IN SYSPLEX #@$#PLEX
Figure 3-17 System now in sysplex
The IXC4181 message is easy to miss because it occurs at the same time as PATHIN and
PATHOUT activity, which generates a large number of messages.
49
If you reply J in the preceding scenario, the system being IPLed would join the active
sysplex. However, after it is running, if the time stamps of the other systems do not become
updated, this system may consider them to be in status update missing (SUM) condition,
and may start partitioning the inactive systems out of the sysplex.
Note: Under certain circumstances, the sysplex becomes locked and message IXC420A
might be issued instead of IXC405A. Those circumstances may include during a disaster
recovery where the CF name has changed; if CDS specifications for the IPLing system do
not appear to match what the current sysplex is using; or if the Sysplex Timers do not
appear to be the same.
In these cases, using J to join is not an option. The only choices offered are I to initialize
the sysplex, or R to specify a new COUPLExx.
3.5 IPLing any system after any type of shutdown

This section describes how any image is IPLed in a Parallel Sysplex following any type of
shutdown. This means that at least one other system image is already active in the Parallel
Sysplex prior to this IPL taking place. You are not necessarily coming from a position of a
total shutdown or outage in this section.
Again, the actions depend on what happened before this IPL:
Whether this system was removed from XCF since it was shut down
Whether this system was not removed from XCF
Note that the overall sequence of events is the same as the previous sections and is not
repeated here. Only the differences are shown.
3.5.1 IPL procedure for any additional system in a Parallel Sysplex

TOD clock setting
The active system or systems in the sysplex will already be in their coded timer mode,
meaning in ETR, SIMETR, or STP mode. Any other systems trying to join should be coded
the same, or at least in a compatible manner.
Timer errors are shown in 3.6.5, Mismatching timer references on page 55.
Joining the sysplex

There are two scenarios that may have preceded any IPL into an active sysplex, as explained
here.
If the system was removed from XCF since it was stopped or shut down
If the system was removed, then there is no residual presence of this system in the sysplex,
and it will join without incident.
If this system was not removed from XCF

During stopping or shutdown of this additional system, there may have been replies left
unanswered in the remaining systems. Examples are shown in Figure 3-18 on page 51.
50
004 IXC102A XCF IS WAITING FOR SYSTEM #@$1 DEACTIVATION. REPLY DOWN
WHEN MVS ON #@$1 HAS BEEN SYSTEM RESET
006 IXC402D #@$1 LAST OPERATIVE AT 20:30:15.

REPLY DOWN AFTER SYSTEM RESET, OR INTERVAL=SSSSS
TO SET A REPROMPT TIME.
Figure 3-18 System requiring DOWN reply
If the WTOR remains outstanding during this attempted IPL, it means that neither the sysplex
partitioning nor the cleanup of the system that was brought down has been performed. This
partitioning and cleanup will have to be performed before the system can rejoin the sysplex at
IPL.
As shown in Figure 3-19, #$@3 has been varied offline, but the DOWN reply to IXC102A is
not given. When the system is re-IPLed, the IPLing system issues the following IXC203
message as it tries to join the sysplex (referring to its previous incarnation).
IXC203I #@$3 IS CURRENTLY ACTIVE IN THE SYSPLEX
IXC218I SYSTEM STATUS FOR SYSPLEX #@$#PLEX AT 06/27/2007 01:27:13:
#@$2
01:27:12 ACTIVE
#@$3
01:23:43 BEING REMOVED
IXC214I COUPLE00 IS THE CURRENT COUPLE PARMLIB MEMBER
IXC240I IF XCF-LOCAL MODE INITIALIZATION IS DESIRED, RE-IPL WITH
"PLEXCFG=XCFLOCAL" AND "COUPLE=**"
IXC207A XCF INITIALIZATION IS RESTARTED. RESPECIFY COUPLE SYSTEM
PARAMETER, REPLY COUPLE=XX.
Figure 3-19 Trying to join when already active
The outstanding DOWN reply, in another system, must be responded to. After replying, wait
for the cleanup. The cleanup activity is highlighted in the IXC105I message text on the
remaining systems, shown in Figure 3-20.
IXC105I SYSPLEX PARTITIONING HAS COMPLETED FOR #@$3
- PRIMARY REASON: OPERATOR VARY REQUEST
- REASON FLAGS: 000004
Figure 3-20 System cleanup after DOWN reply
When IXC105 is issued, reply with your correct COUPLExx member and the IPL will
continue. You can also choose to reinitiate the IPL.
PATHIN and PATHOUT activation

With at least one another system already in the sysplex, the IPLing system can now make
system-to-system connections. These are reflected by message IXC466I, as shown in
51
IXC466I OUTBOUND SIGNAL CONNECTIVITY ESTABLISHED WITH SYSTEM #@$1 669

VIA STRUCTURE IXC_DEFAULT_1 LIST 8
IXC466I OUTBOUND SIGNAL CONNECTIVITY ESTABLISHED WITH SYSTEM #@$1 670
IXC466I INBOUND SIGNAL CONNECTIVITY ESTABLISHED WITH SYSTEM #@$1 671
IXC466I INBOUND SIGNAL CONNECTIVITY ESTABLISHED WITH SYSTEM #@$1 672
Figure 3-21 IXC466 Signal connectivity to other systems
GRS complex initialization

The GRS configuration (RING or STAR) will already be established. The joining system must
be of the same type. An incompatible option will halt the IPL process. This topic is covered in
more detail in 3.6.9, IPL wrong GRS options on page 58.
3.6 IPL problems in a Parallel Sysplex

A number of possible IPL problems are unique to the Parallel Sysplex. The following sections
describe several different scenarios:
Maximum number of systems reached

COUPLExx parmlib member syntax errors
No CDS specified
Wrong CDS names specified
Mismatching timer references
Unable to establish XCF connectivity
IPLing the same system name
Sysplex name mismatch
IPL wrong GRS options
3.6.1 Maximum number of systems reached

When a system tries to IPL into a sysplex, but the sysplex already contains the maximum
number of systems specified, XCF ends initialization and issues the message shown in
Figure 3-22.
IXC202I SYSPLEX sysplex-name IS FULL WITH nnn SYSTEMS
Figure 3-22 Sysplex is full message
This message is followed by the IXC207A WTOR, as shown in Figure 3-23

IXC207A XCF INITIALIZATION IS RESTARTED. RESPECIFY COUPLE SYSTEM
PARAMETER, REPLY couple=xx.
Figure 3-23 Sysplex full suggested action
To determine the cause of the problem:

Check the maximum number of systems specified in the XCF CDS.
Check the number and status of the system images currently in the sysplex.
52
Enter the D XCF,COUPLE command to identify the maximum number of systems specified in
the sysplex. In the information displayed about the primary sysplex CDS, check the
MAXSYSTEM parameter 1 in Figure 3-24.
D XCF,COUPLE
SYSTEM #@$1 DATA
...
PRIMARY
DSN: SYS1.XCF.CDS01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
11/20/2002 16:27:24
1 3
100
(52)
203
(18)
Figure 3-24 MAXSYSTEM value
Enter the command D XCF,S,ALL to obtain the number and status of systems in the sysplex.
See Figure 3-25.
D XCF,S,ALL
SYSTEM
TYPE SERIAL LPAR STATUS TIME
#@$1
2084 6A3A
N/A 07/01/2007 23:30:27
#@$2
2084 6A3A
N/A 07/01/2007 23:30:24
#@$3
2084 6A3A
N/A 07/01/2007 23:30:23
SYSTEM STATUS
ACTIVE
ACTIVE
ACTIVE
TM=SIMETR
TM=SIMETR
TM=SIMETR
Figure 3-25 Current active systems
The action you take will depend on the cause of the problem, as described in Table 3-1.
Table 3-1 MAXSYSTEM suggested actions
Cause
Suggested action
The system being IPLed is trying to join the

wrong sysplex.
Reset the system and verify the load parameter. If it is incorrect,

correct it and reIPL. If it is correct, contact the systems programmer,
who should correct the sysplex parameter in the COUPLxx parmlib
member.
MAXSYSTEM has been reached, but one or

more of the systems are not in the Active state.
This could mean that the sysplex is waiting for
XCF to complete partitioning and cleanup.
Find and respond to any outstanding IXC102A or IXC402D

messages to have XCF complete partitioning and cleanup.
Reply to IXC207A to respecifiy the current COUPLExx parmlib

member.
MAXSYSTEM has been reached, but one or

more of the systems can be removed from the
sysplex to allow the new one to join in.
Use the V XCF,sysname,OFFLINE command to remove one or more

systems from the sysplex. Before you attempt this, check with the
systems programmer.
MAXSYSTEM has been reached, and none of

the systems can be removed from the sysplex
to allow for the new one to join in.
1. The systems programmer must format a new CDS to cope with a

larger number of systems.
2. Make the new CDS the alternate by using the SETCXCF
COUPLE,ACOUPLE,dsn command.
3. Switch the new alternate CDS to the primary using the SETCXCF
COUPLE,PSWITCH command.
53
3.6.2 COUPLExx parmlib member syntax errors

If a system is IPLed with a COUPLExx member containing syntax errors, the IPL will cease at
this point and the system could issue any of the messages in Figure 3-26, depending on what
the exact problem is, as shown here.
IXC205I SYNTAX ERROR IN COUPLExx: text (such as incorrect parentheses)
IXC206I THE COUPLExx text (such as incorrect keyword)
IXC211A SYNTAX ERROR IN COUPLE SYSTEM PARAMETER. REPLY COUPLE=XX

Figure 3-26 Couple data set syntax errors
The text should specify the nature of the error, in some cases specifying the column it
occurred in. If the COUPLExx member is the correct one for the IPL, the syntax errors must
be corrected before the IPL can complete.
If the IPLing system is using the same COUPLExx member as the existing system or
systems, then the COUPLExx must have been changed since they were IPLed. It could,
however, be using a different COUPLExx member.
If all systems in the sysplex are sharing their parmlib definitions, the systems programmer
should be able to log on to one of the active systems and correct the definitions from there.
When the definitions have been corrected, respond to the IXC201A WTOR with COUPLE=xx,
where xx is the suffix of the corrected COUPLExx member in the PARMLIB. You may choose
to start the IPL again.
If the problems cannot be corrected from another system, you must IPL the failing system in
XCF-local mode. Before you attempt this, check with the systems programmer. After e the
system has completed the IPL, the systems programmer will be able to analyze and correct
the problem.
Note: To IPL in XCF-local mode, we recommend that an installation maintains an alternate
COUPLExx member in the PARMLIB containing the definition COUPLE
SYSPLEX(LOCAL).
3.6.3 No CDS specified

XCF requires a primary sysplex CDS. When the system is IPLing, if XCF does not find any
CDS specified in the COUPLExx member, but the PLEXCFG parameter indicates a
monoplex or multisystem configuration, the messages shown in Figure 3-27 on page 55 are
issued.
54
IXC212I SYSTEM WILL CONTINUE IPLING IN XCF-LOCAL MODE. NO PCOUPLE

KEYWORD OR PRIMARY DATA SET NAME WAS SPECIFIED IN THE COUPLE00 PARMLIB
MEMBER.
IXC412I SYSPLEX CONFIGURATION IS NOT COMPATIBLE WITH REQUIRED CONFIGURATION
IXC413I MONOPLEX SYSPLEX CONFIGURATION PREVENTED BY PLEXCFG=MULTISYSTEM
IXC413I XCFLOCAL SYSPLEX CONFIGURATION PREVENTED BY PLEXCFG=MULTISYSTEM
IXC207A XCF INITIALIZATION IS RESTARTED. RESPECIFY COUPLE SYSTEM PARAMETER,
REPLY COUPLE=XX.
Figure 3-27 No CDS specified
When the definitions are corrected, respond to the IXC201A WTOR with COUPLE=xx, where
xx is the suffix of the corrected COUPLExx member in the PARMLIB. You may choose to
start the IPL again.
3.6.4 Wrong CDS names specified

For this particular exercise, primary and alternate sysplex data set names were changed from
CDS01 and CDS02 to CDS03 and CDS04. The resolution messages shown in Figure 3-27
were received.
IXC268I THE COUPLE DATA SETS SPECIFIED IN COUPLE00 ARE IN
INCONSISTENT STATE
IXC275I COUPLE DATA SETS SPECIFIED IN COUPLE00 ARE 098
PRIMARY:
SYS1.XCF.CDS01
ON VOLSER #@$#X1
ALTERNATE: SYS1.XCF.CDS02
ON VOLSER #@$#X2
IXC273I XCF ATTEMPTING TO RESOLVE THE COUPLE DATA SETS
IXC275I RESOLVED COUPLE DATA SETS ARE 100
PRIMARY:
SYS1.XCF.CDS03
ON VOLSER #@$#X1
ALTERNATE: SYS1.XCF.CDS04
ON VOLSER #@$#X2
Figure 3-28 Wrong CDS name specified
The error was resolved by the system and the current values were used. The IPL continued
successfully.
3.6.5 Mismatching timer references

All systems must reference the same timing source. If the mode of an IPLing system is not
the same as the one being used by already active systems, an error will result.
In Figure 3-29 on page 56, the current SIMETRID was changed from 00 to 01 and system
#@$3 was IPLed. Message IXC406I was issued, followed by message IXC420D.
55
IXC406I THIS SYSTEM IS CONNECTED TO ETR NET ID=01. THE OTHER ACTIVE SYS
IN THE SYSPLEX ARE USING ETR NET ID=00.
IXC404I SYSTEM(S) ACTIVE OR IPLING: #@$1
#@$2
IXC419I SYSTEM(S) NOT SYNCHRONIZED: #@$1
#@$2
IXC420D REPLY I TO INITIALIZE SYSPLEX #@$#PLEX, OR R TO REINITIALIZE XC
REPLYING I WILL IMPACT OTHER ACTIVE SYSTEMS.
Figure 3-29 Mismatching SIMETRID
We used SIMETRID in this scenario because all three systems were on the same CEC. We
then commented out the SIMETRID parameter and message IEA261I was issued, as seen in
Figure 3-30. Under other circumstances, a different error messages may be issued.
IEA261I
IEA598I
IEA888A
IEA888A
NO ETR PORTS ARE USABLE. CPC CONTINUES TO RUN IN LOCAL MODE.

TIME ZONE = W.04.00.00
UTC
DATE=2007.198,CLOCK=01.26.56
LOCAL DATE=2007.197,CLOCK=21.26.56 REPLY U, OR UTC/LOCAL TIME
Figure 3-30 No SIMETRID parameter
3.6.6 Unable to establish XCF connectivity

When systems are IPLed into a sysplex, all systems' XCFs check the signalling paths in their
COUPLExx members. If XCF discovers there are not enough signalling paths available to
provide at least one inbound and one outbound path between each of the systems within the
sysplex, it issues error messages IXC454 and IXC453.
In this exercise, the PATHIN and PATHOUT names in member COUPLExx for an IPL of
system #@$2 were rendered invalid, either by using non-existent DSNs, or commenting them
out. The result is shown in Figure 3-31.
IXC454I SIGNALLING CONNECTIVITY CANNOT BE ESTABLISHED FOR SYSTEMS: #@$1
#@$3
IXC453I INSUFFICIENT SIGNALLING PATHS AVAILABLE TO ESTABLISH CONNECTIVITY
REPLY COUPLE=XX.
Figure 3-31 Insufficient signalling paths
System initialization stops. The operator and the systems programmer should check the
following areas to establish the cause of the problem.
Has any system in the sysplex issued message IXC451I, indicating invalid signalling
paths?
In the COUPLExx member in PARMLIB:
Are all systems using the same CDS?
Are there any incorrect or missing CF signalling structure definitions?
Do the signalling path definitions in the IPLing system match their corresponding
PATHINs and PATHOUTs in the other systems in the sysplex configuration?
56
Are the signalling path definitions consistent with the hardware configuration?
Are the CF signalling structures able to allocate the storage they require (check for
IXL013I messages)?
Are there any hardware failures?
The action taken will depend on the cause of the problem.
3.6.7 IPLing the same system name

There are two slightly different scenarios tested here:
IPL a system that is not removed from XCF.
IPL a system that is the same name as one already fully up.
The first scenario could occur for many reasons, and the exact result would depend upon
such factors as SFM being active; the sequence and timing of commands; whether the
system was shut down but not removed; or even it were IPLed over the top. It would also
depend on if SFM was set to PROMPT or not.
In general, however, when a system tries to IPL when not removed from the sysplex and it
gets to the stage of trying to join the sysplex, it issues message IXC203I as shown in
Figure 3-32.
IXC203I sysname IS CURRENTLY {ACTIVEIPLING} IN THE SYSPLEX
Figure 3-32 IXC203I system currently active in sysplex
This scenario is covered in Figure 3-19 on page 51.

Trying to IPL a system of the same name as one that is already fully up is in theory quite
unlikely to happen because the changes needed may involve page data sets and parmlib.
By changing the IPLPARM of #@$2 to resemble #@$3, the messages shown in Figure 3-33
were received, which are the same as the previous scenario.
...
REPLY COUPLE=XX.
Figure 3-33 IPL a system of the same name
3.6.8 Sysplex name mismatch

If the sysplex name in COUPLExx does not match the name of the sysplex that the IPLing
system is joining, then message IXC255 is issued as shown in Figure 3-34 on page 58.
57
IXC255I UNABLE TO USE DATA SET

SYS1.XCF.CDS01
AS THE PRIMARY FOR SYSPLEX:
SYSPLEX NAME #@$#PLEX DOES NOT MATCH THE SYSPLEX NAME IN USE
IXC255I UNABLE TO USE DATA SET
SYS1.XCF.CDS02
AS THE PRIMARY FOR SYSPLEX:
SYSPLEX NAME #@$#PLEX DOES NOT MATCH THE SYSPLEX NAME IN USE
IXC272I XCF WAS UNABLE TO RESOLVE THE COUPLE DATA SETS
REPLY COUPLE=XX.
Figure 3-34 Sysplex name mismatch
The exact cause is not obvious. The initial text is that the sysplex data sets cannot be used,
then the reason given is because of the sysplex name mismatch, saying it does not match the
one in use. The one in use does not refer to the name of the currently running sysplex, it
refers to the name of the one attempting to IPL.
The resolution is also not obvious. Correcting the sysplex name and replying to IXC207 with
the couple member is not effective because the IPL is past the stage where it picks up the
sysplex name. The IPL has to be re-initiated after correction of the sysplex name. That may
involve, for example, specifying a new name, or specifying a new loadparm.
3.6.9 IPL wrong GRS options

Whether the system is using a RING or a STAR is coded into the IEASYSxx member. If a
type that differs from the existing systems is attempting to IPL, message ISG307 is issued. In
this case, with an existing STAR config, the member of #@$2 was changed to RING, and
re-IPLed. The result is shown in Figure 3-35.
ISG307W GRS=TRYJOIN IS INCONSISTENT WITH THE CURRENT STAR COMPLEX.
Figure 3-35 GRS RING or STAR option wrong
The result is the same if the scenario is reversed; that is, if a STAR tries to join a RING. The
IPL stops and a non-restartable 0A3 wait state is loaded. Correct the parms and re-IPL.
GRS Resource Name List (RNL) mismatch

The GRSRNL parm is also coded in IEASYSxx. The list of resources must not only contain
the same names as the other systems, but in the same order as well. A mismatch produces
message ISG312, shown in Figure 3-36
ISG312W GRS INITIALIZATION ERROR. SYSTEMS EXCLUSION RNL MISMATCH
Figure 3-36 GRSRNL mismatch
The IPL stops and a non-restartable 0A3 wait state is loaded. Correct the parms and re-IPL.
58
Chapter 4.
Shutting down z/OS systems in a

Parallel Sysplex
This chapter explains how to shut down a z/OS system image in a Parallel Sysplex.
This chapter concentrates on three areas:
Shutdown overview
Removing z/OS systems from a Parallel Sysplex
Running a stand-alone dump (SAD) on a Parallel Sysplex
59
4.1 Introduction to z/OS system shutdown in a Parallel Sysplex

The process of shutting down a z/OS image in a Parallel Sysplex is very similar to stopping
an image in either a base sysplex or a non-sysplex environment. This chapter highlights the
differences by showing the operator messages and activities that differ within a Parallel
Sysplex.
As mentioned, the focus here is on the following three areas:
Shutdown overview
Removing z/OS systems from a Parallel Sysplex
Running a standalone dump (SAD) on a Parallel Sysplex
For reference, messages that are often seen during a system stop (both controlled and
uncontrolled) are listed here:
IXC101I SYSPLEX PARTITIONING IN PROGRESS FOR sysname REQUESTED BY jobname REASON:
reason
IXC102A XCF IS WAITING FOR SYSTEM sysname DEACTIVATION. REPLY DOWN
WHEN MVS ON sysname HAS BEEN SYSTEM RESET.
IXC105I SYSPLEX PARTITIONING HAS COMPLETED FOR sysname - PRIMARY REASON text REASON FLAGS: flags
IXC371D CONFIRM REQUEST TO VARY SYSTEM sysname OFFLINE. REPLY SYSNAME=sysname TO
REMOVE sysname OR C TO CANCEL.
IXC402D sysname LAST OPERATIVE AT hh:mm:ss. REPLY DOWN AFTER SYSTEM RESET OR
INTERVAL=SSSSS TO SET A REPROMPT TIME.
Note: In this book, the terms normal, clean, scheduled, planned, and controlled are
synonymous.
4.2 Shutdown overview

There are many different shutdown and failure scenarios. These include: planned with
Sysplex Failure Management (SFM); planned without SFM; manually detected failure;
automatically detected failure; and SFM fails to isolate. Some of these scenarios share the
same actions and procedures.
There are eight steps involved in a planned shutdown of a z/OS system in a Parallel Sysplex.
The scenarios are variations of the basic steps, with differing amounts of operator
intervention. Follow your sites procedures. It is possible to use an automated operations
package to perform some of the activities.
1. Shut down the subsystem workload.
2. Shut down the subsystems and possibly restart them on another system.
3. Shut down z/OS.
4. Remove the system from the Parallel Sysplex by issuing:
VARY XCF,sysname,OFFLINE
5. Respond to IXC371D to confirm the VARY command. After responding to this message,
IXC101I is displayed, indicating that sysplex partitioning is starting.
60
6. If IXC102A is issued, perform a hardware system reset on the system being removed from
the sysplex.
7. Reply DOWN to IXC102A.
8. IXC105I will be displayed when system removal is complete.
Assuming each stage completes successfully, the system is now removed from the Parallel
Sysplex. If the shutdown is on the last system in the Parallel Sysplex, the Parallel Sysplex is
shut down completely. However, one or more of the Coupling Facilities may still be active.
4.2.1 Overview of Sysplex Failure Management

Sysplex Failure Management (SFM) is a z/OS component that can automate responses and
actions to situations and write to operator with reply messages (WTORs) generated during
system shutdowns and failures in the Parallel Sysplex environment. SFM can have an effect
on the shutdown process by removing the need for certain operator actions, thus reducing
delays and minimizing impact.
You need to know whether SFM is currently active on your systems, and what settings are in
place if it is active.
Determining whether SFM is active

The terms started and active are synonymous. To establish the status of SFM, issue the
command D XCF with policy type as shown in Figure 4-1.
D XCF,POL,TYPE=SFM
TYPE: SFM
POLICY NOT STARTED
Figure 4-1 Display with SFM not started
This means that SFM will not take part in the shutdown of any of the systems in the sysplex,
and all eight steps in the shutdown overview will be used.
SFM Couple Data Sets and policy need to be configured by your system programmers. When
SFM needs to start, issue the SETXCF START,POLICY,TYPE=SFM command.
In the example shown in Figure 4-2, SFM is active. However, this does not tell you which SFM
settings are in effect.
D XCF,POL,TYPE=SFM
TYPE: SFM
POLNAME:
SFM01
STARTED:
07/02/2007 20:21:59
LAST UPDATED: 05/28/2004 13:44:52
SYSPLEX FAILURE MANAGEMENT IS ACTIVE
Figure 4-2 Display with SFM active
Chapter 4. Shutting down z/OS systems in a Parallel Sysplex
61
Determining which SFM settings are in effect

To identify which settings are in effect, enter the D XCF,COUPLE command, as shown in
Figure 4-3.
D XCF,COUPLE
SYSTEM #@$3 DATA
1INTERVAL
OPNOTIFY
MAXMSG
85
88
2000
SSUM ACTION
3 ISOLATE
SSUM INTERVAL
0
2CLEANUP
15
RETRY
10
WEIGHT
1
CLASSLEN
956
MEMSTALLTIME
NO
...
Figure 4-3 Display SFM settings
1 The (SUM) INTERVAL is 85 seconds.

2 The CLEANUP interval is 15 seconds.
3 The SUM ACTION is set to ISOLATE.
What SFM does and when it does it are explained in subsequent sections. However, be
aware that your own installation may use different values in the SFM policy.
The SSUM default action setting is to PROMPT the operator to intervene by issuing the
IXC402 message.
Note: SSUM is System Status Update Missing, more commonly referred to as just SUM.
This indicates that the heartbeat to the XCF CDS was not received within the defined
INTERVAL.
For more information about this topic, refer to Chapter 5, Sysplex Failure Management on
page 73.
4.3 Removing a z/OS system from a Parallel Sysplex

This section describes the procedure to shut down and remove a z/OS system in a Parallel
Sysplex. In the examples, the following two scenarios are described, and the appropriate
differences are highlighted:
An active system in the Parallel Sysplex is cleanly shut down. The differences in removing
the last or only system in the sysplex are noted as appropriate.
One (or more) of the systems in the Parallel Sysplex is abnormally stopped.
We start with systems #@$1, #@$2, and #@$3 in the sysplex called #@$#PLEX, as shown
in Figure 4-4 on page 63, and we shut system #@$1.
62
D XCF,S,ALL
IXC335I 19.00.10 DISPLAY
SYSTEM
TYPE SERIAL LPAR
#@$3
2084 6A3A
N/A
#@$2
2084 6A3A
N/A
#@$1
2084 6A3A
N/A
XCF 491
STATUS TIME
06/21/2007 19:00:10
06/21/2007 19:00:06
06/21/2007 19:00:07
SYSTEM STATUS
ACTIVE
ACTIVE
ACTIVE
TM=SIMETR
TM=SIMETR
TM=SIMETR
Figure 4-4 Display currently active systems
4.3.1 Procedure for a planned shutdown

This section demonstrates the planned shutdown of a system in a Parallel Sysplex. It
continues from the closure of all subsystems as listed in steps 1, step 2, and step 3 in 4.2,
Shutdown overview on page 60. The amount of shutdown that the operator performs
depends on the automation environment installed. However, the actions and their results
remain the same.
z/OS closure
When all the subsystems and applications have been shut down, the response to the D A,L
command should be similar to the one shown in Figure 4-5. However, this may not be the
case if there were problems during subsystem closure, or due to your sites configuration.
D A,L
IEE114I 19.11.42 2007.177 ACTIVITY 412
JOBS
M/S
TS USERS
SYSAS
INITS
00000
00000
00000
00032
00015
ACTIVE/MAX VTAM
00000/00030
OAS
00001
Figure 4-5 Display for D A,L
To close z/OS cleanly, the operator should issue the end of day command, Z EOD, as shown
in Figure 4-6, prior to removing the system from the Parallel Sysplex. This writes a logrec
dataset error record, and closes the current SMF dataset to preserve statistical data.
Z EOD
IEE334I HALT EOD SUCCESSFUL
Figure 4-6 Z EOD command
Sysplex partitioning
Use the V XCF,sysname,OFFLINE command to remove the closing system from the Parallel
Sysplex. This is shown in Figure 4-7 on page 64. It can be issued on any system in the
sysplex, including on the system that is being removed.
Note: The VARY command (and sysname) should still be used when removing the last or
only system from the sysplex, because there is still some cleanup to be done. However,
you do not receive message IXC102A, because there is no active system to issue it.
63
V XCF,#@$1,OFFLINE
*018 IXC371D CONFIRM REQUEST TO VARY SYSTEM #@$1
* OFFLINE. REPLY SYSNAME=#@$1 TO REMOVE #@$1 OR C TO CANCEL.
Figure 4-7 Initiate sysplex partitioning
The system on which the command is entered issues message IXC371D. This message
requests confirmation of the removal, also shown in Figure 4-7. To confirm removal, this
WTOR must be replied to, as shown in Figure 4-8.
R 18,SYSNAME=#@$1
IEE600I REPLY TO 018 IS;SYSNAME=#@$1
Figure 4-8 Confirm removal system name
We recommend that this should not be performed prior to a SAD. If an XCF component was
causing the problem that necessitated the SAD, then diagnostic data will be lost.
Note: If the confirmation is entered with a sysname that is different from the one requested
(#@$1), then the following message IXC208I is issued, and IXC371 is repeated:
R 18,SYSNAME=#@$2
IXC208I THE RESPONSE TO MESSAGE IXC371D IS INCORRECT: #@$2 IS NOT ONE OF THE
SPECIFIED SYSTEMS
From this point on (having received a valid response to IXC371), the CLEANUP interval
(seen in Figure 4-3 on page 62) starts, sysplex partitioning (also known as fencing) begins,
and the message seen in Figure 4-9 is issued to a randomly chosen system, which monitors
the shutdown.
IXC101I SYSPLEX PARTITIONING IN PROGRESS FOR #@$1 REQUESTED BY
*MASTER*. REASON: OPERATOR VARY REQUEST
Figure 4-9 IXC101I Sysplex partitioning initiated by operator
During this step, XCF group members are given a chance to be removed,
The next step varies, depending on whether or not there is an active SFM policy with system
isolation in effect. The without scenario is described first.
Scenario without an active SFM policy

Without an active SFM policy, message IXC102A is issued. This appears on the system
monitoring the removal. It requests a DOWN reply following a SYSTEM RESET (or equivalent)
on the system being closed; see Figure 4-10.
However, this message is not issued if this is the last or only system in the sysplex that is
being shut, because there is no active system to issue it.
*022 IXC102A XCF IS WAITING FOR SYSTEM #@$1 DEACTIVATION. REPLY DOWN
Figure 4-10 IXC102 System removal waiting for reset
64
When this stage of the cleanup is complete (or if the CLEANUP interval expires), the system
being removed is loaded with a non-restartable WAIT STATE 0A2.
Wait for this state before performing the system reset. Do not reply DOWN yet.
Do not reply to IXC102A until SYSTEM RESET is done: Before replying DOWN to
IXC102A, you must perform a hardware SYSTEM RESET (or equivalent) on the system
being removed. This is necessary to ensure that this system can no longer perform any I/O
operations, and that it releases any outstanding I/O reserves. The SYSTEM RESET
therefore ensures data integrity on I/O devices.
SYSTEM RESET refers to an action on the processor that bars z/OS from doing I/O. The
following are all valid actions for a SYSTEM RESET: Stand Alone Dump (SAD), System
Reset Normal, System Reset Clear, Load Normal, Load Clear, Deactivating the Logical
Partition, Resetting the Logical Partition, Power-on Reset (POR), Processor IML, or
Powering off the CPC.
When the SYSTEM RESET (or its equivalent) is complete, the operator should reply DOWN to
the IXC102A WTOR; see Figure 4-11.
R 22,DOWN
IEE600I REPLY TO 022 IS;DOWN
Figure 4-11 Reply DOWN after system reset
After DOWN has been entered, XCF performs a cleanup of the remaining resources relating
to the system being removed from the sysplex, as seen in Figure 4-12.
IEA257I CONSOLE PARTITION CLEANUP IN PROGRESS FOR SYSTEM #@$1.

CNZ4200I CONSOLE #@$1M01 HAS FAILED. REASON=SYSFAIL
IEA258I CONSOLE PARTITION CLEANUP COMPLETE FOR SYSTEM #@$1
Figure 4-12 Console cleanup
Finally, removal of the system from the sysplex completes, and the following IXC105I
message is issued, as shown in Figure 4-13. The reason flags may vary at your site.
IXC105I SYSPLEX PARTITIONING HAS COMPLETED FOR #@$1 236
- PRIMARY REASON: OPERATOR VARY REQUEST
Figure 4-13 Sysplex partitioning completed
What happens if you do not reply DOWN

If you do not reply to IXC102A (or IXC402D), the system cleanup will not be performed, and
XCF thinks the closing system is still in the sysplex. If you re-IPL the same system, message
IXC203I is issued, followed by IXC207A, as shown in Figure 4-14 on page 66.
65

IXC218I SYSTEM STATUS FOR SYSPLEX #@$#PLEX AT 06/27/2007 00:54:56:
#@$2
00:54:54 ACTIVE
#@$1
00:52:11 BEING REMOVED
REPLY COUPLE=XX.
Figure 4-14 IXC203 system currently active
Check for the IXC102A message and reply DOWN to it. When IXC105I is issued, then reply with
your correct COUPLExx member and the IPL will continue. You can also choose to start the
IPL again.
Scenario with an active SFM policy

In this example, SFM is active and system ISOLATE is coded, as shown in Figure 4-3 on
page 62.
Up to this point, the operator had issued V XCFTh and replied with sysname. After IXC101I is
issued, SFM waits for the CLEANUP interval to expire, then initiates removal of the closing
system. This produces messages including IXC467I, IXC307I, and IXC302I, as shown in
When system isolation successfully completes, message IXC105 is issued (see Figure 4-13
on page 65), and the system is placed into a disabled WAIT STATE. Refer to Wait state
X'0A2' on page 66 for more information about this topic.
It is not necessary after successful SFM partitioning to perform a system reset, because the
isolated system can no longer perform any I/O.
If SFM isolation is unsuccessful

A failed attempt at fencing could occur for any number of reasons. It is detected by SFM and
message IXC102A is issued. As shown in Figure 4-10 on page 64, this requires:
The closing system to be SYSTEM RESET
The reply of DOWN
As usual, IXC105 is issued when isolation is complete.
Last or only system in the sysplex

As noted previously, the V XCF and reply sysname should still be performed on the last or
only system in the sysplex. However, message IXC102A will not be issued. You cannot reply
DOWN. Instead, you:
Observe the 0A2 Wait State
Perform system reset
Completion of removal is flagged by the reset; message IXC105I is not issued because there
is no active system to issue it.
Wait state X'0A2'

If cleanup processing completes before a SYSTEM RESET is performed, or when an active
SFM policy has isolated the system successfully, the system being removed will be placed in
a X'0A2' non-restartable wait state. The console clears, and message IXC220W is issued.
66
*IXC220W XCF IS UNABLE TO CONTINUE: WAIT STATE CODE: 0A2 REASON CODE: 004,
AN OPERATOR REQUESTED PARTITIONING WITH THE VARY XCF COMMAND
Figure 4-15 0A2 non-restartable wait state
Important: Wait for the WAIT STATE before performing SYSTEM RESET.
Sysplex cleanup
With any closure of a system in a Parallel Sysplex, whether controlled or not, the remaining
systems clean up the XCF connections. This activity occurs when the CLEANUP interval (as
shown in Figure 4-3 on page 62) expires.
The default XCF CLEANUP time is sixty seconds. However, thirty seconds is recommended.
Tip: Set XCF CLEANUP interval to 30 seconds.
As shown in Figure 4-16, GRS and system partitioning take place, and these are indicated by
many IXC467I, IXC307I, and IXC302I messages, which may not be seen at the console.
IXC467I STOPPING PATH STRUCTURE IXC_DEFAULT_2 217
RSN: SYSPLEX PARTITIONING OF LOCAL SYSTEM
IXC467I STOPPING PATHOUT STRUCTURE IXC_DEFAULT_2 LIST 8 218
USED TO COMMUNICATE WITH SYSTEM #@$2
RSN: SYSPLEX PARTITIONING OF LOCAL SYSTEM
...
LIST 10 TO COMMUNICATE WITH SYSTEM #@$3 COMPLETED
SUCCESSFULLY: SYSPLEX PARTITIONING OF LOCAL SYSTEM
SUCCESSFULLY: SYSPLEX PARTITIONING OF LOCAL SYSTEM
...
IXC302I STOP PATHIN REQUEST FOR STRUCTURE IXC_DEFAULT_1 116
LIST 0 TO COMMUNICATE WITH SYSTEM #@$1 REJECTED:
UNKNOWN PATH
DIAG037=18 DIAG074=08710000 RC,RSN=00000008 081A0004
LIST 0 TO COMMUNICATE WITH SYSTEM #@$1 REJECTED:
UNKNOWN PATH
DIAG037=18 DIAG074=08710000 RC,RSN=00000008 081A0004
Figure 4-16 PATHIN and PATHOUT cleanup
Checking that the system has been removed

If at least one system is left in the sysplex, reissue the D XCF,S,ALL command to verify that
the target system has been removed.
67
4.3.2 Procedure for an abnormal stop

This section discusses system failure by describing what events occur in a Parallel Sysplex,
as seen by the operator, when:
SFM is active, with system isolation in effect
SFM is inactive
Figure 4-3 on page 62 displays an overview of checking your SFM settings.
Sometimes a failure is detected by the operator or system programmer when the system is
slow, hung, or not responding, but is still able to provide its heartbeat to the XCF couple data
set, and can still communicate with the other systems.
In such cases, a decision can be made to IPL now, or IPL later, and the procedure for
planned shutdown can still be followed, because it is performed from the other systems
which are not affected.
However, during normal operations, there could be occasions when an z/OS system cannot
be closed cleanly due to problems that are beyond operator control. This may be due to an
application or subsystem error, or may be caused by a major hardware or software failure, like
loss of power, or z/OS hanging.
System failure detection

Subject to your installations configuration, an early indication of a system failure may come
from subsystem monitoring. In this test sysplex example, system #@$1 was deliberately
brought down by performing a SYSTEM RESET on the processor. Both IMS and VTAM
messages (respectively) appeared before the XCF message; see Figure 4-17. (This scenario
might not occur at your installation.)
DFS4165W FDR FOR (I#$1) XCF DETECTED TIMEOUT ON ACTIVE IMS SYSTEM,
REASON = SYSTEM
, DIAGINFO = 0C030384
F#$1
DFS4164W FDR FOR (I#$1) TIMEOUT DETECTED DURING LOG AND XCF SURVEILLANCE
OF #$1
IST1494I PATH SWITCH FAILED FOR RTP CNR0000C TO USIBMSC.#@$2M
IST1495I NO ALTERNATE ROUTE AVAILABLE
Figure 4-17 Subsystem detection of potential failure
More commonly, the first indication of system failure is when a Status Update Missing (SUM)
condition is registered by one of the other systems in the sysplex. This occurs when a system
has not issued a status update (heartbeat) to the XCF couple dataset within the INTERVAL
since the last update. This value is defined in the COUPLExx member of PARMLIB, and
shown in Figure 4-3 on page 62. When a SUM condition occurs, the system detecting the
SUM notifies all other systems and issues the IXC101I message with the text as shown in
Figure 4-18. The SUM reason text differs from the OPERATOR VARY REQUEST in the
previous section. There are a dozen different reasons that can cause this alert to appear, but
only two are considered here.
IXC101I SYSPLEX PARTITIONING IN PROGRESS FOR #@$1 REQUESTED BY XCFAS.
REASON: SFM STARTED DUE TO STATUS UPDATE MISSING
Figure 4-18 IXC101 sysplex partitioning initiated by XCF
68
When a system failure is detected, one of the following actions occurs:

SFM initiates partitioning of the failing system, and it works.
SFM initiates partitioning of the failing system, and it does not work.
If SFM is inactive, message IXC402D (rather than IXC102) is issued.
SFM initiates partitioning of the failing system, and it works

An active SFM system policy to ISOLATE automatically initiates system partitioning when it
detects the SUM condition. This is notified by an IXC101I message, issued after the
INTERVAL has elapsed. Partitioning triggers the cleanup of resources (such as consoles,
GRS, XCF paths) for the failing system.
If partitioning is successful, the usual IXC105I completion message is issued. There is no
need to RESET because the isolated system can no longer perform any I/O.
SFM initiates partitioning of the failing system, and it doesnt work

If system isolation fails, then SFM issues the IXC102A WTOR, instead of the IXC402D
WTOR, after the XCF CLEANUP time has elapsed; see Figure 4-19.
031 IXC102A XCF IS WAITING FOR SYSTEM #@$1 DEACTIVATION. REPLY DOWN
Figure 4-19 If SFM fails to isolate
Do not reply to IXC102A until SYSTEM RESET is done: Before replying DOWN to
IXC102A, you must perform a hardware SYSTEM RESET (or equivalent) on the system
being removed. This is necessary to ensure that this system can no longer perform any I/O
operations, and that it releases any outstanding I/O reserves. The SYSTEM RESET,
therefore, ensures data integrity on I/O devices.
SYSTEM RESET refers to an action on the processor that bars z/OS from performing I/O.
The following are all valid actions for a SYSTEM RESET: Stand-alone Dump (SAD),
System Reset Normal, System Reset Clear, Load Normal, Load Clear, Deactivating the
Logical Partition, Resetting the Logical Partition, Power-on Reset (POR), Processor IML,
or Powering off the CEC.
SFM is inactive
When a sysplex without an active SFM policy becomes aware of the system failure, message
IXC402D is issued which alerts the operator that a system (#@$1 in this exercise) is not
operative, shown in Figure 4-20.
006 R #@$3
*006 IXC402D #@$1 LAST OPERATIVE AT 20:30:15.

REPLY DOWN AFTER SYSTEM RESET, OR INTERVAL=SSSSS
TO SET A REPROMPT TIME.
Figure 4-20 SUM condition without SFM
With the IXC402D WTOR, the operator is requested to reply DOWN when the system has
been reset. The INTERVAL option allows the operator to specify a period of time for system
operation recovery.
If the system has not recovered within this period, message IXC402D is issued again. The
INTERVAL reply can be in the range 0 to 86400 seconds (24 hours).
69
After the SYSTEM RESET has been performed, you can reply DOWN to IXC402D, as shown in
Figure 4-21. It is only this reply that starts the partitioning.
R 06,DOWN
Figure 4-21 DOWN reply after IXC402
Important: As for an IXC102A message, with IXC402D do not reply DOWN until a
SYSTEM RESET has been performed.
SFM without isolation

There is another possible but unusual scenario, which is to have an active SFM policy that
specifies PROMPT (which is the default) instead of ISOLATE. Although it may seem to defeat
the purpose of SFM to have it prompt, it is nonetheless possible. When SFM is prompted
during an SUM condition (it is not invoked during normal shutdown), it issues the IXC402D
message. Actions are thus the same as for SFM is inactive on page 69.
System cleanup
No matter how it is arrived at, after the DOWN reply, XCF performs a cleanup of resources
relating to the system being removed. This activity starts and ends with the IEA257I and
IEA258I messages interspersed with IEE501I messages, as shown in Figure 4-22.
IEA257I CONSOLE PARTITION CLEANUP IN PROGRESS FOR SYSTEM #@$1.
CNZ4200I CONSOLE #@$1M01 HAS FAILED. REASON=SYSFAIL
IEA258I CONSOLE PARTITION CLEANUP COMPLETE FOR SYSTEM #@$1
IEE501I CONSOLE #@$1M01 FAILED, REASON=SFAIL . ALL ALTERNATES
UNAVAILABLE, CONSOLE IS NOT SWITCHED
Figure 4-22 Console cleanup
When system removal completes, IXC105I message is issued. The system is now out of the
sysplex, as shown in Figure 4-23.
Note: When you receive message IXC105I, the RSN text may be different.
IXC105I SYSPLEX PARTITIONING HAS COMPLETED FOR #@$1 790

- PRIMARY REASON: SYSTEM REMOVED BY SYSPLEX FAILURE MANAGEMENT BECAUSE
ITS STATUS UPDATE WAS MISSING
Figure 4-23 Sysplex partitioning completed
Sysplex cleanup
With any closure of a system in a Parallel Sysplex (except the last or only), whether controlled
or not, the remaining systems clean up the XCF connections. This activity occurs at the same
time as GRS and system partitioning take place. This is indicated by many IXC302 1,
IXC307I 2, and IXC467I 3 messages, which may not be seen at the console; see Figure 4-24
on page 71.
70
IXC467I STOPPING PATHIN STRUCTURE IXC_DEFAULT_2 LIST 12 789

RSN: SYSPLEX PARTITIONING OF REMOTE SYSTEM
...
SUCCESSFULLY: SYSPLEX PARTITIONING OF REMOTE SYSTEM
...
Figure 4-24 Sysplex cleanup
4.4 Running a stand-alone dump on a Parallel Sysplex

A stand-alone dump (SAD) is diagnostic tool. It takes a snapshot of the running environment
of an system for later analysis. It is normally performed when a system has failed, most often
at the request of the system programmer.
During system failure, the sooner the SAD is performed, the better the diagnostic information
collected and the sooner the system is back up. It is good practice for Operations to know
who can request a SAD. From an operational perspective, a SAD may need to be performed
at a moment's notice. Your site should have in place up-to-date procedures in a prominent
location. All operators should familiarize themselves with the documentation. It is good
practice to schedule test SADs.
There may be standards on IPL profiles, or dump volumes, or dump names. The dump
screen may need to be captured. Consult your installation's documentation for guidance
about these topics.
This section details two different scenarios when SADs are required:
SAD required during planned removal of a system from a sysplex.
SAD required during unplanned removal of a system from a sysplex with SFM active.
An example of running a stand-alone dump in a Parallel Sysplex is provided in Appendix C,
Stand-alone dump on a Parallel Sysplex example on page 503.
4.4.1 SAD required during planned removal of a system

The recommended procedure for taking a SAD of an z/OS image running in a sysplex, and
then removing that system from the sysplex, is as follows:
1. Perform a hardware STOP function to place the system CPUs into a stopped state.
2. IPL the stand-alone dump program.
71
3. Issue the VARY XCF,sysname,OFFLINE command from another system in the sysplex (if
message IXC402D or IXC102A is not already present).
4. Reply DOWN to message IXC402D or IXC102A, without performing a SYSTEM RESET,
because this has already taken place through the IPL of the SADMP program, and you
would cause the SADMP to fail.
Note the following points:
If this is the last or only system in the sysplex, then step 1 and step 2 apply.
If the system has already been removed from the sysplex, then only step 2 applies.
You do not need to wait for the SAD to complete before continuing with step 3.
Performing steps 3 and 4 immediately after IPLing the SAD will speed up sysplex
recovery, allowing resources held by the IPLing system to be released quickly.
If there is a delay between steps 2 and 3, then messages IXC402D or IXC102A may be
issued by another system detecting the loss of connectivity with the IPLing system.
After the SAD program is IPLed, IXC402D or IXC102A will be issued, even if an active
SFM policy is in effect. This happens because z/OS is unable to automatically partition the
failing system using SFM.
4.4.2 SAD required during unplanned removal of a system with SFM active
If SFM is active and system isolation is in effect, then it will detect the system failure and start
sysplex partitioning. In this case, follow this procedure:
1. Perform a hardware STOP function to place the failing systems CPUs into a stopped
state (this is not strictly required, but is good practice).
2. IPL the standalone dump program on the failing system,
3. If message IXC102A is present, reply DOWN without performing a SYSTEM RESET.
72
Chapter 5.
Sysplex Failure Management

This chapter provides information about the Sysplex Failure Management (SFM) function. It
introduces commands that you can use to determine the status of SFM in your sysplex.
This chapter explains:
Why to use SFM
How SFM reacts to various failure situations
How to control SFM
73
5.1 Introduction to Sysplex Failure Management

Sysplex Failure Management (SFM) is an availability function that is integrated into the z/OS
base. It allows you to define a sysplex-wide policy that specifies the actions z/OS is to take
when certain failures occur in the sysplex, such as when a system loses access to a Coupling
Facility.
When SFM is active, it is invoked automatically to minimize the impact that a failing system
might have on the workloads running in a sysplex. It does this by automating the recovery
actions, which shortens the recovery time and reduces the time the sysplex is impacted by
the failure. If these recovery actions are not performed quickly, there may be an extended
impact to the other systems and the workloads running on them. For instance, if a system
fails while holding an exclusive GRS enqueue, work on the other systems may wait until the
enqueue is released before they can continue their processing.
Some sysplex delays may also occur during recovery processing. If a system fails while
holding some locks, XES will transfer the management of the locks to one of the other
systems during the cleanup processing. During this transition, Cross-System Extended
Service (XES) will quiesce all activity to the associated lock structures to preserve data
integrity. This delay may impact the subsystems that use these lock structures, such as
IRLM/DB2, VSAM RLS and IRLM/IMS.
SFM is invoked for both planned and unplanned conditions, when:
A system is varied out of the sysplex using the VARY XCF command
A system enters a Status Update Missing condition
An XCF signalling connectivity failure occurs
A system loses connectivity to a Coupling Facility
When a failure is detected, a recovery action is initiated, such as:
Isolating the failed image
Deactivating logical partitions
Reallocating real storage
These recovery actions can be initiated automatically and completed without operator
intervention, or a message can be issued, prompting the operator to perform the recovery
actions manually. The recovery actions are defined in the SFM policy.
For a sysplex to take advantage of SFM, all systems must have connectivity to the SFM
couple datasets and the SFM policy must be started.
For additional information about SFM, see MVS Setting up a Sysplex, SA22-7625.
5.2 Status Update Missing condition

For each z/OS system, XCF updates the sysplex couple dataset (CDS) with its status every
few seconds; the status consists of a time stamp and it is sometimes referred to as a
heartbeat. For example, if you have 10 systems in a sysplex, all 10 systems update their
respective heartbeats in the sysplex CDS every few seconds.
In addition to writing the heartbeat time stamps to the sysplex CDS, XCF on each z/OS
system monitors the heartbeat time stamps of the other z/OS systems in the sysplex CDS. If
74
any z/OS system's heartbeat time stamp is older than the current time minus that system's
INTERVAL value from the COUPLExx parmlib member, that system is considered to have
failed in some way. When this occurs, the failed system is considered to be in a Status
Update Missing (SUM) condition. All systems are notified when a SUM condition occurs. The
recovery actions which are taken when a SUM condition occurs depend on the recovery
parameters that you specify in your SFM policy. They could be:
Prompt the operator to perform the recovery actions manually.
Remove the system from the sysplex without operator intervention by:
Using the Coupling Facility fencing services to isolate the system.
System resetting the failing systems LPAR.
Deactivating the failing systems LPAR.
5.3 XCF signalling failure

XCF monitors XCF signalling connectivity to ensure that systems in the sysplex can
communicate with each other. If XCF detects that two or more systems can no longer
communicate with each other because some signalling paths have failed, then SFM will
determine which systems should be partitioned out of the sysplex and proceed to remove
them from the sysplex in one of two ways:
Automatically, without operator intervention
By prompting the operator to perform the recovery actions manually
SFMs recovery action is controlled by two parameters in your SFM policy:
WEIGHT
The WEIGHT parameter from the SFM policy allows you to indicate the relative
importance of each system in the sysplex. SFM uses this value to determine which
systems should be partitioned out of the sysplex when signalling connectivity failures
occur.
CONNFAIL
The CONNFAIL parameter controls SFMs recovery actions if signalling connectivity fails
between one or more systems in the sysplex:
If you specify YES, then SFM performs sysplex partitioning actions using the WEIGHT
values assigned to each system in the sysplex. SFM determines the best set of
systems to remain in the sysplex and which systems to remove from the sysplex, and
then attempts to implement that decision by system isolation.
If you specify NO, then SFM prompts the operator to decide which system or systems
to partition from the sysplex.
5.4 Loss of connectivity to a Coupling Facility

A Coupling Facility (CF) link failure or certain types of failures within a CF can cause one or
more systems to lose connectivity to a CF. This means these systems will also lose
connectivity to the structures residing in that CF.
In this situation, z/OS initiates a structure rebuild, which is a facility that allows structures to
be rebuilt into another CF. If the structure supports rebuild, it rebuilds into another CF and the
structure exploiters connect to it in the new location.
Chapter 5. Sysplex Failure Management
75
If the structure supports rebuild, you can influence when it should be rebuilt by using the
REBUILDPERCENT parameter in the structures definition in the Coupling Facility Resource
Management (CFRM) policy:
The structure is rebuilt if the weight of the system that lost connectivity is equal to or
greater than the REBUILDPERCENT value you specified.
The structure is not rebuilt if the weight of the system that lost connectivity is less than the
REBUILDPERCENT value you specified. In this case, the affected system will go into
error handling to recover from the connectivity failure.
If the structure supports user-managed rebuild and you used the default value of 1% for
REBUILDPERCENT, the structure rebuilds when a loss of connectivity occurs.
During the rebuild, to ensure that the rebuilt structure has better connectivity to the systems in
the sysplex than the old structure, the CF selection process will factor in the SFM system
weights and the connectivity that each system has to the CF. However, if there is no SFM
policy active, all the systems are treated as having equal weights when determining the
suitability of a CF for the new structure allocation.
5.5 PR/SM reconfiguration

After a system running in an LPAR is removed from the sysplex, SFM allows the remaining
systems in the sysplex to take the processor storage that had been in use by the failed
system and make it available for their own use.
5.6 Sympathy sickness

The Sysplex Failure Management (SFM) function in z/OS is enhanced to support a new
policy specification for how long a system should be allowed to remain in the sysplex when it
appears unresponsive because it is not updating its system status on the Sysplex Couple
Data Set, yet it is still sending XCF signals to other systems in the sysplex. A system that is in
this state is definitely not completely inoperable (because it is sending XCF signals), and yet
it may not be fully functional either, so it may be causing sysplex sympathy sickness problems
for other active systems in the sysplex.
The new SFM policy externally provides a way for installations to limit their exposure to
problems caused by such systems, by automatically removing them from the sysplex after a
specified period of time. The Sysplex Failure Management (SFM) function in z/OS is
enhanced to support a new policy specification to indicate that, after a specified period of
time, the system may automatically terminate XCF members that have been identified as
stalled and who also appear to be causing sympathy sickness problems.
If allowed to persist, these stalled members can lead to sysplex-wide hangs or other
problems, not only within their own XCF group, but also for any other system or application
functions that depend on the impacted function. Automatically terminating these members is
intended to provide improved application availability within the sysplex.
76
5.7 SFM configuration

For SFM to be active in a sysplex, all systems must have connectivity to the SFM CDSs and
an SFM policy must be started. When SFM is invoked, the actions it performs are determined
by parameters you have defined in your COUPLExx parmlib member and your SFM policy:
The COUPLExx parmlib member provides basic failure information, such as when to
consider a system has failed or when to notify the operator of the failure.
SFM policy defines how XCF handles systems failures, signalling connectivity failures, or
PR/SM reconfigurations.
5.7.1 COUPLExx parameters used by SFM

SFM uses three parameters from the COUPLExx parmlib member: INTERVAL 1, OPNOTIFY
2 and CLEANUP 3. You can view these SFM parameters in the output from the D XCF,COUPLE
command, as shown in Figure 5-1.
D XCF,COUPLE
SYSTEM #@$2 DATA
INTERVAL
OPNOTIFY
MAXMSG
1 85
2 88
2000
SSUM ACTION
4 ISOLATE
. . .
CLEANUP
3 15
SSUM INTERVAL
5 0
RETRY
10
WEIGHT
6 19
CLASSLEN
956
MEMSTALLTIME
NO
Figure 5-1 SFM parameters from D XCF,COUPLE output
The INTERVAL 1 is otherwise known as the failure detection interval. This specifies
when the failing system is considered to have entered a SUM condition.
The OPNOTIFY 2 specifies when SFM notifies the operator that a system has not updated
its status. The timer for both INTERVAL and OPNOTIFY start at the same time and the
value for OPNOTIFY must be greater than or equal to the value specified for INTERVAL.
The CLEANUP 3 interval specifies how long XCF group members can perform clean-up
for the z/OS system being removed from the sysplex. The intention of the cleanup interval
is to give XCF group members on the system being removed a chance to exit gracefully
from the system. The XCF CLEANUP interval only applies to planned system shutdowns,
when the VARY XCF command is used to remove a z/OS system from a sysplex.
5.7.2 SFM policy

You can view some of the SFM policy parameters in the output from the D XCF,COUPLE
The SSUM ACTION 4 indicates the recovery action taken when a SUM condition occurs. The
options are:
PROMPT - Prompt the operator to perform the recovery actions manually. This is the
default value.
Remove the system from the sysplex without operator intervention:
ISOLATE - Using the Coupling Facility fencing services to isolate the system. We
recommend using this value.
77
RESET - System resetting the failing systems LPAR.

DEACTIVATE - Deactivating the failing systems LPAR.
In this example, the option is ISOLATE, which causes SFM to initiate automatic removal of
the failing system from the sysplex using CF fencing services.
The SSUM INTERVAL 5 indicates how soon after the failure detection interval expires that
the recovery action occurs. In this example, the value is 0. Thus, the failing system will be
removed from the sysplex as soon as it enters a SUM condition; 85 seconds after the last
status update.
The WEIGHT 6 is used if signalling connectivity errors occur:
The WEIGHT parameter allows you to indicate the relative importance of each system in
the sysplex. SFM uses this value to determine which systems should be partitioned out of
the sysplex if a signalling connectivity failure occurs. This can be a value between 1 and
9999.
SFM determines whether to initiate a rebuild for a structure in a Coupling Facility to which
a system has lost connectivity. SMF uses the assigned weights in conjunction with the
REBUILDPERCENT value specified in the CFRM policy.
The policy information also specifies whether SFM is used to automatically recover XCF
signaling connectivity failures and what reconfiguration actions are to be taken when the
PR/SM Automatic Reconfiguration Facility is being used. These options cannot be displayed
via an operator command; however, the contents of the active SFM policy can be listed using
a batch job.
The sample JCL for this batch job is shown in Figure 5-2. Consult your system programmer
before running this job to ensure you have the correct security access.
//DEFSFMP1 JOB (0,0),'SFM POLICY',CLASS=A,MSGCLASS=X,
//
NOTIFY=&SYSUID
//*JOBPARM SYSAFF=#@$3
//*****************************************************
//STEP10
EXEC PGM=IXCMIAPU
//STEPLIB DD
DSN=SYS1.MIGLIB,DISP=SHR
//SYSPRINT DD
SYSOUT=*
//SYSIN
DD
*
DATA TYPE(SFM) REPORT(YES)
/*
Figure 5-2 SFM policy report sample JCL
5.7.3 Access to the SFM CDSs

For SFM to be active in the sysplex, all systems must have connectivity to the SFM CDSs and
the SFM policy must be started. If any system loses connectivity to both the primary and
alternate SFM CDS, SFM becomes inactive in the sysplex. SFM automatically becomes
active again when all systems regain access to either the primary or alternate SFM CDS.
78
5.8 Controlling SFM

The following commands can be used to control SFM. For additional information about these
commands, see z/OS MVS System Commands, SA22-7627.
5.8.1 Displaying the SFM couple datasets

To display the SFM CDS configuration, use the following command:
D XCF,COUPLE,TYPE=SFM
An example of the response to this command is shown in Figure 5-3. You can see:
There are two SFM CDSs; a primary 1 and an alternate 2 CDS.
Both CDSs are connected to all systems 3.
SFM COUPLE DATA SETS
PRIMARY
DSN: SYS1.XCF.SFM01 1
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 16:08:53
4
ADDITIONAL INFORMATION:
FORMAT DATA
POLICY(9) SYSTEM(16) RECONFIG(4)
ALTERNATE DSN: SYS1.XCF.SFM02 2
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
11/20/2002 16:08:53
4
FORMAT DATA
SFM IN USE BY ALL SYSTEMS 3
Figure 5-3 SFM couple dataset configuration
5.8.2 Determining whether SFM is active

To see whether SFM is active in the sysplex, use the following command:
D XCF,POLICY,TYPE=SFM
An example of the response to this command when SFM is active is shown in Figure 5-4 on
page 80. You can see:
SFM is active, as shown in the last line of the output 2.
The name of the current SFM policy, SFM01 1, and when it was started.
79
D XCF,POL,TYPE=SFM
TYPE: SFM
POLNAME:
SFM01 1
STARTED:
07/02/2007 20:21:59
LAST UPDATED: 05/28/2004 13:44:52
Figure 5-4 SFM policy display when SFM is active
An example of the response to this command when SFM is not active is shown in Figure 5-5.
The last line of the output shows that SFM is not active 3.
TYPE: SFM
POLICY NOT STARTED 3
Figure 5-5 SFM policy display when SMF is inactive
You can also use the following command to determine if SFM is active in the sysplex:
D XCF,COUPLE
An example of the response to this command when SFM is active and when SFM is not
active is shown in Figure 5-6. When SFM is active, the SSUM ACTION, SSUM INTERVAL,
WEIGHT, and MEMSTALLTIME fields 1 are populated with values from the SFM policy.
D XCF,COUPLE
SYSTEM #@$2 DATA
INTERVAL
OPNOTIFY
MAXMSG
85
88
2000
SSUM ACTION
ISOLATE
. . .
SSUM INTERVAL
0
CLEANUP
15
RETRY
10
CLASSLEN
956
WEIGHT MEMSTALLTIME
19
NO
When SFM is not active, these fields contain N/A 2, as shown in Figure 5-7.
D XCF,COUPLE
SYSTEM #@$2 DATA
INTERVAL
OPNOTIFY
MAXMSG
85
88
2000
SSUM ACTION
N/A
. . .
SSUM INTERVAL
N/A
CLEANUP
15
Figure 5-7 SFM policy display when SFM is inactive
80
RETRY
10
CLASSLEN
956
WEIGHT MEMSTALLTIME
N/A
N/A
5.8.3 Starting and stopping the SFM policy

To start an SFM policy, use the following command. In this example we are starting an SFM
policy called SFM01:
SETXCF START,POLICY,TYPE=SFM,POLNAME=SFM01
An example of the system response to this command is shown in Figure 5-8. The system
responses show which SFM values were taken from the SFM policy 1 and 2, and which
values are system defaults 3.
THE ACTION WAS SPECIFIED FOR THIS SYSTEM. 1
19 SPECIFIED BY SPECIFIC POLICY ENTRY 2
IXC614I SFM POLICY SFM01 INDICATES MEMSTALLTIME(NO) FOR SYSTEM #@$2 AS
SPECIFIED BY SYSTEM DEFAULT 3
IXC601I SFM POLICY SFM01 HAS BEEN STARTED BY SYSTEM #@$2
Figure 5-8 Console messages when starting SFM policy
If your system programmer asks you to stop the current SFM policy, use the following
command:
SETXCF STOP,POLICY,TYPE=SFM
An example of the system response to this command is shown in Figure 5-9. This command
stops the SFM policy on all systems in the sysplex. After the SFM policy is stopped, its status
in the sysplex changes to POLICY NOT STARTED, as shown in Figure 5-5 on page 80.
IXC607I SFM POLICY HAS BEEN STOPPED BY SYSTEM #@$2
Figure 5-9 Console messages when stopping SFM policy
5.8.4 Replacing the primary SFM CDS

The process to replace a primary CDS is described in Chapter 8, Couple Data Set
management on page 165.
5.8.5 Shutting down systems when SFM is active

The process to shut down a system when SFM is active is described in Chapter 4, Shutting
down z/OS systems in a Parallel Sysplex on page 59.
SFM isolation failure

When SFM does not successfully isolate the system being removed, you have to perform the
recovery actions manually.
There are several reasons why SFM may not isolate the system being removed, such as:
There is no CF in the configuration.
A system-to-CF link is inoperative or not configured.
81
The system being removed was SYSTEM RESET or IPLed.

The logical partition where the system being removed resides was deactivated.
If SFM is unable to isolate the system being removed, it will issue the message IXC102A
when the XCF CLEANUP interval expires, as shown in Figure 5-10.
IXC102A XCF IS WAITING FOR SYSTEM #@$1 DEACTIVATION. REPLY DOWN

Figure 5-10 IXC102A message
When you receive message IXC102A, perform the following:

A SYSTEM RESET on the z/OS system being removed from the sysplex, if you have not
already done so.
Reply DOWN to IXC102A.
Message IXC105I will be issued when system removal is complete.
82
Chapter 6.
Automatic Restart Manager

This chapter discusses Automatic Restart Manager (ARM) scenarios. It also examines
problems that may be encountered with ARM policies.
How to define an ARM policy
What happens during an ARM restart
Finally, the chapter provides an example that illustrates how to define a simple task to ARM;
see 6.5, Defining SDSF as a new ARM element on page 91.
83
6.1 Introduction to Automatic Restart Manager

As explained in Chapter 5, Sysplex Failure Management on page 73, Sysplex Failure
Management (SFM) handles system failures in a sysplex. In contrast, Automatic Restart
Manager (ARM) is a z/OS recovery function to improve the availability of specific batch jobs
or started tasks. The goals of SFM and ARM are complementary. SFM keeps the sysplex
running, and ARM keeps specific work in the sysplex running. If a job or task fails, ARM
restarts it.
The purpose of ARM is to provide fast, efficient restarts for critical applications when they fail.
ARM improves the time required to restart an application by automatically restarting the batch
job or started task (STC) when it unexpectedly terminates. These unexpected outages may
be the result of an abend, system failure, or the removal of a system from the sysplex.
For a batch job or started task (STC) to use ARM, two criteria must be met:
It needs to be defined to the active ARM policy.
It needs to register to ARM when starting.
ARM will attempt to restart any job or STC that meets these criteria, if it abnormally fails.
Any ARM configuration should be closely coordinated with any automation products to avoid
duplicate startup attempts and to monitor any ARM restart failures.
The utility IXCMIAPU is used to define ARM policies. The operator command
SETXCF START is used to activate the policies. Figure 6-1 illustrates an ARM configuration.
For further information about IXCMIAPU, ARM, and policy setup, refer to MVS Setting Up a
Sysplex, SA22-7625. For information about how some subsystems are controlled in an ARM
environment, refer to the following sections:
For CICS, refer to CICS and ARM on page 362
For DB2, refer to Automatic Restart Manager on page 394
For IMS, refer to IMS use of Automatic Restart Manager on page 421
Figure 6-1 ARM configuration
84
6.2 ARM components

There are three components to ARM, as explained here.
Each system in the sysplex must be connected to an ARM couple dataset.
There must be an active ARM policy.
A Parallel Sysplex is governed by various policies, such as CFRM policies and WLM
policies. A policy is a set of rules and actions that systems in a sysplex are to follow. A
policy allows MVS to manage specific resources in compliance with your system and
resource requirements, but with little operator intervention.
The ARM policy allows you to define how MVS is to manage automatic restarts of started
tasks and batch jobs that are registered as elements of Automatic Restart Manager. There
can be multiple ARM policies defined but only one ARM policy active at any one time.
The STC or job must register to ARM.
Registration is how the program communicates its restart requirements to ARM. A
program calls the ARM API, using the IXCARM macro. When a program registers, it is in
one of these states:
Starting
The element is executing and has registered.
Available
The element is executing, has registered, and has indicated that it is

ready for work.
Available-to
The element was restarted and has registered, but has not indicated it is
ready to work. After a time-out period has expired, ARM will consider it
available.
Failed
The element is registered and has terminated without deregistering. ARM

has not yet restarted it, or is beginning to restart it.
Restarting
The element failed, and ARM is restarting it. The element may be
executing and has yet to register again with ARM, or job scheduling
factors may be delaying its start.
WaitPred
The element is waiting for all predecessor programs to complete

initialization.
Recovering
The element has been restarted by ARM and has registered, but has not
indicated it is ready for work.
Note: A batch job or started task registered with Automatic Restart Manager can only be
restarted within the same JES XCF group. That is, it can only be restarted in the same
JES2 MAS or the same JES3 complex.
A system is considered ARM-enabled if it is connected to an ARM Couple Data Set. During
an IPL of an ARM-enabled system, the system indicates which ARM datasets it has
connected to, as shown in Figure 6-3 on page 86.
Chapter 6. Automatic Restart Manager
85
IXC286I COUPLE DATA SET 129

SYS1.XCF.ARM01,
VOLSER #@$#X1, HAS BEEN ADDED AS THE PRIMARY
FOR ARM ON SYSTEM #@$3
SYS1.XCF.ARM02,
VOLSER #@$#X2, HAS BEEN ADDED AS THE ALTERNATE
FOR ARM ON SYSTEM #@$3
. . .
IXC811I SYSTEM #@$3 IS NOW ARM CAPABLE
Figure 6-2 ARM messages issued during IPL
The D XCF,POLICY command displays the currently active ARM policy. Figure 6-3 displays a
system with the ARM policy ARMPOL01 active.
D XCF,POLICY,TYPE=ARM
TYPE: ARM
POLNAME:
ARMPOL01
STARTED:
06/22/2007 03:26:23
LAST UPDATED: 06/22/2007 03:25:58
1
2
3
Figure 6-3 Displaying ARM Policy
1 Currently active ARM policy

2 When the policy was activated or started
3 When the policy was defined
Use the D XCF,COUPLE command to display the ARM Couple Data Sets. Figure 6-4 on
page 87 displays a sysplex with two Couple Data Sets, namely SYS1.XCF.ARM01 and
SYS1.XCF.ARM02.
86
D XCF,COUPLE,TYPE=ARM
ARM COUPLE DATA SETS
PRIMARY
DSN: SYS1.XCF.ARM01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 15:08:01
4
FORMAT DATA
VERSION 1, HBB7707 SYMBOL TABLE SUPPORT
POLICY(20) MAXELEM(200) TOTELEM(200)
ALTERNATE DSN: SYS1.XCF.ARM02
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
11/20/2002 15:08:04
4
FORMAT DATA
VERSION 1, HBB7707 SYMBOL TABLE SUPPORT
POLICY(20) MAXELEM(200) TOTELEM(200)
ARM IN USE BY ALL SYSTEMS
Figure 6-4 Displaying ARM Couple Data Sets
1 The primary ARM couple dataset

2 The alternate ARM couple dataset
Figure 6-3 on page 86 shows that the current policy is ARMPOL01. To obtain more
information about this policy or any other defined but inactive policies, use the IXCMIAPU
utility as shown in Figure 6-5. This would normally be run by the system programmer.
//SFARMRPT JOB (999,SOFTWARE),'LIST ARM POLICY',CLASS=A,MSGCLASS=S,
//
NOTIFY=&SYSUID,TIME=1440,REGION=6M
//*
//STEP1
EXEC PGM=IXCMIAPU
//SYSPRINT DD
SYSOUT=*
//SYSABEND DD
SYSOUT=*
//SYSIN
DD
*
DATA TYPE(ARM) REPORT(YES)
//
Figure 6-5 Sample IXCMIAPU JCL to display the ARM policy report
Figure 6-6 on page 88 displays an extract from the ARM policy report. The report shows
items such as a RESTART_GROUP 1, an ELEMENT 2 and a TERMTYPE 3.
A restart_group is a logical connected group of elements that need to be restarted
together if the system they are on running fails. Not all the elements in a restart_group
need to be running on the same system, nor do they all need to be running.
We recommend that you set up a default group (RESTART_GROUP(DEFAULT)) with
RESTART_ATTEMPTS(0), so that any elements that are not defined as part of another
restart group are not restarted. All elements that do not fall into a specific restart group in
the policy are in the DEFAULT restart group.
The figure displays three restart groups: CICS#@$1, DB2DS1, and the default group.
87
RESTART_GROUP(CICS#@$1)
ELEMENT(SYSCICS_#@$CCC$1)
TERMTYPE(ELEMTERM)
. . .
RESTART_GROUP(DB2DS1)
ELEMENT(D#$#D#$1)
ELEMENT(DR$#IRLMDR$1001)
. . .
/* NO OTHER ARM ELEMENTS WILL BE RESTARTED */
RESTART_GROUP(DEFAULT)
ELEMENT(*)
RESTART_ATTEMPTS(0)
RESTART_ATTEMPTS(0,)
RESTART_TIMEOUT(120)
TERMTYPE(ALLTERM)
RESTART_METHOD(BOTH,PERSIST)
1
2
3
Figure 6-6 Sample job output displaying the active ARM policy
An element specifies a batch job or started task that can register as an element of
Automatic Restart Manager. The element name can use wild card characters of ? and * as
well as two system symbols, &SYSELEM. and &SYSSUF.
The termtype has two options:
ALLTERM - indicates restart if either the system or the element fails.
ELEMTERM - indicates restart only if the element fails. If the system fails, do not
restart.
For more information about the ARM policy, refer to MVS Setting Up a Sysplex, SA22-7625.
6.3 Displaying ARM status

To display the status of ARM on the console, issue the D XCF,ARMS command as shown in
Figure 6-7.
D XCF,ARMS
-------------- ELEMENT STATE SUMMARY -------------1
2
3
4
5
STARTING AVAILABLE FAILED RESTARTING RECOVERING
0
11
0
0
0
-TOTAL6
11
-MAX7
200
Figure 6-7 Output from D XCF,ARMSTATUS command
As displayed in Figure 6-7, D XCF,ARMS shows:

1 The total number of batch jobs and started tasks that are currently registered as elements
of ARM and are in the STARTING state.
of ARM which are in AVAILABLE state. (This also includes elements listed in AVAILABLE-TO
state.)
of ARM which are in FAILED state.
88
of ARM that are in RESTARTING state.
of ARM that are in RECOVERING state.
of ARM.
7 The maximum number of elements that can register. This information is determined by the
TOTELEM value when the ARM couple data set was formatted.
To see more detail, enter the command D XCF,ARMS,DETAIL. A significant amount of useful
detail can be displayed, as shown in Figure 6-8.
D XCF,ARMS,DETAIL
-------------- ELEMENT STATE SUMMARY -------------- -TOTAL- -MAXSTARTING AVAILABLE FAILED RESTARTING RECOVERING
0
11
0
0
0
11
200
RESTART GROUP:DEFAULT
PACING :
0
FREECSA:
0
ELEMENT NAME :EZA$1TCPIP
JOBNAME :TCPIP
STATE
:AVAILABLE
CURR SYS :#@$1
JOBTYPE :STC
ASID
:0023
INIT SYS :#@$1
EVENTEXIT:*NONE*
ELEMTYPE:SYSTCPIP LEVEL
:
1
TOTAL RESTARTS :
0
INITIAL START:06/22/2007 22:49:47
RESTART THRESH :
0 OF 0
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
PACING :
0
FREECSA:
0
JOBNAME :TCPIP
STATE
:AVAILABLE
CURR SYS :#@$2
JOBTYPE :STC
ASID
:0023
INIT SYS :#@$2
EVENTEXIT:*NONE*
:
1
TOTAL RESTARTS :
0
INITIAL START:06/22/2007 22:04:21
RESTART THRESH :
0 OF 0
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
PACING :
0
FREECSA:
0
JOBNAME :TCPIP
STATE
:AVAILABLE
CURR SYS :#@$3
JOBTYPE :STC
ASID
:0023
INIT SYS :#@$3
EVENTEXIT:*NONE*
:
1
TOTAL RESTARTS :
0
INITIAL START:06/22/2007 22:00:23
RESTART THRESH :
0 OF 0
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
PACING :
0
FREECSA:
0
. . .
1A
2A
3A
1B
2B
3B
1C
2C
3C
Figure 6-8 Displaying detailed ARM information
This portion of the display, which was restricted to TCPIP, shows the following details:
TCPIP belongs to the default ARM group 1A, 1B, 1C
TCPIP has an STC on each LPAR 2A, 2B, 2C
TCPIP has never been restarted by ARM 3A, 3B, 3C
89
6.4 ARM policy management

ARM policies may be changed for different reasons. For example, you might not want ARM
enabled during some maintenance windows. However, even while automatic restarts are
disabled, jobs and STCs still register to ARM when the system is ARM enabled; that is, when
there is an active ARM Couple Data Set.
6.4.1 Starting or changing the ARM policy

Before changing or activating a new ARM policy, it is best to know the current policy. Issue a
D XCF,POL,TYPE=ARM command. Figure 6-9 shows that the currently active ARM policy is
ARMPOL01 1.
D XCF,POL,TYPE=ARM
TYPE: ARM
POLNAME:
ARMPOL01
STARTED:
06/24/2007 21:17:30
LAST UPDATED: 06/16/2004 19:22:27
Figure 6-9 Display current ARM policy
To start or change an ARM policy at the request of a system programmer, issue the SETXCF
command. Figure 6-10 shows the ARM policy changed to ARMPOL02.
SETXCF START,POLICY,TYPE=ARM,POLNAME=ARMPOL02
IXC805I ARM POLICY HAS BEEN STARTED BY SYSTEM #@$2.
POLICY NAMED ARMPOL02 IS NOW IN EFFECT.
Figure 6-10 Starting or changing an ARM policy
If SETXCF START is issued without the POLNAME parameter, the ARM defaults are used.
You can find the default values in MVS Setting Up a Sysplex, SA22-7625. Figure 6-11
displays an example of starting an ARM policy without specifying a POLNAME.
SETXCF START,POLICY,TYPE=ARM
IXC805I ARM POLICY HAS BEEN STARTED BY SYSTEM #@$2.
POLICY DEFAULTS ARE NOW IN EFFECT.
Figure 6-11 Starting the default ARM policy
To stop an ARM policy without activating another one, issue the command SETXCF
STOP,POLICY,TYPE=ARM as shown in Figure 6-12. This would be done at the request of a
system programmer, usually during a maintenance window or disaster recovery exercise.
SETXCF STOP,POLICY,TYPE=ARM
IXC806I ARM POLICY HAS BEEN STOPPED BY SYSTEM #@$2
Figure 6-12 Stopping the ARM policy
90
6.4.2 Displaying the ARM policy status

The ARM policy is sysplex in scope. This means that when the ARM policy on system #@$2
was stopped, it was also stopped on systems #@$1 and #@$3. This can be seen when you
display the policy on all systems, as shown in Figure 6-13. Normally you would only need to
issue the command on a single system.
RO *ALL,D XCF,POL,TYPE=ARM
IEE421I RO *ALL,D XCF,POL,TYPE=A 638
#@$1
RESPONSES ----------------IXC364I 21.48.37 DISPLAY XCF 118
TYPE: ARM
POLICY NOT STARTED
#@$2
TYPE: ARM
POLICY NOT STARTED
#@$3
TYPE: ARM
POLICY NOT STARTED
Figure 6-13 Verify the ARM is stopped
6.5 Defining SDSF as a new ARM element

This section illustrates how to create a new ARM policy defining a new element, SDSF in this
case. It also demonstrates how ARM can work.
SDSF is fully documented in SDSF Operation and Customization, SA22-7670. Part of SDSF
is the server started task (STC). This is called SDSF in the following example.
6.5.1 Defining an ARM policy with SDSF

Figure 6-14 on page 92 shows part of the initial ARM policy used in this example, without
SDSF. Specifically, the RESTART_GROUP(DEFAULT) section 1 has restart_attempts(0)
defined.
91
RESTART_GROUP(CICS#@$1)
ELEMENT(SYSCICS_#@$CCC$1)
TERMTYPE(ELEMTERM)
ELEMENT(SYSCICS_#@$CCM$1)
TERMTYPE(ELEMTERM)
. . .
/* NO OTHER ARM ELEMENTS WILL BE RESTARTED */
RESTART_GROUP(DEFAULT)
ELEMENT(*)
RESTART_ATTEMPTS(0)
RESTART_ATTEMPTS(0,)
RESTART_TIMEOUT(120)
TERMTYPE(ALLTERM)
Figure 6-14 ARM policy without SDSF
There was no RESTART_GROUP for SDSF, which means that the command
C SDSF,ARMRESTART resulted in ARM not restarting SDSF, as shown in 2 in Figure 6-15.
C SDSF,ARMRESTART
IEA989I SLIP TRAP ID=X222 MATCHED. JOBNAME=SDSF
. . .
$HASP395 SDSF
ENDED
IEA989I SLIP TRAP ID=X33E MATCHED. JOBNAME=*UNAVAIL, ASID=0024.
IXC804I JOBNAME SDSF, ELEMENT ISFSDSF@$2 WAS NOT RESTARTED. 025
THE RESTART ATTEMPTS THRESHOLD HAS BEEN REACHED.
Figure 6-15 Cancel SDSF,ARMRESTART with no SDSF element
To create an ARM policy for SDSF, an updated ARM policy had to be created. The changes
made are shown in Figure 6-16.
/* SDSF */
RESTART_GROUP(SDSF)
ELEMENT(ISFSDSF*)
RESTART_METHOD(ELEMTERM,STC,'S SDSF')
RESTART_ATTEMPTS(3,60)
1
2
3
4
5
Figure 6-16 SDSF ARM policy changes
1 Add a comment that describes the section.

2 Define a group name. The group is only used by the ARM policy and can be anything you
like. It is recommended that you use a meaningful name. For instance, we could have created
a name AAA1 but instead we used a meaningful group name of SDSF. A group is a list of
jobs and STCs that all need to run on the same system. For example, there might be some
CICS regions that are tied into a DB2 region. By setting up a group that contains both the
DB2 and CICS regions, ARM is being told that if it starts DB2 on system A, it also needs to
start the CICS regions on system A. In this example, SDSF is a stand-alone element.
92
3 The element here is ISFSDSF*, where * is the standard wildcard matching 0 or more
characters. The element ID must match the registration ID. SDSF documentation states that
the registration ID for SDSF is ISFserver-name@&sysclone. Thus, on system #@$2, where
&SYSCLONE = $2, the registration ID is ISFSDSF$2. We could have defined three different
groups, one per system, but it is cleaner to create a wildcard entry that matches each system.
4 In this example we specify ELEMTERM, indicating that ARM is only to attempt a restart
when the element fails, that is, if the STC SDSF fails. The alternatives are to specify
SYSTERM, which means the restart only applies when the system fails, or to specify BOTH,
which means the restart method applies if either a system or an element fails. The second
part of this parameter says SDSF is a STC to be restarted via the S SDSF command.
5 In this example ARM will attempt to restart SDSF three times in 60 seconds. After the third
attempt, ARM will produce a message and not try to restart it. Automation should be set up to
trap on the IXC804I message; Figure 6-22 on page 95 shows this situation.
6.5.2 Starting SDSF

There are two ways to start SDSF, with ARM registration or without ARM registration. SDSF
Operation and Customization, SA22-7670, describes the two options displayed in
Figure 6-17.
S SDSF,. . .,ARM or
S SDSF,. . .,NOARM
ARM specifies that ARM registration will be done if ARM is active in the
system.
The server will register using the following values:
element name: ISFserver-name@&sysclone
element type: SYSSDSF
termtype: ELEMTYPE
NOARM specifies that ARM registration will not be done.
Figure 6-17 Start options for SDSF
When SDSF is started using the defaults, it registers itself to ARM as shown at 1 in
Figure 6-19 on page 94. Even if ARM is inactive when an STC or job such as SDSF starts, it
still successfully registers to ARM. Notice that nothing in Figure 6-18 indicates whether ARM
is active or inactive. Instead, it is the state of ARM and the ARM policy when the STC or job
fails that determines what happens.
S SDSF
ISF724I SDSF level HQX7730 initialization complete for server SDSF.
ISF726I SDSF parameter processing started.
ISF170I Server SDSF ARM registration complete for element type SYSSDSF,
element name ISFSDSF@$2
1
ISF739I SDSF parameters being read from member ISFPRM00 of data set
SYS1.PARMLIB
ISF728I SDSF parameters have been activated
Figure 6-18 Starting SDSF
93
6.5.3 Cancelling SDSF,ARMRESTART with no active ARM policy

When a job is cancelled with a cancel command, ARM will not attempt to restart it. The
ARMRESTART parameter is needed before ARM will attempt to restart it.
Figure 6-19 displays the result of the C SDSF,ARMRESTART command with no active ARM policy.
SDSF is not restarted, as shown at A.
C SDSF,ARMRESTART
. . .
IEF450I SDSF SDSF - ABEND=S222 U0000 REASON=00000000
$HASP395 SDSF
ENDED
IXC804I JOBNAME SDSF, ELEMENT ISFSDSF@$2 WAS NOT RESTARTED.
ARM RESTARTS ARE NOT ENABLED.
Figure 6-19 C SDSF,ARMRESTART with no active ARM policy
6.5.4 Cancelling SDSF,ARMRESTART with active ARM policy

SDSF not defined to ARM policy
Activating the ARM policy shown in Figure 6-14 on page 92 without an SDSF entry results in
SDSF using the default restart group (0 restart attempts). When the C SDSF,ARMRESTART
command is issued, SDSF is not restarted, as shown at B in Figure 6-20. This is because the
default restart_group has restart_attempts(0) defined.
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
Figure 6-20 C SDSF without SDSF defined correctly to the ARM policy
If you define an ARM policy with incorrect elements, such as ELEMENT(SDSF) and
ELEMENT(ISFSDSF#@$2), then issuing a C SDSF,ARMRESTART command produces the same
results as shown in Figure 6-20.
SDSF defined to ARM policy

By activating the ARM policy shown in Figure 6-16 on page 92 and issuing the C
SDSF,ARMRESTART command, SDSF is restarted by ARM, as shown in Figure 6-21 on page 95.
94
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
S SDSF
IXC812I JOBNAME SDSF, ELEMENT ISFSDSF@$2 FAILED.
THE ELEMENT WAS RESTARTED WITH OVERRIDE START TEXT.
IXC813I JOBNAME SDSF, ELEMENT ISFSDSF@$
WAS RESTARTED WITH THE FOLLOWING START TEXT:
S SDSF
THE RESTART METHOD USED WAS DETERMINED BY THE ACTIVE POLICY.
$HASP100 SDSF
ON STCINRDR
1
2
3
4
Figure 6-21 C SDSF,ARMRESTART with ARM restart working

1
2
3
4
ARM has determined a registered element has failed.

ARM restarts the failing element with the options specified in the policy.
ARM displays the restart text used.
SDSF has restarted.
6.5.5 ARM restart_attempts

If an error is causing the job or STC to continually fail, then the restart count comes into play.
Figure 6-16 on page 92 shows the SDSF restart_group defined with
RESTART_ATTEMPTS(3,60). This indicates that ARM is only to attempt three restarts in 60
seconds.
Figure 6-22 shows that the C SDSF,ARMRESTART command was issued repeatedly, and that
after the fourth cancel of SDSF within 60 seconds, the restart_attempts value in the ARM
policy took effect. As indicated at 1, ARM identified this as a problem and did not restart it.
It is useful to have an automation package trap message IXC804I and produce a highlighted
message for operators to act on. After the problem is rectified, the job or STC can be
restarted manually.
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
C SDSF,ARMRESTART
. . .
$HASP395 SDSF
ENDED
Figure 6-22 Multiple C SDSF,ARMRESTART
95
6.6 ARM and ARMWRAP

IBM developed a program called ARMWRAP that provides the ability to exploit ARM without
having to make changes to the application code. To prevent any program from using the ARM
facilities, SAF control was added to ARMWRAP. This allows RACF or other security products
to control its usage.
6.7.2, Cross-system restarts on page 100 explains how ARM can restart a job or STC on
different systems. In normal circumstances there is an instance of the SDSF server running
on every system; thus, it does not make sense to configure SDSF for a cross-system restart.
Instead, a procedure known as SLEEPY was created. It runs a program that loops through
24 stimer calls of 10 minutes each, so that it essentially sleeps for 4 hours and then finishes.
The SLEEPY program does not perform its own ARM registration. Instead, it makes use of
the facility ARMWRAP.
Figure 6-23 displays the JCL for PROC SLEEPY without making use of the ARMWRAP
facility.
//SLEEPY
//SLEEPY
//*
PROC
EXEC
PGM=SLEEPY
Figure 6-23 SLEEPY without ARMWRAP
To make use of the ARMWRAP facility, two steps must be added to the proc. These steps
can be seen in Figure 6-24.
//SLEEPY PROC
//*
//* Register element 'SLEEPY' element type 'APPLTYPE' with ARM
//* Requires access to SAF FACILITY IXCARM.APPLTYPE.SLEEPY
//*
//ARMREG EXEC PGM=ARMWRAP,
//
PARM=('REQUEST=REGISTER,READYBYMSG=N,',
//
'TERMTYPE=ALLTERM,ELEMENT=SLEEPY,','ELEMTYPE=APPLTYPE')
//*
//SLEEPY EXEC PGM=SLEEPY
//*
//* For normal termination, deregister from ARM
//*
//ARMDREG EXEC PGM=ARMWRAP,PARM=('REQUEST=DEREGISTER')
//SYSABEND DD SYSOUT=*
1
2
3
Figure 6-24 SLEEPY with ARMWRAP
1 This is the first step in the new PROC that runs the program ARMWRAP.
2 ARMWRAP takes parameters to register and define the ARM values such as ELEMTYPE,
ELEMENT and TERMTYPE.
3 The step (or steps) that form the proc are left as they were before.
4 When the proc finishes normally, it needs to deregister.
After SLEEPY is configured to work with ARMWRAP, it must be added to the ARM policy. In
our case, because we wanted SLEEPY to move to a different system in the event of a system
failure, we added the lines seen in Figure 6-25 on page 97 to the current policy.
96
It is not sufficient to add ARMWRAP to the STC or batch job. Instead, you must define it the
ARM policy. Figure 6-25 shows that we added a RESTART_GROUP for SLEEPY.
/* Sleepy */
RESTART_GROUP(SLEEPY)
TARGET_SYSTEM(#@$2,#@$3)
ELEMENT(SLEEPY)
RESTART_METHOD(BOTH,STC,'S SLEEPY')
RESTART_ATTEMPTS(3,60)
Figure 6-25 SLEEPY ARM policy with restart on another system
1 As shown in Figure 6-25, the element defined in the ARM policy must match the element
defined in the ARMWRAP parameters. We could have coded a generic element in the ARM
policy such as ELEMENT(SL*).
$HASP373 SLEEPY
STARTED
+ARMWRAP IXCARM REGISTER
RC = 000C RSN = 0168
-JOBNAME STEPNAME PROCSTEP
RC
EXCP
CPU
-SLEEPY
STARTING ARMREG
12
1
.00
Figure 6-26 Start SLEEPY without RACF profiles
When an attempt was made to start SLEEPY without defining the appropriate RACF profile,
the startup messages shown in Figure 6-26 were received. This attempt failed with an error 1.
z/OS V1R10.0 MVS Programming: Sysplex Services Reference, SA22-7618, identifies
IXCARM RC=12, RSN=168 as a security error, as shown in Figure 6-27.
Equate Symbol: IXCARMSAFNOTDEFINED
Meaning: Environmental error. Problem state and problem key users cannot use
IXCARM without having a security profile.
Action: Ensure that the proper IXCARM.elemtype.elemname resource profile for the
unauthorized application is defined to RACF or another security product.
Figure 6-27 IXCARM RC=12 RSN=168
The RACF commands used to protect this resource are shown in Figure 6-28.
RDEFINE FACILITY IXCARM.APPLTYPE.SLEEPY
PE
IXCARM.APPLTYPE.SLEEPY ID(SLEEPY) AC(UPDATE) CLASS(FACILITY)
SETROPTS RACLIST(FACILITY) REFRESH
Figure 6-28 RACF commands to protect ARM - SLEEPY
When this is done, an attempt to start SLEEPY works, as shown in Figure 6-29 on page 98.
97
IEF695I START SLEEPY

WITH JOBNAME SLEEPY
IS
$HASP373 SLEEPY
STARTED
RC = 0000 RSN = 0000
+ARMWRAP IXCARM READY
RC = 0000 RSN = 0000
1
2
Figure 6-29 Start SLEEPY with RACF profiles set up
1 Sleepy has registered with ARM.

2 Sleepy is now ready to work with ARM.
To verify that Sleepy has registered to ARM, issue the D XCF,ARMS command as shown in
Figure 6-30.
D XCF,ARMS,ELEMENT=SLEEPY,DETAIL
-------------- ELEMENT STATE SUMMARY --------------TOTAL- -MAXSTARTING AVAILABLE FAILED RESTARTING RECOVERING
0
1
0
0
0
1
200
RESTART GROUP:SLEEPY
PACING :
0
FREECSA:
0
0
ELEMENT NAME :SLEEPY
JOBNAME :SLEEPY
STATE
:AVAILABLE
CURR SYS :#@$2
JOBTYPE :STC
ASID
:004C
INIT SYS :#@$2
JESGROUP:XCFJES2A TERMTYPE:ALLTERM
EVENTEXIT:*NONE*
ELEMTYPE:APPLTYPE LEVEL
:
2
TOTAL RESTARTS :
0
INITIAL START:06/25/2007 19:31:15
RESTART THRESH :
0 OF 0
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
1
2
3
Figure 6-30 D XCF,ARMS,ELEMENT=SLEEPY,DETAIL
1 There is one available element.

2 The restart group is SLEEPY.
3 The Element name is SLEEPY and it is an STC.
6.7 Operating with ARM

Depending on the situation and the setup, ARM will restart subsystems on the same system
or on an alternative system. The following sections describe both scenarios.
6.7.1 Same system restarts

To invoke ARM for a particular application, as previously seen with SDSF, it must be
cancelled or forced with the ARMRESTART parameter. This is used in the scenario where a
critical application is hung and recovery time is crucial.
Note: If you require documentation for support, use the DUMP parameter as well. Issue
the C jobname,DUMP,ARMRESTART command.
During restart, you will to see the STATE change to RESTARTING 1, as shown in Figure 6-31
on page 99.
98
RESTART GROUP:SDSF
ELEMENT NAME :ISFSDSF@$2
CURR SYS :#@$2
INIT SYS :#@$2
EVENTEXIT:*NONE*
TOTAL RESTARTS :
1
RESTART THRESH :
0 OF 3
RESTART TIMEOUT:
300
PACING :
0
FREECSA:
0
0
JOBNAME :SDSF
STATE
:RESTARTING
JOBTYPE :STC
ASID
:004A
ELEMTYPE:SYSSDSF LEVEL
:
2
INITIAL START:06/25/2007 03:36:28
FIRST RESTART:06/25/2007 03:39:16
LAST RESTART:06/25/2007 03:39:16
Figure 6-31 Status of element during restart by ARM
After the element is started, the ARM status changes to AVAILABLE 1, as shown in
Figure 6-32. It will also increment the Restart Thresh count 2.
RESTART GROUP:SDSF
CURR SYS :#@$1
INIT SYS :#@$1
EVENTEXIT:*NONE*
TOTAL RESTARTS :
2
RESTART THRESH :
1 OF 3
RESTART TIMEOUT:
300
PACING :
0
FREECSA:
0
JOBNAME :SDSF
STATE
:AVAILABLE
JOBTYPE :STC
ASID
:0022
:
2
INITIAL START:06/22/2007 22:49:17
FIRST RESTART:06/25/2007 03:49:16
LAST RESTART:06/25/2007 03:49:16
0
1
Figure 6-32 Status of element after restart by ARM
Note: ARM does not restart elements in the following instances. Therefore the operator, or
an automation product, must manually intervene.
Canceling a job without the ARMRESTART parameter.
*F J=jobname,C (JES3 cancel command without the ARMRESTART parameter).
Batch jobs in a JES3 DJC net.
During a system shutdown and the policy for the element has
TERMTYPE(ELEMTERM). A TERMTYPE(ALLTERM) is required for system failure
restarts.
During a system shutdown when there is only one target system defined in the policy
for that element.
If the element has not been restarted by ARM, the initial start, first restart, and last restart
output is updated in the display. Figure 6-33 on page 100 shows that, for System #@$1 the
fields FIRST RESTART: and LAST RESTART: have the value *NONE*, which indicates that ARM
has not restarted SDSF on this system. In contrast, System #@$2 has times for these fields.
Likewise, note that the total number of restarts is 0 for #@$1 and it is 2 for #@$2.
99
System #@$1
RESTART GROUP:SDSF
CURR SYS :#@$1
INIT SYS :#@$1
EVENTEXIT:*NONE*
TOTAL RESTARTS :
0
RESTART THRESH :
0 OF 3
RESTART TIMEOUT:
300
PACING :
0
FREECSA:
0
JOBNAME :SDSF
STATE
:AVAILABLE
JOBTYPE :STC
ASID
:0022
:
2
INITIAL START:06/22/2007 22:49:17
LAST RESTART:*NONE*
System #@$2
RESTART GROUP:SDSF
CURR SYS :#@$2
INIT SYS :#@$2
EVENTEXIT:*NONE*
TOTAL RESTARTS :
2
RESTART THRESH :
0 OF 3
RESTART TIMEOUT:
300
PACING :
0
FREECSA:
0
JOBNAME :SDSF
STATE
:AVAILABLE
JOBTYPE :STC
ASID
:004B
:
2
INITIAL START:06/25/2007 03:36:28
FIRST RESTART:06/25/2007 03:39:16
LAST RESTART:06/25/2007 03:41:05
Figure 6-33 Comparative ARM displays
6.7.2 Cross-system restarts

If SLEEPY is active on system #@$2 and we issue a C SLEEPY,ARMRESTART command, then it
restarts on #@$2 in a manner similar to that seen with SDSF. However, what happens if
system #@$2 fails or is not stopped cleanly? Figure 6-34 displays the log from system #@$3.
Notice that system #@$2 was removed from the sysplex and then SLEEPY was started.
#@#3 Syslog
IXC105I SYSPLEX PARTITIONING HAS COMPLETED FOR #@$2
. . .
INTERNAL S SLEEPY
. . .
$HASP373 SLEEPY
STARTED
RC = 0000 RSN = 0000
+ARMWRAP IXCARM READY
RC = 0000 RSN = 0000
1
2
Figure 6-34 SLEEPY restarting on #@$3
1 System #@$2 is partitioned out of the sysplex.

2 XCF issues an S SLEEPY command.
3 SLEEPY is registered onto system #@#3.
ARM restarts SLEEPY on another system because SLEEPY did not de-register. If system
#@#2 was shut down cleanly, then SLEEPY would finish and successfully deregister itself to
ARM; thus, ARM would not restart it.
100
Chapter 7.
Coupling Facility considerations

in a Parallel Sysplex
This chapter provides details of operational considerations of a Coupling Facility. It includes:
Overview of the CF
Displaying the CF
Displaying structures in the CF
Managing the CF
Rebuilding and moving structures
101
7.1 Introduction to the Coupling Facility

The Coupling Facility (CF) plays a key role in the Parallel Sysplex infrastructure. Whether the
CF is implemented as stand-alone or in an LPAR, it allows multisystem access to data.
The stand-alone CF provides the most robust CF capability, because the CEC is wholly
dedicated to running the CFCC microcode. All of the processors, links, and memory are for
Coupling Facility use only.
Running the CFCC in a PR/SM LPAR on a server is the same as running it on a stand-alone
model. Distinctions are mostly in terms of price/performance, maximum configuration
characteristics, CF link options, and recovery characteristics.
The Coupling Facility architecture uses hardware, specialized Licensed Internal Code (LIC),
and enhanced z/OS and subsystem code. All these elements form an integral part of a
Parallel Sysplex configuration.
7.2 Overview of the Coupling Facility

The Coupling Facility has two aspects, software and hardware, as explained here:
Software
Consists of CFCC Licensed Internal Code (conceptually similar to an operating
system).
CF levels
XES - the interface from z/OS
Structures
Lock: For serialization of data with high granularity. Global Resource Serialization
(GRS) is an example of a Lock structure exploiter.
List: For shared queues and shared status information. System Logger is an
example of a List structure exploiter.
Cache: For storing data and maintaining local buffer pool coherency information.
RACF database sharing is an example of a Cache structure exploiter.
For a current list of CF structure names and exploiters, refer to Appendix B, List of
structures on page 499.
Hardware
Processor
Channels (links) and subchannels
Storage
The Coupling Facility can be configured either stand-alone or in an LPAR on a CEC
alongside operating systems such as z/OS and z/VM. The Coupling Facility does not have
any connected I/O devices, and the only console interface to it is through the HMC.
Connectivity to the CF is with CF links, which can be a combination of the following:
Inter-System Channel (ISC)
Integrated Cluster Bus (ICB)
Internal Coupling Channel (IC)
102
A description of the various System z channel and CHPID types can be found at:
http://www.redbooks.ibm.com/abstracts/tips0086.html?Open
A Coupling Facility possesses unique attributes:
You can shut it down, upgrade it, and bring it online again without impacting application
availability.
You can potentially lose a CF without impacting the availability of the applications that are
using that CF.
The amount of real storage in a CF depends on several factors:
Space for the Coupling Facility Control Code (CFCC)

Dump space
Space for allocated structures
Space for failover of structures from another CF
Space for growth
These space requirements will vary with each CF level.

For the most current information about CF levels and the enhancements introduced at each
CF level, refer to:
http://www.ibm.com/systems/z/pso/cftable.html#HDRCFLVLCN
7.3 Displaying a Coupling Facility

This section includes operator commands that you can use to display and monitor a CF.
7.3.1 Displaying the logical view of a Coupling Facility

By issuing the command shown in Figure 7-1, you can display the logical view of the CF. To
determine the names of the CF in the Parallel Sysplex, issue the MVS D XCF,CF command as
seen in Figure 7-1.
D XCF,CF
CFNAME
COUPLING FACILITY
FACIL01 1 SIMDEV.IBM.EN.0000000CFCC1
PARTITION: 00
CPCID: 00
FACIL02
PARTITION: 00
CPCID: 00
SITE
N/A
N/A
Figure 7-1 Logical view of all CFs
1 Name of the CF
To display the logical view of one of the CFs identified in Figure 7-1, issue the command D
XCF,CF,CFNAME=cfname as seen in Figure 7-2 on page 104.
Chapter 7. Coupling Facility considerations in a Parallel Sysplex
103
CFNAME: FACIL01 1
COUPLING FACILITY
:
SIMDEV.IBM.EN.0000000CFCC1 2
PARTITION: 00 3 CPCID: 00 4
SITE
:
N/A
2000 K 5
2048 K 6
256 K 7
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES: 9
D#$#_LOCK1(OLD)
DFHNCLS_#@$CNCS1
IXC_DEFAULT_2
D#$#_SCA(OLD)
DFHXQLS_#@$STOR1
SYSTEM_OPERLOG
DFHCFLS_#@$CFDT1
IRRXCF00_P001
Figure 7-2 Logical view of a particular CF
1 The name of the CF

2 Node descriptor (Type.Manufacturer.Plant.Sequence)
3 Partition identifier
4 CPC identifier
5 Policy dump space size
6 Actual dump space size
7 Storage increment size
8 Systems connected to this CF
9 Structures that are currently residing in this CF
7.3.2 Displaying the physical view of a Coupling Facility

By issuing the MVS D CF command, you can display the physical view of the CF; see
Figure 7-3 on page 105. The output of this command displays information about CF
connectivity to systems and the CF characteristics.
104
D CF
. . .
SENDER PATH
09
0E
PHYSICAL
ONLINE
ONLINE
LOGICAL
ONLINE
ONLINE

TOTAL:
6
IN USE:
6
NOT USING:
DEVICE
SUBCHANNEL
STATUS
4030
0004
OPERATIONAL
4031
0005
OPERATIONAL
4032
0006
OPERATIONAL
4033
0007
OPERATIONAL
4034
0008
OPERATIONAL
4035
0009
OPERATIONAL
CHANNEL TYPE
ICP 1
ICP
NOT USABLE:
REMOTELY CONNECTED COUPLING FACILITIES

CFNAME
COUPLING FACILITY
--------------------------------FACIL02 2
CHPIDS ON FACIL01 CONNECTED TO REMOTE FACILITY
RECEIVER:
CHPID
TYPE
F0
ICP 3
SENDER:
CHPID
E0
TYPE
ICP
. . .
Figure 7-3 Physical view of a CF
If the name of the CF is known, you can expand on the command in Figure 7-3 by issuing D
CF,CFNAME=FACIL01.
The output from the D CF command in Figure 7-3 displays information that includes the CF
link channel type in use.
1 CF channel type in use on FACIL01 is an Internal Coupling Channel Peer mode.
2 FACIL01 CF is connected to another CF named FACIL02.
3 The CF CHPIDs on FACIL01 are connected to CF FACIL02 using Internal Coupling
Channel Peer mode links.
7.3.3 Displaying Coupling Facility structures

The D XCF,STRUCTURE command provides a complete list of all structures defined in the active
CFRM Policy and the current status of each structure. A sample display of the output from
this command is shown in Figure 7-4 on page 106.
105
D XCF,STR
STRNAME
ALLOCATION TIME
CIC_DFHLOG_001
--CIC_DFHSHUNT_001
--CIC_GENERAL_001
--D#$#_GBP0
--D#$#_GBP1
--D#$#_GBP32K
--D#$#_GBP32K1
--D#$#_LOCK1
06/20/2007 03:32:17
STATUS
TYPE
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
ALLOCATED (NEW)3
LOCK
DUPLEXING REBUILD
D#$#_LOCK1
06/20/2007 03:32:15 ALLOCATED (OLD)
LOCK
DUPLEXING REBUILD
D#$#_SCA
06/20/2007 03:32:12 ALLOCATED (NEW)
LIST
DUPLEXING REBUILD
D#$#_SCA
06/20/2007 03:32:10 ALLOCATED (OLD)
LIST
DUPLEXING REBUILD
DFHCFLS_#@$CFDT1 06/21/2007 01:47:27 ALLOCATED 1
LIST
DFHNCLS_#@$CNCS1 06/21/2007 01:47:24 ALLOCATED
LIST
DFHXQLS_#@$STOR1 06/21/2007 01:47:22 ALLOCATED
LIST
IRRXCF00_B001
06/22/2007 21:59:18 ALLOCATED 2
CACHE
IRRXCF00_P001
06/22/2007 21:59:17 ALLOCATED
CACHE
. . .
Figure 7-4 Display of all CF structures
1 DFHCFLS_#@$CFDT1 was allocated at 01:47:27 on 06/21/2007 and is a List structure.

2 IRRXCF00_B001 was allocated at 21:59:18 on 06/22/2007 and is a Cache structure.
3 D#S#_LOCK1 was allocated at 03:32:17 on 06/20/2007 and is a Lock structure.
To display the structures that are currently defined to a particular CF, issue the MVS D
XCF,CF,CFNAME=cfname command, where cfname is the name of the CF; see Figure 7-5 on
page 107.
106
CFNAME: FACIL01
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
2
CIC_DFHSHUNT_001
D#$#_LOCK1(OLD)
DFHNCLS_#@$CNCS1
IXC_DEFAULT_2
D#$#_GBP0(NEW)
D#$#_SCA(OLD)
DFHXQLS_#@$STOR1
SYSTEM_OPERLOG
D#$#_GBP1(NEW)
DFHCFLS_#@$CFDT1
IRRXCF00_P001
Figure 7-5 Display of structures in a particular CF
In Figure 7-5, the display shows a number of CF structures that are present in the CF named
FACIL01.
1 Systems #@$1, #@$2 and #@$3 are currently connected to this CF.
2 List of structures that are currently residing in this CF.
7.3.4 Displaying information about a specific structure

To gather detailed information on a particular CF structure, issue the MVS command shown
in Figure 7-6 on page 108.
107
D XCF,STR,STRNAME=SYSTEM_OPERLOG 1
STRNAME: SYSTEM_OPERLOG
STATUS: ALLOCATED 2
TYPE: LIST 3
POLICY INFORMATION:
POLICY SIZE
: 16384 K
FULLTHRESHOLD : 0
ALLOWAUTOALT
: NO
REBUILD PERCENT: N/A
DUPLEX
: DISABLED
PREFERENCE LIST: FACIL01 FACIL02 4
ENFORCEORDER
: NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/18/2007 03:43:48
CFNAME
: FACIL01
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 9216 K
ENTRIES: IN-USE:
6118 TOTAL:
6118, 100% FULL
ELEMENTS: IN-USE:
12197 TOTAL:
12341, 98% FULL
PHYSICAL VERSION: C0C39A43 CCB4260C
LOGICAL VERSION: C0C39A43 CCB4260C
DISPOSITION
: DELETE 5
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 1 6
7
CONNECTION NAME ID VERSION SYSNAME JOBNAME ASID STATE
---------------- -- -------- -------- -------- ---- ---------------IXGLOGR_#@$1
01 000100CC #@$1
IXGLOGR 0016 FAILED-PERSISTENT
DIAGNOSTIC INFORMATION:
STRNUM: 0000000D STRSEQ: 00000001


Figure 7-6 Displaying detailed information for a particular CF structure
1 The name of the structure that detailed information is being gathered for. In this example,
detailed information for the SYSTEM_OPERLOG structure is being requested.
2 Identifies whether the structure is allocated or not.
3 The structure type. In this example, it is a List structure.
108
4 The Preference List as defined in the active Coupling Facility Resource manager (CFRM)
Policy. It displays the desired order of CFs as to where the structure should normally be.
5 The disposition of the structure.
6 The number of systems that are connected to this structure.
7 The connection names, system names, jobnames and states of the connection.
Note: The D XCF,STR,STRNM=ALL command displays all defined structures in detail for all
CFs.
7.3.5 Structure and connection disposition

Each structure has a disposition associated with it, and each connection also has a
disposition. This section explains the differences for each.
Structure disposition
There are two disposition types for structures:
DELETE
This implies that as soon as the last connected exploiter disconnects

from the structure, the structure is deallocated from the CF processor
storage. Examples of structures that use this disposition include
SYSTEM_OPERLOG and ISGLOCK. A deallocation of the structure
occurs when all address spaces related to the structure are shut
down.
KEEP
This indicates that even though there are no more exploiters

connected to the structure, because of normal or abnormal
disconnection, the structure is to remain allocated in the CF processor
storage. Examples of structures that use this disposition include
JES2_CHKPT1 and IGWLOCK00 (VSAM/RLS). To manually
deallocate a structure with a disposition of keep, you must force the
structure out of the CF using the SETXCF FORCE command.
Attention: Use the SETXCF FORCE command with caution. Inform support staff before
proceeding.
Connection state and disposition

The connection of a structure can be one of four states:
UNDEFINED
The connection is not established.
ACTIVE
The connection is currently being used.
FAILED-PERSISTENT
The connection has abnormally terminated but is logically

remembered, although it is not physically active.
DISCONNECTING or FAILING
The connection has disconnected or failed.
At connection time, another parameter indicates the disposition of the connection. The state
of the connection depends on the disposition of the structure.
A connection with a disposition of keep is placed in a failed-persistent state if it terminates
abnormally, or if the owner of the structure has defined it this way (for example IMS). When in
the failed-persistent state, a connection becomes active again as soon as the connectivity to
the structure is recovered. The failed-persistent state can be thought of as a placeholder for
the connection to be recovered. Note that in some special cases, a connection with a
disposition of keep may be left in an undefined state even after an abnormal termination.
109
A connection with a disposition of delete is placed in an undefined state when it terminates

normally. When the connectivity to the structure is recovered, the exploiter must establish a
new connection.
Connections can be displayed by using the DISPLAY XCF,STRUCTURE command. For more
information about displaying connection attributes, see 7.3.6, Displaying connection
attributes on page 110.
7.3.6 Displaying connection attributes

The DISPLAY XCF,STRUCTURE,STRNM=strname,CONNM=connm command can be used to display
connection attributes. To identify the connection name, we will issue the MVS command to
display the required structure.
In the example in Figure 7-7 on page 111, we use the SYSTEM_OPERLOG structure.
110
D XCF,STR,STRNM=SYSTEM_OPERLOG
STATUS: ALLOCATED
TYPE: LIST
POLICY INFORMATION:
POLICY SIZE
: 16384 K
FULLTHRESHOLD : 0
ALLOWAUTOALT
: NO
DUPLEX
: DISABLED
ENFORCEORDER
: NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/18/2007 03:43:48
CFNAME
: FACIL01
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 9216 K
ENTRIES: IN-USE:
3246 TOTAL:
6118,
ELEMENTS: IN-USE:
6846 TOTAL:
12341,
PHYSICAL VERSION: C0C39A43 CCB4260C
LOGICAL VERSION: C0C39A43 CCB4260C
DISPOSITION
: DELETE
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3
53% FULL
55% FULL
1 CONNECTION NAME ID VERSION SYSNAME JOBNAME ASID STATE

---------------- -- -------- -------- -------- ---- -----------IXGLOGR_#@$1
03 00030077 #@$1
IXGLOGR 0016 ACTIVE
IXGLOGR_#@$2
01 000100CE #@$2
IXGLOGR 0016 ACTIVE
IXGLOGR_#@$3
02 00020065 #@$3
IXGLOGR 0016 ACTIVE
DIAGNOSTIC INFORMATION: STRNUM: 0000000D STRSEQ: 00000000
Figure 7-7 Displaying connection attributes for a particular CF structure
1 The output displayed in Figure 7-7 identifies all of the connection names from the various
systems.
111
Now that the connection names have been identified for the SYSTEM_OPERLOG structure,
we can display the individual connection attributes for the structure. In this example, we
display the ConnectionName from our system #@$3.
D XCF,STR,STRNM=SYSTEM_OPERLOG,CONNM=IXGLOGR_#@$3
...
CONNECTION NAME : IXGLOGR_#@$3
ID
: 02
VERSION
: 00020065 1
CONNECT DATA
: 00000001 00000000
SYSNAME
: #@$3
JOBNAME
: IXGLOGR
2
ASID
: 0016
STATE
: ACTIVE
3
FAILURE ISOLATED FROM CF
CONNECT LEVEL
: 00000000 00000000
INFO LEVEL
: 01
CFLEVEL REQ
: 00000001
NONVOLATILE REQ : YES
CONDISP
: KEEP
4
ALLOW REBUILD
: YES
5
ALLOW DUPREBUILD: NO
ALLOW AUTO
: YES
SUSPEND
: YES
ALLOW ALTER
: YES
6
USER ALLOW RATIO: YES
USER MINENTRY
: 10
USER MINELEMENT : 10
USER MINEMC
: 25
DIAGNOSTIC INFORMATION: STRNUM: 0000000D STRSEQ: 00000000
Figure 7-8 Displaying connection name details for a particular CF structure
1 The version of this connection. This is needed to distinguish it from other connections with
the same name on the same system for the same jobname; this was done, for example, after
a connection failure that was recovered.
2 This connection was done for the jobname IXGLOGR.
3 The connection is active.
4 The connection IXGLOGR_#@$3 has a connection disposition of KEEP.
5 The connection supports REBUILD.
6 The connection supports ALTER.
7.4 Structure duplexing

Not all CF exploiters provide the ability to recover from a structure failure. For those that do,
certain types of structure failures require disruptive recovery processes. Recovery from a
structure failure can be time-consuming, even for exploiters that provide recovery support. To
address these concerns, structure duplexing is used.
112
There are two types of structure duplexing:

User-managed duplexing
System-managed duplexing
User-managed duplexing is only used by DB2 for its Group Buffer Pools. System-managed
duplexing is available for any structure that supports system-managed processes.
7.4.1 System-managed Coupling Facility (CF) structure duplexing

System-managed CF structure duplexing is designed to provide a general purpose,
hardware- assisted mechanism for duplexing CF structure data. This can provide a robust
recovery mechanism for failures, such as a loss of a single structure or CF, or loss of
connectivity to a single CF, through rapid failover to the other structure instance of the duplex
pair.
Benefits of CF duplexing include:
Availability
Manageability and usability
Configuration benefits (failure isolation)
With system-managed CF structure duplexing, two instances of the structure exist, one on
each of the CFs. This eliminates the single point of failure when a data sharing structure is on
the same server as one of its connectors.
Figure 7-9 on page 114 depicts how a request to a system-managed duplexed structure is
processed.
Note: User-managed duplexing of DB2 group buffer pools does not operate this way.
A request is sent to XES from the application or subsystem that is connected to a
duplexed structure. The exploiter does not need to know if the structure is duplex or not.
XES sends requests 2a and 2b separately to the primary and secondary structures in the
CF. XES will make both either synchronous or asynchronous.
Before the request is processed by the CFs, a synchronization point is taken between the
two CFs, 3a and 3b.
Note: In user-managed duplexing mode, the request to the secondary structure is always
asynchronous.
The request is then executed by each CF.
After the request is performed, a synchronization point is taken between the CFs again.
The result of the request is returned from each CF to XES. These requests, 6a and 6b, are
checked for consistency.
Finally, the result is returned to the exploiter.
113
3a+b. Exchange of Ready to Execute signals
CF2
CF1
5a+b. Exchange of Ready to Complete signals
4. Execute request
2a. Split request out
4. Execute request
6a. Response
z/OS
2b. Split request out
XES
(split/merge)
6b. Response
1. Request in
Exploiter
7. Response out
Figure 7-9 Request process for system-managed CF structure duplexing
7.4.2 Rebuild support history

User-managed rebuild
Introduced in MVS 5.1
Support for recovery from CF or CF link failure
Driven by support in the exploiter
Introduced in OS/390 2.6
Supports duplex copies of Group Buffer Pool (GBP) structures
Driven by support in the exploiter
System-managed rebuild
Introduced in OS/390 2.8
Much easier for exploiters to use the CF
Does not support recovery from a CF failure
Introduced in z/OS V1.2
When used with system-managed rebuild structures, provides recovery from a CF
failure
Transparent to exploiter, but performance must be considered
7.4.3 Difference between user-managed and system-managed rebuild

This section highlights the difference between these two types of rebuild.
114
User-managed rebuild:
It involves complex programming to ensure structure connectors communicate with each
other and XES to move the structure contents from one CF to another.
It requires that 12 specific events must be catered for, as well as handling error and
out-of-sequence situations.
The level of complex programming can be time-consuming, expensive, and error-prone.
The structure can only be moved to another CF if there is still one connector active.
Each exploiter of a CF structure must design and code its own solution. Therefore, some
exploiters do not provide rebuild capability (for example, JES2).
It leads to complex and differing operational procedures to handle planned and unplanned
CF outages.
System-managed rebuild:
It removes most of the complexity from the applications.
The actual movement of structure contents is handled by XES. This means that every
structure that supports system-managed rebuild is handled consistently.
Failure and out-of-sequence events are handled by XES.
Rebuild support is easier for CF exploiters.
It provides consistent operational procedures.
It can rebuild a structure when there are no active connectors.
It provides support for planned CF reconfigurations.
It is not for recovery scenarios.
7.4.4 Enabling system-managed CF structure duplexing

The relevant structure definitions need to be updated in a new CFRM policy and activated to
include the DUPLEX keyword.
Note: Although there is value in CF duplexing, IBM does not recommend its use in all
situations, nor should it necessarily be used in every environment for the structures that
support it.
The updated CFRM Policy will then need to be activated using the command SETXCF
START,POL,POLNM=policyname,TYPE=CFRM.
Additional information can be found in the IBM Technical Paper titled System -Managed CF
Structure Duplexing. It is available at the following URL:
http://www.ibm.com/servers/eserver/zseries/library/techpapers/gm130103.html
The CF duplexing function for a structure can be started and stopped by MVS commands.
There are two ways to start duplexing:
Activate a new CFRM policy with DUPLEX(ENABLED) keyword for the structure.
If the old structure is currently allocated, then z/OS will automatically initiate the process to
establish duplexing as soon as you activate the policy.
If the structure is not currently allocated, then the duplexing process will be initiated
automatically when the structure is allocated.
115
Activate a new CFRM policy with DUPLEX(ALLOWED) keyword for the structure.
This method allows the structures to be duplexed; however, the duplexing must be
initiated by command because z/OS will not automatically duplex the structure.
Duplexing can then be initiated using the SETXCF START,REBUILD,DUPLEX command or
programmatically via the IXLREBLD STARTDUPLEX programming interface.
Duplexing can be manually stopped by using the SETXCF STOP,REBUILD,DUPLEX command or
programmatically via the IXLREBLD STOPDUPLEX programming interface. When you need
to stop duplexing structures, you must first decide which is to remain as the surviving simplex
structure.
You can also stop duplexing of a structure in a particular CF by issuing the command SETXCF
STOP,REBUILD,DUPLEX,CFNAME=cfname
For further information about each system command of CF Duplexing, including SETXCF,
refer to z/OS MVS System Commands, SA22-7627.
7.4.5 Identifying which structures are duplexed

To identify which structures are defined in the active CFRM Policy as being system-managed
CF structure duplexed, issue the MVS D XCF,STR,STRNM=strname command that is shown in
116
D XCF,STR,STRNM=ALL
STRNAME: CIC_DFHLOG_001 1
STATUS: NOT ALLOCATED
POLICY INFORMATION:
POLICY SIZE
: 32756 K
FULLTHRESHOLD : 80
ALLOWAUTOALT
: NO
DUPLEX
: DISABLED 2
ENFORCEORDER
: NO
STRNAME: D#$#_GBP0 3
POLICY INFORMATION:
POLICY SIZE
: 8192 K
FULLTHRESHOLD : 80
ALLOWAUTOALT
: YES
DUPLEX
: ENABLED 4
ENFORCEORDER
: NO
STRNAME: LOG_FORWARD_001 5
POLICY INFORMATION:
POLICY SIZE
: 16384 K
FULLTHRESHOLD : 0
ALLOWAUTOALT
: NO
DUPLEX
: ALLOWED 6
ENFORCEORDER
: NO
...
Figure 7-10 Displaying all CF structures
1 Structure name of CIC_DFHLOG_001.

2 Duplexing has been DISABLED for structure CIC_DFHLOG_001 in the active CFRM Policy.
3 Structure name of D#$#_GBP0.
4 Duplexing has been ENABLED for structure D#$#_GBP0 in the active CFRM Policy.
5 Structure name of LOG_FORWARD_001.
6 Duplexing has been ALLOWED for structure LOG_FORWARD_001 in the active CFRM Policy.
117
As shown in Figure 7-10 on page 117, the DUPLEX field can have a value of DISABLED,
ENABLED, and ALLOWED, as explained here:
DISABLED
ENABLED
ALLOWED
Duplexing cannot be started for this structure.

Duplexing of this structure may be started either manually or automatically.
Duplexing of this structure may be started manually but will not be started
automatically.
To obtain detailed duplexing information about a particular structure, you can use the output
from Figure 7-10 on page 117 and use the MVS command D XCF,STR,STRNAME=strname.
D XCF,STR,STRNAME=D#$#_GBP0
STRNAME: D#$#_GBP0
STATUS: REASON SPECIFIED WITH REBUILD START:
POLICY-INITIATED
DUPLEXING REBUILD
METHOD: USER-MANAGED
PHASE: DUPLEX ESTABLISHED 1
...
DUPLEXING REBUILD NEW STRUCTURE
------------------------------ALLOCATION TIME: 06/26/2007 19:37:34
CFNAME
: FACIL01 2
PARTITION: 00
CPCID: 00
...
DUPLEXING REBUILD OLD STRUCTURE
------------------------------ALLOCATION TIME: 06/26/2007 19:37:31
CFNAME
: FACIL02 3
PARTITION: 00
CPCID: 00
...
4 CONNECTION NAME ID VERSION SYSNAME JOBNAME
---------------- -- -------- -------- -------DB2_D#$1
02 00020045 #@$1
D#$1DBM1
DB2_D#$2
03 0003003E #@$2
D#$2DBM1
DB2_D#$3
01 00010048 #@$3
D#$3DBM1
...
ASID
---004B
004C
0024
STATE
---------------ACTIVE NEW,OLD
ACTIVE NEW,OLD
ACTIVE NEW,OLD
Figure 7-11 Display of a duplexed structure
1 Duplexing status for the structure.

2 Duplexed D#$#_GBP0 structure allocated in CF FACIL01.
3 Original D#$#_GBP0 structure allocated in CF FACIL02.
4 DB2 subsystems from z/OS systems #@$1, #@$2, and #@$3 are currently connected to
the D#$#_GBP0 duplexed structure.
118
7.5 Structure full monitoring

Structure full monitoring adds support for the monitoring of objects within a Coupling Facility
structure. Its objective is to determine the level of usage for objects that are monitored within
a CF, and to issue a warning message if a structure full condition is imminent. This will allow
an installation to intervene, either manually or through automation, so that the appropriate
diagnostic or tuning actions can be taken to avoid a structure full condition.
Structure full monitoring, running on a given system, will periodically retrieve structure
statistics for each of the active structure instances from each of the Coupling Facilities that it
is currently monitoring. The retrieved information will indicate the in-use and total object
counts for the various monitored objects. These counts will be used to calculate a percent full
value.
When structure full monitoring observes that a structure's percent full value is at or above a
percent full threshold in terms of any of the structure objects that it contains, highlighted
message IXC585E, as shown in Figure 7-12, will be issued to the console and to the system
message logs. You can review this message manually and take whatever action is necessary,
such as adjusting the structure size or making changes in the workload that is being sent to
the structure. As an alternative, you can define message automation procedures to diagnose
or relieve the structure full condition.
IXC585E STRUCTURE IXC_DEFAULT1 IN COUPLING FACILITY FACIL01, 235
PHYSICAL STRUCTURE VERSION C0D17D82 29104C88,
IS AT OR ABOVE STRUCTURE FULL MONITORING THRESHOLD OF 80%.
ENTRIES: IN-USE:
22 TOTAL:
67, 32% FULL
ELEMENTS: IN-USE:
43 TOTAL:
51, 84% FULL
Figure 7-12 IXC585E message
For each structure in the CFRM policy, the percent full threshold can be specified by the
installation to be any percent value between 0 and 100. Specifying a threshold value of zero
(0) means that no structure full monitoring will take place. If no threshold value is specified,
then the default value of 80% is used as the full threshold percent value.
When the utilization of all monitored structures falls below the structure full threshold,
message IXC586I, as shown in Figure 7-13, will be issued to the console and to the system
message logs to indicate that the full condition was relieved. Message IXC585E will be
deleted.
IXC586I STRUCTURE IXC_DEFAULT1 IN COUPLING FACILITY FACIL01, 295
PHYSICAL STRUCTURE VERSION C0D17D82 29104C88,
IS NOW BELOW STRUCTURE FULL MONITORING THRESHOLD.
Figure 7-13 IXC586I message
7.6 Managing a Coupling Facility

This section provides an overview of managing the Coupling Facility. This includes:
Adding a Coupling Facility
Removing a Coupling Facility for maintenance
Restoring a Coupling Facility
119
For more information about Coupling Facility CF recovery, refer to z/OS System z Parallel
Sysplex Recovery, which is available at:
http://www.ibm.com/servers/eserver/zseries/zos/integtst/library.html
7.6.1 Adding a Coupling Facility

If this is a new Coupling Facility, verify the following items with your system programmer:
The new CF is defined in a new IODF and the new IODF is activated.
A new CFRM policy with the new CF definitions is created and activated.
The CF links are physically connected.
The reset and image profiles on the HMC are customized for the new CF.
Before a CF can be used, the LPAR that it will run in must be defined and activated.
If the CEC that the CF is running on is deactivated, you need to activate the CEC by
performing a Power-on Reset (POR). Refer to Activating a Coupling Facility partition on
page 124 for more information about n activating a CEC.
You can use the procedure in this section to add one or more CFs.
Verifying CF status and definitions

You can use the following display commands to determine whether the CF is defined in the
active CFRM policy and whether the CF is already active.
For more information about the display command, see 7.3.1, Displaying the logical view of a
Coupling Facility on page 103. By issuing this command, you can display the logical view of
the CF. This command queries the information in your CFRM couple data set, where cfname
is the name of the new CF.
Determine the current active CFRM policy, as shown in Figure 7-14.
D XCF,POL,TYPE=CFRM
TYPE: CFRM
POLNAME:
CFRM02 1
STARTED:
06/12/2007 17:28:40
LAST UPDATED: 06/12/2007 17:22:42
Figure 7-14 Displaying the active CFRM policy name
1 This is the active policy name from the CFRM Couple Data Set.
If the relevant RACF authority has been granted, you can execute the IXCMIAPU utility to list
the CFRM Policies contained within the CFRM Couple Data Set. Sample JCL can be found in
SYS1.SAMPLIB (IXCCFRMP). An example is shown in Figure 7-15 on page 121.
120
//STEP1
EXEC PGM=IXCMIAPU
//SYSPRINT DD
SYSOUT=*
//SYSOUT
DD
SYSOUT=*
//SYSABEND DD
SYSOUT=*
//SYSIN
DD
*
DATA TYPE(CFRM) REPORT(YES)
Figure 7-15 Sample JCL for the CFRM Administrative utility to report on the CFRM policies
After the IXCMIAPU utility has been executed successfully, output that is similar to
Figure 7-16 will be displayed.
DEFINE POLICY NAME(CFRM02 )
/* Defined: 06/12/2007 17:22:42.862596 User: ROBI
/* 55 Structures defined in this policy */
/* 2 Coupling Facilities defined in this policy */
*/
CF NAME(FACIL01) DUMPSPACE(2000) PARTITION(00) CPCID(00)

TYPE(SIMDEV) MFG(IBM) PLANT(EN) SEQUENCE(0000000CFCC1)
CF NAME(FACIL02) DUMPSPACE(2000) PARTITION(00) CPCID(00)
TYPE(SIMDEV) MFG(IBM) PLANT(EN) SEQUENCE(0000000CFCC2)
STRUCTURE NAME(CIC_DFHLOG_001) SIZE(32756)
INITSIZE(16384)
PREFLIST(FACIL02, FACIL01)
STRUCTURE NAME(CIC_DFHSHUNT_001) SIZE(16000)
INITSIZE(9000)
PREFLIST(FACIL01, FACIL02)
...
Figure 7-16 Sample output from the CFRM Administrative utility
To see sample output of the command used to display the logical view of the CF, refer to the
D XCF,CF command output shown in Figure 7-2 on page 104.
If you receive the response shown in Figure 7-17, then it means that your new CF is not
defined in the active CFRM policy.
D XCF,CF,CFNAME=CFT1
NO COUPLING FACILITIES MATCH THE SPECIFIED CRITERIA
Figure 7-17 Display a new CF
Contact your system programmer to define and activate a new CFRM policy that includes the
new CF.
After the CFRM Policy has been updated and activated, issue the command to display the
logical view of the new CF, as shown in Figure 7-18 on page 122.
121
D XCF,CF,CFNM=CFT1
CFNAME: CFT1
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
N/A
N/A
NO SYSTEMS ARE CONNECTED TO THIS COUPLING FACILITY
NO STRUCTURES ARE IN USE BY THIS SYSPLEX IN THIS COUPLING FACILITY
Figure 7-18 Display the logical view of the new CF
Verifying the image profile on the HMC

Before you can activate a CF image, you must verify that the image profile is customized to
load the Coupling Facility Control Code (CFCC). We are assuming that your system
programmer has customized the reset profile for the CEC and that the reset profile has been
assigned for activation.
To view the image profile, change the work area to display the CF image you want to activate.
By following this procedure, you can only view the image profile for the new CF. If you need to
customize this image profile, contact your systems programmer for assistance.
Double-click the CF object in HMC Image Group View; see Figure 7-19.
Figure 7-19 HMC Image Group View
Single-click Change Options, as shown in Figure 7-20 on page 123, to display the list of
profiles.
122
Figure 7-20 HMC image detail panel
Select the image profile you want to view and single-click View to display the image profile.
Figure 7-21 HMC Change Object Options
Ensure that the mode Coupling facility is selected on the General view panel. Single-click
Processor on the left of the screen to display the processor view, as shown in Figure 7-22 on
page 124.
123
Figure 7-22 HMC Image Profile (Processor)
Verify these processor settings with your system programmer. If this is a production CF, you
will normally dedicate one or more processors. It is possible to share one or more processors,
but then you must also assign a processor weight. Unlike z/OS, the CFCC is running in a
continuous loop and if its processor resources are shared with other images, it can cause a
performance degradation on the other images if you do not assign the correct processor
weight value.
Tip: We recommend that you do not enable capping for a CF image.
Single-click Storage, on the left of the screen, to display the storage view as shown in HMC
Image Profile (Storage) in Figure 7-23.
Figure 7-23 HMC Image Profile (Storage)
Verify the storage settings with your system programmer.

Single-click Cancel, on the bottom of the screen, to exit.
Activating a Coupling Facility partition

To activate the CF, use the following procedure:
If the CEC is deactivated, you can activate it by performing a Power-on Reset (POR).
Drag and drop the CEC you want to activate to the activate task in the task area.
Click yes on the confirmation panel to start the activation process. This will result in a
POR.
124
Important: Do not activate an already active CEC if there are multiple images
defined on the CPC. This will reload all the images and they might contain active
operating systems.
If the CEC is already activated but the CF image is deactivated, you can activate the CF
image to load the CFCC, as described here:
Drag and drop the CF image you want to activate to the activate task in the task area.
Click yes on the confirmation panel to start the activation process. This will load the
CFCC code.
As soon as z/OS detects connectivity to the CF, you will see the messages as shown in
Figure 7-24 for all the images in the sysplex that are connected to this CF.
IXL157I PATH 09 IS NOW OPERATIONAL TO CUID: 0309 095
IXL157I PATH 0E IS NOW OPERATIONAL TO CUID: 0309 096
IXL157I PATH 0F IS NOW OPERATIONAL TO CUID: 030F 097
IXL157I PATH 10 IS NOW OPERATIONAL TO CUID: 030F 098
...
PARTITION: 00
CPCID: 00
NAMED FACIL01
PARTITION: 00
CPCID: 00
NAMED FACIL02
Figure 7-24 z/OS connectivity messages to CF
After you receive these messages, your Coupling Facility is ready to be used.
7.6.2 Removing a Coupling Facility

You might need to remove a CF from the sysplex for hardware maintenance or an upgrade.
Depending on which structures are residing in the CF, it is possible to remove a CF
non-disruptively, assuming there is another CF available to move the structures to.
As part of your planning to remove a CF, keep the following considerations in mind:
Perform maintenance for the CF during off-peak periods.
Remove only one CF at a time.
To prevent structures from being created in the CF that you are removing, activate a new
CFRM policy that does not include any reference to the CF you want to remove.
125
Resolve failed persistent and no connector structure conditions before shutting down the
CF.
Ensure all systems in the sysplex that are currently using the structures in the CF you
want to remove have connectivity to the alternate Coupling Facility.
To allow structures to be rebuilt on an alternate CF, ensure that enough capacity, such as
storage and CPU cycles, exists on the alternate CF.
Removing a Coupling Facility when multiple Coupling Facilities exist

To remove a CF when multiple CFs exist in the configuration, use the following procedure.
Determine the status of the CFs

See Figure 7-25 on page 127 to display the physical view of the CFs to determine if there is
enough spare storage capacity available on the target CF to move the structures.
126
D CF
NAMED FACIL01
ALLOCATED SPACE
STRUCTURES:
209920 K 1
DUMP SPACE:
2048 K
TABLE COUNT:
FREE SPACE:
511488 K
FREE DUMP SPACE:
TOTAL SPACE:
723456 K
TOTAL DUMP SPACE:
VOLATILE:
YES
CFLEVEL:
14
BUILT ON 03/26/2007 AT 17:58:00
0
0
2048
2048
0
256
K
K
K
K
K
COUPLING FACILITY SPACE CONFIGURATION

IN USE
FREE
TOTAL
CONTROL SPACE:
211968 K
511488 K
723456 K
NON-CONTROL SPACE:
0 K
0 K
0 K
...
CONTROL UNIT ID: 030F
NAMED FACIL02
ALLOCATED SPACE
STRUCTURES:
122368 K
0 K
DUMP SPACE:
2048 K
TABLE COUNT:
0
FREE SPACE:
599040 K
FREE DUMP SPACE:
2048 K
TOTAL SPACE:
723456 K
TOTAL DUMP SPACE:
2048 K
0 K
VOLATILE:
YES
256 K
CFLEVEL:
14
BUILT ON 03/26/2007 AT 17:58:00
IN USE
CONTROL SPACE:
124416 K
NON-CONTROL SPACE:
0 K
FREE
599040 K 2
0 K
TOTAL
723456 K
0 K
Figure 7-25 Physical view of CF storage capacity
For example, in our case, we want to move all the structures from Coupling Facility FACIL01
to FACIL02.
There is 209920 KB of storage used by structures on 1 FACIL01. There is 599040 KB of
storage available on 2 FACIL02. There is sufficient storage available on FACIL02 to move all
the structures from FACIL01 to FACIL02.
127
You can use the output of the command shown in Figure 7-25 on page 127 to determine
whether there is enough storage available on the alternate CF. If there is more than one
alternate CF available, the sum of the free storage for these CFs must be enough to
accommodate all the structures you want to move.
If you do not have enough free storage available on your CF, you can disable certain
subsystem functions from using the Coupling Facility. Refer to Removing a Coupling Facility
when only one Coupling Facility exists on page 131 for more information.
The command shown in Figure 7-26 displays the logical view of the CF. When you want to
remove a CF, use this command to determine which structures are allocated in the CF.
CFNAME: FACIL02
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
1
D#$#_GBP0(OLD)
D#$#_GBP1(OLD)
I#$#RM
I#$#VSAM
IRRXCF00_B001
ISGLOCK
IXC_DEFAULT_1
I#$#LOCK1
IGWLOCK00
ISTGENERIC
Figure 7-26 Displaying structures allocated to CF
1 This is the list of structures allocated to the CF.
Activating a new CFRM policy (optional)

To be able to move a structure from one CF to an alternate CF, the cfname of the alternate
CF must be specified in the active CFRM policy preference list. This new CFRM policy will
also remove all references to the CF you want to remove. Depending on your installation, you
might need to activate a new CFRM policy to be able to rebuild the structures in an alternate
CF. For more information on your CFRM policies, contact your system programmer.
Issue the command shown in Figure 7-27 to record the information of the current active
CFRM policy name. You will need to activate this CFRM policy again when you restore the CF
you are about to remove from the sysplex.
D XCF,POLICY,TYPE=CFRM
TYPE: CFRM
POLNAME:
CFRM02
STARTED:
06/25/2007 01:34:07
LAST UPDATED: 06/12/2007 17:22:42
Figure 7-27 Displaying information about the active CFRM policy
Issue the command shown in Figure 7-28 on page 129 to activate the new CFRM policy.
128
SETXCF START,POLICY,TYPE=CFRM,POLNAME=new_policy_name
Figure 7-28 Activate a new CFRM Policy
Use a name for the new CFRM policy that is different from the name of the original CFRM
policy so that when CF maintenance is complete, the original policy can be restored.
After the command shown in Figure 7-28 has been issued, you may see the messages
displayed in Figure 7-29.
IXC511I START ADMINISTRATIVE POLICY CFRM03 FOR CFRM ACCEPTED
IXC512I POLICY CHANGE IN PROGRESS FOR CFRM 876
TO MAKE CFRM03 POLICY ACTIVE.
2 POLICY CHANGE(S) PENDING.
Figure 7-29 Policy change pending messages
If you receive the messages shown in Figure 7-29, issue the command in Figure 7-30 to
identify structures that are in a policy change pending state.
Note: When you start a new CFRM policy, the allocated structures that are affected by the
new policy change enter a policy change pending state. Structures that enter a policy
change pending state remain in that state until the structure is deallocated and reallocated
through a rebuild. Structures that reside on CFs that are not being removed might remain
in a policy change pending state until the original policy is restored.
D XCF,STR,STATUS=POLICYCHANGE
STRNAME
ALLOCATION TIME
STATUS
TYPE
IXC_DEFAULT_1
06/26/2007 22:25:22 ALLOCATED
LIST
POLICY CHANGE PENDING - CHANGE
IXC_DEFAULT_2
06/26/2007 22:25:13 ALLOCATED
LIST
POLICY CHANGE PENDING - CHANGE
Figure 7-30 Identify structures in a policy change pending state
Moving structures to the alternate Coupling Facility

The structures as shown in Figure 7-26 on page 128 need to be moved before the CF can be
shut down. Some structures can be moved by issuing the command shown in Figure 7-31.
SETXCF START,REBUILD,{STRNAME=strname|LOC=cfname}
Figure 7-31 Rebuild a structure to another CF
Other structures must be moved by the owning subsystem (for example, the JES2 checkpoint
structure). See 7.8, Managing CF structures on page 147, for information about moving
these structures and deallocating persistent structures.
129
For a list of structures that either support or do not support rebuild, refer to Appendix B, List
of structures on page 499.
To verify that no structures remain in the CF that is being removed, issue the command in
Figure 7-32.
D XCF,CF,CFNM=FACIL02
IXC362I 19.43.13 DISPLAY XCF
CFNAME: FACIL02
...
1 NO STRUCTURES ARE IN USE BY THIS SYSPLEX IN THIS COUPLING FACILITY
Figure 7-32 Displaying information about the CF to be removed
If there are no structures allocated to this 1 CF, you can continue to configure the sender
paths offline and to deactivate the CF.
Configuring sender paths offline

You must repeat this procedure for all the systems in the sysplex that are connected to this
CF. (Figure 7-26 on page 128 shows the command to use to list of all the systems that are
connected to the CF.)
Identify the sender paths for the CF you want to remove by issuing the command shown in
Figure 7-33.
D CF
NAMED FACIL02
...
SENDER PATH
PHYSICAL
LOGICAL
0F
ONLINE
ONLINE
10
ONLINE
ONLINE
...
CHANNEL TYPE
ICP
ICP
Figure 7-33 Displaying CF sender paths
Issue the command shown in Figure 7-34, where nn is one of the sender paths you need to
configure offline.
CONFIG CHP(nn),OFFLINE,UNCOND
Figure 7-34 Configure command for a CF chpid
Note: The UNCOND parameter is only needed if it is the last sender path that is connected
to the CF.
130
CONFIG CHP(10),OFFLINE,UNCOND
IEE712I CONFIG PROCESSING COMPLETE
Figure 7-35 Configure a CF chpid offline
To ensure that the sender paths were taken offline, issue the command in Figure 7-36.
D CF
NAMED FACIL02
...
SENDER PATH
PHYSICAL
LOGICAL
0F
ONLINE
ONLINE
10
1 NOT OPERATIONAL
ONLINE
CHANNEL TYPE
ICP
ICP
NO COUPLING FACILITY SUBCHANNEL STATUS AVAILABLE 2

...
Figure 7-36 Displaying CF Sender paths
The output of the command in Figure 7-36 indicates that the physical status of the sender
path is 1 NOT OPERATIONAL and that there are 2 NO SUBCHANNELS AVAILABLE as a result of the
configure offline command.
Deactivating the Coupling Facility

Important: When the CF is deactivated, any remaining structure data is lost.
To deactivate the CF, drag and drop the CF image you want to deactivate to the deactivate
task in the task area. Click Yes on the confirmation panel to start the deactivation process.
Removing a Coupling Facility when only one Coupling Facility exists

Note: Having a single CF in a production data sharing environment is not recommended.
For example, having a GRS Star environment in a single CF means that when any
maintenance work needs to be carried out on the CF, a sysplex-wide outage is required
because there is no alternative CF to rebuild the ISGLOCK structure to.
It is always recommended to have more than one CF to avoid a single point of failure. Where
only one CF is implemented, the operator must be aware of the implications for each
application that accesses the CF. For example, if a RACF structure is present and the CF is
removed, RACF goes into read only mode and no one is able to update the database or
change a password. If you are removing the only CF from the sysplex, you cannot continue to
run subsystems such as IMS and DB2 that participate in data sharing.
Determining the status of the Coupling Facility

Issue the command shown in Figure 7-38 on page 132 to determine which structures, if any,
remain in the CF that you want to remove.
131
CFNAME: FACIL02
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
Figure 7-37 Identify structures allocated in CF
Stopping the active CFRM policy

Issue the command in Figure 7-38 to stop the CFRM policy.
SETXCF STOP,POLICY,TYPE=CFRM
Figure 7-38 Stopping a CFRM policy
Figure 7-39 displays the output of the command to stop the CFRM policy.
SETXCF STOP,POLICY,TYPE=CFRM
IXC510I STOP POLICY FOR CFRM ACCEPTED
IXC512I POLICY CHANGE IN PROGRESS FOR CFRM
TO MAKE NULL POLICY ACTIVE.
11 POLICY CHANGE(S) PENDING. 1
Figure 7-39 Stopping an active CFRM policy
Note: When you stop the active CFRM policy, allocated structures will enter a 1 policy
change pending state.
The following examples explain how to remove CF structure exploiters.
Note: Each CF structure exploiter may have an explicit way of removing its use of a
particular structure. Therefore, you may need to reference the relevant documentation to
obtain further specific information about the process required.
Removing the JES2 checkpoint structure

Use the JES2 reconfiguration dialog to move the checkpoint structure to a data set. You can
use the command shown in Figure 7-40 to invoke the JES2 reconfiguration dialog.
$TCKPTDEF,RECONFIG=YES
Figure 7-40 JES2 checkpoint reconfig
132
Use the SETXCF FORCE command to delete the persistent JES2 structure after the
reconfiguration is completed successfully.
Removing the system logger structures

If any subsystem is using the system logger, provide a normal shutdown of the
subsystem.
Issue the vary command shown in Figure 7-41 to activate SYSLOG and deactivate
OPERLOG.
V SYSLOG,HARDCPY
V OPERLOG,HARDCPY,OFF
Figure 7-41 Removing use of OPERLOG for syslog
See Chapter 14, Managing consoles in a Parallel Sysplex on page 283 for more
information on managing OPERLOG.
Issue the command shown in Figure 7-42 to stop log stream recording of LOGREC and
revert to the disk based data set, if a disk version exists.
SETLOGRC DATASET
Figure 7-42 LOGREC medium reverted back to disk data set
If a disk-based data set is not available, you can request that recording logrec error and
environmental records be disabled by issuing the command shown in Figure 7-43.
SETLOGRC IGNORE
Figure 7-43 Disable LOGREC recording
Removing the XCF signalling structures

We are assuming that there are CTC connections available to XCF.
Issue the commands shown in Figure 7-44 to stop the XCF signalling through the structures
on all the images in the sysplex.
RO *ALL,SETXCF STOP,PATHIN,STRNAME=strname
RO *ALL,SETXCF STOP,PATHOUT,STRNAME=strname
Figure 7-44 Stopping XCF Pathin/Pathout structures
Disable RACF data sharing

Issue the command shown in Figure 7-45 from either TSO or the console to disable RACF
data sharing. If entering the command in TSO, you will issue it as shown here.
RVARY NODATASHARE
Figure 7-45 Disabling RACF data sharing in TSO
133
If entering the command using the console, you will need to first identify the correct
Command Recognition Character (CRC) in use by the RACF subsystem. Issue the command
in Figure 7-46 to display the various CRCs that are in use.
D OPDATA
PREFIX
OWNER
SYSTEM
SCOPE
...
%
RACF
#@$1
SYSTEM
%
RACF
#@$2
SYSTEM
% 1
RACF
#@$3 2
SYSTEM 3
REMOVE
FAILDSP
NO
NO
NO
PURGE
PURGE
PURGE
Figure 7-46 Display CRCs in use
1 The Command Recognition Character (CRC) defined.

2 The system where this CRC is in use.
3 The scope of the CRC. It will either be a scope of system or sysplex.
After you have identified the CRC to use, issue the command from the console, as shown in
Figure 7-47.
%RVARY NODATASHARE
Figure 7-47 Disabling RACF data sharing with a console command
Note: Regardless of which method you use to disable RACF data sharing mode, you will
be required to enter the RVARY password to authorize this command.
Removing persistent structures

For any persistent structures or failed persistent connectors in the CF, use the SETXCF FORCE
command.
Configuring sender paths offline

Determine whether all the structures have been removed from the CF by using the command
CFNAME: FACIL02
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
Figure 7-48 Display structures allocated to CF
134
D CF,CFNAME=FACIL02
NAMED FACIL02
...
SENDER PATH
PHYSICAL
LOGICAL
10 1
ONLINE
ONLINE
TOTAL:
6
IN USE:
6
NOT USING:
DEVICE
SUBCHANNEL
STATUS
5030
000A
OPERATIONAL
5031
000B
OPERATIONAL
5032
000C
OPERATIONAL
5033
000D
OPERATIONAL
5034
000E
OPERATIONAL
5035
000F
OPERATIONAL
CHANNEL TYPE
ICP
NOT USABLE:
Figure 7-49 Display physical view of CF
Issue the command in Figure 7-50, where nn is the 1 sender path that you need to configure
offline.
CONFIG CHP(nn),OFFLINE
Figure 7-50 Configure the CF Sender Path offline
If all structures are removed from the CF, you can use the UNCOND parameter.
Note: The FORCE and UNCOND parameters are only necessary if it is the last sender
path that is connected to the CF.
CF CHP(10),OFF,FORCE
1 IXL126I CONFIG WILL FORCE OFFLINE LAST CHP(10) TO COUPLING FACILITY FACIL02
06 2 IXL127A REPLY CANCEL OR CONTINUE
R 06,CONTINUE
IEE600I REPLY TO 06 IS;CONTINUE
IEE503I CHP(10),OFFLINE
IEE712I CONFIG
PROCESSING COMPLETE
3 IXC518I SYSTEM #@$3 NOT USING
CONTROL UNIT ID: 030F 2
NAMED FACIL02
REASON: CONNECTIVITY LOST.
REASON FLAG: 13300002.
Figure 7-51 Configuring a CHPID offline
135
You will receive message 1 IXL126I when you specify the FORCE keyword. Reply
CONTINUE on message 2 IXL127A to configure the sender path offline. You will receive
message 3 IXC518I as soon as all the sender paths are configured offline. To ensure that the
sender paths were taken offline, issue the command as shown in Figure 7-52.
D CF,CFNAME=FACIL02
NAMED FACIL02
...
SENDER PATH
PHYSICAL
LOGICAL
10
NOT OPERATIONAL 1
ONLINE
CHANNEL TYPE
ICP
NO COUPLING FACILITY SUBCHANNEL STATUS AVAILABLE 2

Figure 7-52 Physical display of CF
The output of the command in Figure 7-52 is indicating that the physical status of the sender
path is 1 NOT OPERATIONAL and that there is 2 NO SUBCHANNEL AVAILABLE as a result of the
configure offline command.
Deactivating the Coupling Facility

Important: When the CF is deactivated, any remaining structure data is lost.
To deactivate a CF, drag and drop the CF image you want to deactivate to the deactivate task
on the HMC. Click Yes on the confirmation panel to start the deactivation process.
7.6.3 Restoring the Coupling Facility to the sysplex

The following procedure describes how to restore a CF that has been removed from the
sysplex.
Activating the original CFRM policy

Activate the original CFRM policy by issuing the command shown in Figure 7-53.
SETXCF START,POLICY,TYPE=CFRM,POLNAME=polname
Figure 7-53 Starting the new policy to restore a CF
Figure 7-54 shows the messages that are issued when the new CFRM policy is activated.
SETXCF START,POLICY,TYPE=CFRM,POLNAME=CFRM03
IXC513I COMPLETED POLICY CHANGE FOR CFRM.
CFRM03 POLICY IS ACTIVE.
Figure 7-54 CFRM policy messages
136
Activating the Coupling Facility

If the CF is not already added to the sysplex, follow the procedure described in 7.6.1, Adding
a Coupling Facility on page 120.
Moving the structures

After the CF has become active, you should then ensure that the various structures are
located in the correct CF as defined in the current active CFRM policy. This can be achieved
by using either the POPULATECF or REALLOCATE parameters of the SETXCF START
command to relocate the Coupling Facility structures.
For a detailed discussion about REALLOCATE and REBUILD with POPULATECF, refer to
REALLOCATE or REBUILD POPULATECF command on page 154.
Important: The use of the REALLOCATE process and the POPULATECF command are
mutually exclusive.
You can also manually rebuild the structures that have been rebuilt to an alternate CF. Issue
the command shown in Figure 7-55 to rebuild the structures to the original CF.
SETXCF START,REBUILD,STRNAME=strname
Figure 7-55 SETXCF START Rebuild command
The LOC=OTHER parameter may be needed, depending on the CFRM policy structure
preference list.
For more information about the rebuild command, refer to 7.8.1, Rebuilding structures that
support rebuild on page 147.
For more information and a list of structures that do not support rebuild, refer to Appendix B,
List of structures on page 499.
Restart any other subsystems that have been stopped during the CF shutdown procedure.
Note: After the CF and CFRM policy are restored, some functions might reconnect to the
CF automatically, depending on the method used to remove the structures.
7.7 Coupling Facility Control Code (CFCC) commands

The CFCC supports a limited number of operator commands. These commands can only be
issued using the HMC (Operating System Messages) interface. For more information about
these commands, refer to HMC online documentation.
To invoke the operating messages panel on the HMC, drag and drop the CF image object on
the operating system messages task.
7.7.1 CFCC display commands

Note: The HMC used to issue the CFCC Display commands is at HMC Driver Level 64 and
Coupling Facility Control Code (CFCC) level 15.
137
The following commands are used to display CF resource information. These commands can
only be issued to the CF using the Operating System Messages interface of the HMC.
When the CF is initially activated, you see the messages displayed in Figure 7-56.
Figure 7-56 CFCC overview
Display CHPids command

Issue the Display CHPids command to display the channel paths currently configured online
to the CF, as shown in Figure 7-57 on page 139.
138
Figure 7-57 CFCC Display CHIPDs
The output of this command displays the paths currently configured online to the CF.
Display Mode command

Issue the Display Mode command to display the CF volatility mode, as shown in Figure 7-58.
Figure 7-58 CFCC Display Mode
139
The volatility mode of the CF can either be NONVOLATILE or VOLATILE, as explained here:
NONVOLATILE
This specifies that the CF should be run in this mode if there is a

uninterruptable power supply (UPS) available for the processor
complex that the CF is running on. The CF does not monitor the
installation or availability of a UPS, but maintains a nonvolatile status
for the CF.
VOLATILE
This specifies that the CF should be run in this mode regardless of the
actual volatility state of the CF. Coupling Facility storage contents are
lost if a power failure occurs or the CF is turned off. This is the
preferred mode for CF operation without a UPS backup.
CFCC Display level command

Issue the CFCC Display level command to display the CFCC release, service level, build
date, build time and facility operational level, as shown in Figure 7-59.
Figure 7-59 CFCC Display level
Display Rideout command

Issue the Display Rideout command to display the rideout interval (in seconds) for the
POWERSAVE volatility mode, as shown in Figure 7-60 on page 141.
140
Figure 7-60 CFCC Display Rideout
Display Resources command

Issue the Display Resources command to display the number of processors, receiver paths
(CFRs) and storage available to the CF, as shown in Figure 7-61.
Figure 7-61 CFCC Display Resources
Display Timezone command

Issue the Display Timezone command to display the hours east or west of Greenwich Mean
Time (GMT) used to adjust time stamps in messages, as shown in Figure 7-62 on page 142.
141
Figure 7-62 CFCC Display Timezone
Display Dyndisp
Issue the Display Dyndisp command to turn Dynamic Coupling Facility Dispatching on or off
for the CF, as shown in Figure 7-63.
Figure 7-63 CFCC Display Dyndisp
Display CPs
Issue the Display CPs command to display the online and standby central processors
assigned to the CF partition, as shown in Figure 7-64 on page 143.
142
Figure 7-64 CFCC Display CPs
Display Help
Issue the Display Help command to display CF command syntax for the command you enter,
as shown in Figure 7-65.
Figure 7-65 CFCC Display Help
7.7.2 CFCC control commands

This section explains the usage of CFCC control commands.
143
Dynamic CF dispatching
You can enable dynamic CF dispatching for a CF image in order to use it as a backup CF if
the primary CF fails. Issue the command shown in Figure 7-66 to enable or disable dynamic
CF dispatching.
DYNDISP ON|OFF
Figure 7-66 Enabling/disabling dynamic CF dispatching
The message shown in Figure 7-67 will be displayed if you attempt to enable dynamic CF
dispatching on dedicated CF processors.
DYNDISP ON
CF0505I DYNDISP command cancelled
Command has no effect with dedicated CP's
Figure 7-67 Message when enabling dynamic CF dispatching on dedicated CF processors
With dynamic CF dispatching enabled, you see the following:

It uses minimal processor resources despite its assigned processing weight.
Its unused processor resources are shared with other active logical partitions until it is
needed as a backup CF.
It automatically becomes a backup CF if the primary CF fails.
It uses its full share of processor weight only while it is in use as a backup CF.
Change mode of CF operation

This defines the volatility mode to be used for CF operation.
POWERSAVE
This specifies that the CF runs in POWERSAVE mode and that CF storage contents are
nonvolatile if the battery backup feature is installed and its battery is online and charged.
Issue the command shown in Figure 7-68 to enable POWERSAVE.
MODE POWERSAVE
CF0102I MODE is POWER SAVE. Current status is NONVOLATILE.
Power-Save resources are available.
Figure 7-68 Enabling POWERSAVE mode
If the volatility state of the CF remains nonvolatile, running in POWERSAVE mode assures
the following:
If a utility power failure occurs and utility power does not return before a rideout interval
completes, the CF will cease operation and save its storage contents across the utility
power failure. When power is restored in the CF, CF storage contents will be intact and do
not have to be rebuilt.
If a utility power failure occurs and utility power returns before a rideout interval completes,
CF operation continues. You specify the length of the rideout interval using the RIDEOUT
command. POWERSAVE is the default volatility mode.
144
VOLATILE
This specifies that the CF runs in volatile mode regardless of the actual volatility state of the
CF. CF storage contents are lost if a power failure occurs or if CF power is turned off. This is
the preferred mode for CF operation without an UPS backup or internal battery feature (IBF).
Issue the command shown in Figure 7-69 to change the CF to volatile mode.
MODE VOLATILE
CF0100I MODE is VOLATILE
Figure 7-69 Changing mode to VOLATILE
NONVOLATILE
This specifies that the CF runs in nonvolatile mode. This should be used if a UPS is available
for the processor complex that the CF is running on. The CF does not monitor the installation
or availability of a UPS, but maintains a nonvolatile status for the CF. Issue the command
shown in Figure 7-70 to change the CF to nonvolatile mode.
MODE NONVOLATILE
CF0100I MODE is NONVOLATILE
Figure 7-70 Changing mode to NONVOLATILE
Change the rideout time interval

This defines the rideout interval for a CF operating in POWERSAVE mode. The rideout
interval is the consecutive amount of time that utility power must be off before the CF begins
to shut down. When the rideout interval completes, the CF shuts down and battery power is
diverted to preserve CF storage contents in CF storage until power is restored. The default
interval is 10 seconds. Issue the command shown in Figure 7-71 to change the rideout
interval.
RIDEOUT 20
CF0162I Rideout is set to 20 seconds.
Figure 7-71 Changing the rideout time interval
Shutdown CFCC
This ends CF operation and puts all CF logical central processors (CPs) into a disabled wait
state. Issue the command shown in Figure 7-72 to shut down the CF.
SHUTDOWN
CF0082A If SHUTDOWN is confirmed, shared data will be lost;
CF0090A Do you really want to shut down the Coupling Facility? (YES/NO)
YES
Figure 7-72 Shutdown command
Attention: By responding YES to the prompt in Figure 7-72, any shared data that remains
in the CF will be lost.
Enter YES to confirm the shutdown.
145
7.7.3 CFCC Help commands

The following help commands are available.
General Help
Issue the command shown in Figure 7-73 to obtain a list of available CFCC commands.
Figure 7-73 General CFCC Help command
Specific Help
Issue the command shown in Figure 7-74 to obtain help regarding a specific CFCC
command.
HELP command
Figure 7-74 Specific HELP for a CFCC command
As an example, the output shown in Figure 7-75 displays help information about the
CONFIGURE command.
HELP CONFIGURE
CF0403I Configure command formats:
CONfigure xx ONline
xx OFFline
Where xx is a hex CHPID number.
Example:
configure 11 offline
Figure 7-75 Requesting specific HELP for the CONFIGURE command
146
7.8 Managing CF structures

CF structures may be moved from one CF to another if there is enough storage available and
if defined on the preference list for that structure. Also, rebuilding or reinitializing the structure
in the same CF may be required if a connection has been lost or the owner of that structure
needs to restart it.
All structures support automatic rebuild, with these exceptions:
DB2 Cache structure for Group Buffer Pools
CICS Temporary Storage Queue Pool structure
7.8.1 Rebuilding structures that support rebuild

This section uses the example of a typical structure ISGLOCK, which is a LOCK type
structure used for Global Resource Serialization (GRS) in a sysplex configuration.
In normal processing, a structure may be rebuilt and messages similar to those in Figure 7-76
may be seen. This would occur, for example, if access to a CF is removed and an automatic
rebuild of the structure is to be attempted
.
*IXL158I PATH 0F IS NOW NOT-OPERATIONAL TO CUID: 030F 502

*IXL158I PATH 10 IS NOW NOT-OPERATIONAL TO CUID: 030F 503
. . .
IXC518I SYSTEM #@$3 NOT USING 504
PARTITION: 00
CPCID: 00
NAMED FACIL02
REASON: CONNECTIVITY LOST.
REASON FLAG: 13300002.
. . .
IXC521I REBUILD FOR STRUCTURE ISGLOCK 421
HAS BEEN STARTED
. . .
IXC526I STRUCTURE ISGLOCK IS REBUILDING FROM 530
COUPLING FACILITY FACIL02 TO COUPLING FACILITY FACIL01.
REBUILD START REASON: CONNECTIVITY LOST TO STRUCTURE
INFO108: 00000003 00000000.
. . .
ALLOCATION SIZE IS WITHIN CFRM POLICY DEFINITIONS
. . .
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE ISGLOCK 532
WAS SUCCESSFUL. JOBNAME: GRS ASID: 0007
CONNECTOR NAME: ISGLOCK##@$3 CFNAME: FACIL01
Figure 7-76 ISGLOCK structure rebuild after CF connectivity lost
If required, the structure rebuild process can be initiated using operator commands including
the command shown in Figure 7-77 on page 148.
147
SETXCF START,REBUILD,STRNAME=strname.
Figure 7-77 SETXCF START REBUILD command for a structure
There are several reasons why a structure may have to be rebuilt, including performance,
structure size changes, or maintenance.
Rebuilding Coupling Facility structures in either Coupling Facility

Issue the D XCF,STR,STRNM=strname command to check the status of the structure and which
CF it is attached to; see Figure 7-78.
D XCF,STR,STRNM=ISGLOCK
STRNAME: ISGLOCK
STATUS: ALLOCATED
TYPE: LOCK
POLICY INFORMATION:
POLICY SIZE
: 8704 K
FULLTHRESHOLD : 80
ALLOWAUTOALT
: NO
REBUILD PERCENT: 1
DUPLEX
: DISABLED
ENFORCEORDER
: NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/28/2007 00:53:21
CFNAME
: FACIL02
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 8448 K
LOCKS:
TOTAL:
1048576
PHYSICAL VERSION: C0D006D0 BD410282
LOGICAL VERSION: C0D006D0 BD410282
XCF GRPNAME
: IXCLO007
DISPOSITION
: DELETE
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3
CONNECTION NAME
---------------ISGLOCK##@$1
ISGLOCK##@$2
ISGLOCK##@$3
ID
-03
01
02
VERSION
-------00030069
00010090
00020063
SYSNAME
-------#@$1
#@$2
#@$3
JOBNAME
-------GRS
GRS
GRS
ASID
---0007
0007
0007
Figure 7-78 Displaying a structure allocated and its connections
148
STATE
-------ACTIVE
ACTIVE
ACTIVE
Issue the D XCF,CF,CFNAME=cfname command to determine which structures are allocated in

the CF; see Figure 7-79.
CFNAME: FACIL02
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
D#$#_GBP0(NEW)
I#$#OSAM
IGWLOCK00
ISTGENERIC
D#$#_GBP1(NEW)
I#$#RM
IRRXCF00_B001
IXC_DEFAULT_1
I#$#LOCK1
I#$#VSAM
ISGLOCK
Figure 7-79 Display the structures in a CF
Refer to the output in Figure 7-80 and determine if any of the structures that you want to
rebuild has a number of connections of zero (0) (no-connector condition exists), or if the
number of connections is not zero (0) and all connectors are failed persistent.
D XCF,STR,STRNM=SYSTEM_OPERLOG
. . .
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 07/01/2007 19:33:55
CFNAME
: FACIL01
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 9216 K
ENTRIES: IN-USE:
704 TOTAL:
6118,
ELEMENTS: IN-USE:
1780 TOTAL:
12341,
PHYSICAL VERSION: C0D4C6E1 1ADD1885
LOGICAL VERSION: C0D4C6E1 1ADD1885
DISPOSITION
: DELETE
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3 1
. . .
11% FULL
14% FULL
Figure 7-80 Display connections to a structure
1 There are currently three connections to this structure.
149
The command shown in Figure 7-81 illustrates how to display a structure with a
failed-persistent connection.
D XCF,STR,STRNM=IGWLOCK00
STRNAME: IGWLOCK00
STATUS: ALLOCATED
POLICY SIZE
: 160000 K
REBUILD PERCENT: 1
DUPLEX
: DISABLED
...
# CONNECTIONS : 1
---------------- -- -------- -------- -------- ---- ---------------ZZZZZZZZ#@$3
01 00010091 #@$3
SMSVSAM 000A FAILED-PERSISTENT
Figure 7-81 Displaying a structure with a failed-persistent connection
Enter SETXCF START,REBUILD,STRNM=strname to rebuild the structure into the CF that is first
displayed on the PREFERENCE LIST, as defined in the CFRM policy.
An example of the messages from the REBUILD is shown in Figure 7-82 on page 151.
150
SETXCF START,REBUILD,STRNAME=IGWLOCK00
IXC521I REBUILD FOR STRUCTURE IGWLOCK00 209
HAS BEEN STARTED
IXC367I THE SETXCF START REBUILD REQUEST FOR STRUCTURE 210
IGWLOCK00 WAS ACCEPTED.
IGW457I DFSMS REBUILD PROCESSING HAS BEEN 181
INVOKED FOR LOCK STRUCTURE IGWLOCK00
PROCESSING EVENT: REBUILD QUIESCE
. . .
IXC526I STRUCTURE IGWLOCK00 IS REBUILDING FROM 183
REBUILD START REASON: OPERATOR INITIATED
INFO108: 00000003 00000003.
IXC582I STRUCTURE IGWLOCK00 ALLOCATED BY COUNTS. 184
PHYSICAL STRUCTURE VERSION: C0D4E45C 2FCFBC4E
STRUCTURE TYPE:
LOCK
CFNAME:
FACIL01
ALLOCATION SIZE:
14336 K
POLICY SIZE:
20480 K
POLICY INITSIZE:
14336 K
POLICY MINSIZE:
0 K
IXLCONN STRSIZE:
0 K
ENTRY COUNT:
36507
LOCKS:
2097152
ALLOCATION SIZE IS WITHIN CFRM POLICY DEFINITIONS
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE IGWLOCK00 185
WAS SUCCESSFUL. JOBNAME: SMSVSAM ASID: 000A
CONNECTOR NAME: ZZZZZZZZ#@$2 CFNAME: FACIL01
IXL015I REBUILD NEW STRUCTURE ALLOCATION INFORMATION FOR 186
STRUCTURE IGWLOCK00, CONNECTOR NAME ZZZZZZZZ#@$2
CFNAME
---------------------------------------FACIL02
RESTRICTED BY REBUILD OTHER
FACIL01
STRUCTURE ALLOCATED AC001800
...
PROCESSING EVENT: REBUILD PROCESS COMPLETE
IXC579I PENDING DEALLOCATION FOR STRUCTURE IGWLOCK00 IN 463
PARTITION: 00
CPCID: 00
HAS BEEN COMPLETED.
PHYSICAL STRUCTURE VERSION: C0D006C8 BE46ECC5
INFO116: 13088068 01 6A00 00000003
TRACE THREAD: 00002B57.
PROCESSING EVENT: REBUILD PROCESS COMPLETE
Figure 7-82 Messages from REBUILD command
Note: The structure was moved from FACIL02 to FACIL01 in this example, because
FACIL01 was the first system in the preference list.
151
Rebuilding the structure in another Coupling Facility

Issue the SETXCF START,REBUILD,STRNM=strname,LOC=OTHER command to rebuild the structure
into any CF other than the one to which it is currently attached; see Figure 7-83.
SETXCF START,REBUILD,STRNAME=IGWLOCK00,LOC=OTHER
HAS BEEN STARTED
IXC367I THE SETXCF START REBUILD REQUEST FOR STRUCTURE 210
PROCESSING EVENT: REBUILD QUIESCE
...
IXC526I STRUCTURE IGWLOCK00 IS REBUILDING FROM 183
...
Figure 7-83 REBUILD command with LOC=OTHER specified
Issue the D XCF,STR,STRNM=strname command, as shown in Figure 7-84 on page 153. Check
whether the structure is active in another Coupling Facility.
152
STRNAME: IGWLOCK00
STATUS: ALLOCATED
TYPE: LOCK
POLICY INFORMATION:
POLICY SIZE
: 20480 K
FULLTHRESHOLD : 80
ALLOWAUTOALT
: NO
DUPLEX
: DISABLED
ENFORCEORDER
: NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 07/01/2007 22:02:55
CFNAME
: FACIL01 2
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 14336 K
ENTRIES: IN-USE:
0 TOTAL:
36507,
LOCKS:
TOTAL:
2097152
PHYSICAL VERSION: C0D4E82E BDEF7B05
LOGICAL VERSION: C0D4E82E BDEF7B05
XCF GRPNAME
: IXCLO000
DISPOSITION
: KEEP
ACCESS TIME
: 0
NUMBER OF RECORD DATA LISTS PER CONNECTION: 16
MAX CONNECTIONS: 4
# CONNECTIONS : 3
CONNECTION NAME
---------------ZZZZZZZZ#@$1
ZZZZZZZZ#@$2
ZZZZZZZZ#@$3
ID
-03
02
01
VERSION
-------00030077
0002006C
00010091
SYSNAME
-------#@$1
#@$2
#@$3
JOBNAME
-------SMSVSAM
SMSVSAM
SMSVSAM
0% FULL
ASID
---000A
000A
000A
STATE
---------ACTIVE
ACTIVE
ACTIVE
STRNUM: 00000000 STRSEQ: 00000000


Figure 7-84 Displaying the location of a structure
The preference list is 1 FACIL02, FACIL01, and the CFNAME is 2 FACIL01.
153
REALLOCATE or REBUILD POPULATECF command

There are two supported commands for rebuilding structures from one CF to another.
Tip: We recommend using the REALLOCATE command. The REALLOCATE command is
integrated into z/OS 1.8, and via APAR OA08688 for z/OS 1.4 and above.
The POPULATECF function and the REALLOCATE process are mutually exclusive.
Note: The REALLOCATE process will not be started when XCF discovers an active
system in the sysplex without the prerequisite z/OS operating system support.
The REALLOCATE process provides a simple, broad-based structure placement optimization
via an MVS command. It simplifies many CF maintenance procedures.
The most significant advantage of the REALLOCATE process applies to environments that
have any of these conditions:
More than two CFs
Duplexed structures, such as DB2 Group Buffer Pools
Installations wanting structures to always reside in specific CFs whenever possible
Configurations with CFs having different characteristics, such as different CF levels or
processor speeds
The REALLOCATE process will:
Clear all CFRM policy change pending conditions.
Move all simplex structures into their most preferred CF location.
Move all duplexed structure instances into their two most preferred CF locations, in the
correct order. It automatically corrects reversal of the primary and secondary structure
locations
Act on one structure at a time to minimize any disruption caused by reallocation actions.
Issue a message describing the evaluation process for each allocated structure.
Issue a summary message upon completion of all structures and summarizing actions
taken.
Simplify CF structure movement during disruptive CF maintenance or upgrades.
When the REALLOCATE process does not select an allocated structure, message IXC544I is
issued with an explanation. See Figure 7-85 on page 155 for an example of the message.
154
...
IXC544I REALLOCATE PROCESSING FOR STRUCTURE CIC_DFHSHUNT_001 883
WAS NOT ATTEMPTED BECAUSE
STRUCTURE IS ALLOCATED IN PREFERRED CF
IXC574I EVALUATION INFORMATION FOR REALLOCATE PROCESSING 884
OF STRUCTURE DFHXQLS_#@$STOR1
SIMPLEX STRUCTURE ALLOCATED IN COUPLING FACILITY: FACIL01
ACTIVE POLICY INFORMATION USED.
CFNAME
STATUS/FAILURE REASON
---------------------------FACIL01
PREFERRED CF 1
INFO110: 00000003 AC007800 0000000E
FACIL02
PREFERRED CF ALREADY SELECTED
INFO110: 00000003 AC007800 0000000E
Figure 7-85 IXC544I message when CF structure is not selected for reallocation
When the entire REALLOCATE process completes for all structures, the processing issues
message IXC545I and a report summarizing the actions that were taken as a whole. See
Figure 7-86 for an example of the messages issued.
...
IXC545I REALLOCATE PROCESSING RESULTED IN THE FOLLOWING: 904
0 STRUCTURE(S) REALLOCATED - SIMPLEX
2 STRUCTURE(S) REALLOCATED - DUPLEXED
0 STRUCTURE(S) POLICY CHANGE MADE - SIMPLEX
0 STRUCTURE(S) POLICY CHANGE MADE - DUPLEXED
28 STRUCTURE(S) ALREADY ALLOCATED IN PREFERRED CF - SIMPLEX
0 STRUCTURE(S) ALREADY ALLOCATED IN PREFERRED CF - DUPLEXED
0 STRUCTURE(S) NOT PROCESSED
25 STRUCTURE(S) NOT ALLOCATED
145 STRUCTURE(S) NOT DEFINED
-------200 TOTAL
0 ERROR(S) ENCOUNTERED DURING PROCESSING
IXC543I THE REQUESTED START,REALLOCATE WAS COMPLETED. 905
Figure 7-86 IXC545I message issued after completion of the REALLOCATE command
Consider the following when you use the SETXCF START,REALLOCATE command:
Move structures out of a Coupling Facility following a CFRM policy change that deletes or
changes that Coupling Facility (for example, following a Coupling Facility upgrade or add).
Move structures back into a Coupling Facility following a CFRM policy change that adds or
restores the Coupling Facility (for example, following a Coupling Facility upgrade or add).
Clean up pending CFRM policy changes that may have accumulated for whatever reason,
even in the absence of any need for structure relocation.
Clean up simplex or duplexed structures that were allocated in or moved into the wrong
Coupling Facilities (for example, if the right Coupling Facility was not accessible at the
time of allocation).
Clean up duplexed structures that have primary and secondary reversed because of a
prior condition which resulted in having duplexing stopped.
155
You can also use the REBUILD POPULATECF command to move structures between CFs.
SETXCF START,REBUILD,POPULATECF=cfname.
Figure 7-87 POPULATECF command
This rebuilds all structures defined in the current CFRM policy that are not in their preferred
CF. Sample output is shown in Figure 7-88 after the REBUILD POPULATECF command was
issued.
SETXCF START,REBUILD,POPCF=FACIL02
HAS BEEN STARTED
IXC540I POPULATECF REBUILD FOR FACIL02 REQUEST ACCEPTED. 460
THE FOLLOWING STRUCTURES ARE PENDING REBUILD:
IGWLOCK00
ISGLOCK
IXC_DEFAULT_1
ISTGENERIC
I#$#RM
I#$#LOCK1
I#$#VSAM
I#$#OSAM
IRRXCF00_B001
Figure 7-88 Messages from REBUILD POPULATECF
The method of emptying a CF using the SETXCF START,REBUILD command has some
disadvantages:
All rebuilds are started at the same time, resulting in contention for the CFRM couple
dataset. This contention elongates the rebuild process for all affected structures, thus
making the rebuilds more disruptive to ongoing work
The IXC* (XCF signalling structures) do not participate in that process, but must instead
be separately rebuilt via manual commands on a structure-by-structure basis.
A duplexed structure cannot be rebuilt out of the target CF, so a separate step is needed
to explicitly unduplex it so that it can be removed from the target CF.
Figure 7-89 on page 157 illustrates the disadvantages of using the SETXCF START,REBUILD
command. In the figure, our CF named 1 FACIL02 has 2 DB2 Group Buffer Pool duplexed
structures and an 3 IXC* XCF signalling structure located in it.
156
CFNAME: FACIL02 1
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
CIC_DFHLOG_001
I#$#LOCK1
I#$#VSAM
ISGLOCK
D#$#_GBP0(OLD) 2
I#$#OSAM
IGWLOCK00
ISTGENERIC
D#$#_GBP1(OLD)
I#$#RM
IRRXCF00_B001
IXC_DEFAULT_1 3
Figure 7-89 Structures in CF FACIL02
We now issue the SETXCF START,REBUILD,CFNAME=FACIL02,LOC=OTHER command and the

majority of the structures will be rebuilt to the other CF FACIL01. After the REBUILD has
completed, we issue the D XCF,CF,CFNAME=FACIL02 command to display if there are any
structures remaining in CF FACIL02; see Figure 7-90.
CFNAME: FACIL02
COUPLING FACILITY
:
PARTITION: 00
CPCID: 00
SITE
:
N/A
2000 K
2048 K
256 K
CONNECTED SYSTEMS:
#@$1
#@$2
#@$3
STRUCTURES:
D#$#_GBP0(OLD) 1
D#$#_GBP1(OLD) 2
IXC_DEFAULT_1 3
Figure 7-90 Remaining structures after a REBUILD
1 and 2 DB2 Group Buffer Pool duplexed structure

3 XCF signalling structure
Figure 7-90 illustrates that the DB2 Group Buffer Pool structures and the IXC* XCF signalling
structure remain and will need to be rebuilt using manual commands.
Maintenance mode
z/OS V1.9 includes support for placing Coupling Facilities into a new state called
maintenance mode. When a CF is in maintenance mode, it is logically ineligible for CF
157
structure allocation purposes, as if it had been removed from the CFRM Policy entirely
(although no CFRM Policy updates are required to accomplish this).
Subsequent rebuild or REALLOCATE processing will also tend to remove any CF structure
instances that were already allocated in that CF at the time it was placed into maintenance
mode.
In conjunction with the REALLOCATE command, the new maintenance mode support can
greatly simplify operational procedures related to taking a CF down for maintenance or
upgrade in a Parallel Sysplex. In particular, now the need is avoided to laboriously update or
maintain several alternate copies of the CFRM Policy that omit a particular CF to be removed
for maintenance.
Here we illustrate the maintenance mode command. In Figure 7-91, a display of the
ISTGENERIC structure shows that it is currently allocated in 1 CF2 and the CFRM Policy has
a preference list of 2 CF2 and then CF1.
D XCF,STR,STRNAME=ISTGENERIC
STRNAME: ISTGENERIC
STATUS: ALLOCATED
...
PREFERENCE LIST: CF2
CF1 2
ENFORCEORDER : NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/08/2007 15:39:47
CFNAME
: CF2 1
COUPLING FACILITY: 002094.IBM.02.00000002991E
PARTITION: 1D
CPCID: 00
ACTUAL SIZE
: 16384 K
...
Figure 7-91 Display of structure prior to invoking maintenance mode
In Figure 7-92, the SETXCF command is issued to rebuild ISTGENERIC from CF2 to CF1.
SETXCF START,REBUILD,STRNAME=ISTGENERIC,LOC=OTHER
IXC521I REBUILD FOR STRUCTURE ISTGENERIC
HAS BEEN STARTED
IXC367I THE SETXCF START REBUILD REQUEST FOR STRUCTURE
ISTGENERIC WAS ACCEPTED.
IXC526I STRUCTURE ISTGENERIC IS REBUILDING FROM
COUPLING FACILITY CF2 TO COUPLING FACILITY CF1.
INFO108: 00000028 00000028.
HAS BEEN COMPLETED
Figure 7-92 Rebuild ISTGENERIC structure to CF1
On completion of the rebuild, a display of the ISTGENERIC structure shows that it has been
reallocated into 1 CF1, as shown in Figure 7-93 on page 159.
158
STRNAME: ISTGENERIC
STATUS: ALLOCATED
...
PREFERENCE LIST: CF2
CF1
...
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 07/18/2007 20:26:24
CFNAME
: CF1 1
COUPLING FACILITY: 002094.IBM.02.00000002991E
PARTITION: 0F CPCID: 00
. . .
Figure 7-93 Structure after being reallocated to CF1
Issuing the SETXCF command as shown in Figure 7-94, the CF named CF2 will be placed into
maintenance mode.
SETXCF START,MAINTMODE,CFNAME=CF2
IXC369I THE SETXCF START MAINTMODE REQUEST FOR COUPLING FACILITY
CF2 WAS SUCCESSFUL.
Figure 7-94 Invoking maintenance mode for CF2
On completion of the maintenance mode command, a display of CF2, as shown in

Figure 7-95, will display that it is now in 1 maintenance mode and no allocations are
permitted. Note that there are still 2 structures located in CF2, even though the CF is in
maintenance mode.
D XCF,CF,CFNAME=CF2
CFNAME: CF2
COUPLING FACILITY
:
002094.IBM.02.00000002991E
PARTITION: 1D CPCID: 00
SITE
:
N/A
2048 K
2048 K
512 K
ALLOCATION NOT PERMITTED
MAINTENANCE MODE
CONNECTED SYSTEMS:
SC63
SC64
SC65
STRUCTURES: 2
IXC_DEFAULT_2
SYSTEM_LOGREC(NEW)
SC70
IXC_DEFAULT_4
SYSTEM_OPERLOG(OLD)
SYSARC_PLEX0_RCL
SYSZWLM_991E2094
Figure 7-95 Display of CF after maintenance mode command issued
In Figure 7-96 on page 160, attempting to rebuild a structure back to a CF while the CF is still
in maintenance mode will not succeed and error messages are issued.
159
SETXCF START,REBUILD,STRNAME=ISTGENERIC,LOC=OTHER
HAS BEEN STARTED
ISTGENERIC WAS ACCEPTED.
IXC522I REBUILD FOR STRUCTURE
ISTGENERIC IS BEING STOPPED
TO FALL BACK TO THE OLD STRUCTURE DUE TO
NO COUPLING FACILITY PROVIDING BETTER OR EQUIVALENT CONNECTIVITY
HAS BEEN STOPPED
Figure 7-96 Attempting to allocate structure into CF while still in maintenance mode
With the CF in maintenance mode and structures still located in the CF, issue the SETXCF
START,REALLOCATE command to relocate these structures into an alternative CF, as shown in
Figure 7-97.
SETXCF START,REALLOCATE
IXC543I THE REQUESTED START,REALLOCATE WAS ACCEPTED.
. . .
IXC521I REBUILD FOR STRUCTURE IXC_DEFAULT_2
HAS BEEN STARTED
IXC526I STRUCTURE IXC_DEFAULT_2 IS REBUILDING FROM
COUPLING FACILITY CF2 TO COUPLING FACILITY CF1.
INFO108: 00000028 00000028.
IXC521I REBUILD FOR STRUCTURE IXC_DEFAULT_2
HAS BEEN COMPLETED
. . .
Figure 7-97 REALLOCATE while in maintenance mode
Figure 7-98 shows that after the REALLOCATE command is completed, the CF will have no 1
structures located in it and still be in 2 maintenance mode.
D XCF,CF,CFNAME=CF2
CFNAME: CF2
COUPLING FACILITY
:
002094.IBM.02.00000002991E
PARTITION: 1D CPCID: 00
SITE
:
N/A
2048 K
2048 K
512 K
ALLOCATION NOT PERMITTED
MAINTENANCE MODE 2
CONNECTED SYSTEMS:
SC63
SC64
SC65
SC70
1 NO STRUCTURES ARE IN USE BY THIS SYSPLEX IN THIS COUPLING FACILITY

Figure 7-98 Structures reallocated to alternate CF while in maintenance mode
160
To remove maintenance mode from the CF, issue the SETXCF command as shown in
Figure 7-99.
SETXCF STOP,MAINTMODE,CFNAME=CF2
IXC369I THE SETXCF STOP MAINTMODE REQUEST FOR COUPLING FACILITY
CF2 WAS SUCCESSFUL.
Figure 7-99 Turn off maintenance mode
7.8.2 Stopping structure rebuild

The rebuild process can be stopped by using the SETXCF STOP,REBUILD command or
internally by the application; see Figure 7-100.
SETXCF STOP,REBUILD,STRNM=IGWLOCK00
IXC522I REBUILD FOR STRUCTURE IGWLOCK00 IS BEING STOPPED DUE TO 718
REQUEST FROM AN OPERATOR
IXC367I THE SETXCF STOP REBUILD REQUEST FOR STRUCTURE 719
IXL014I IXLCONN REQUEST FOR STRUCTURE IGWLOCK00 WAS SUCCESSFUL. 720
JOBNAME: SMSVSAM ASID: 000A CONNECTOR NAME: ZZZZZZZZ#@$3
CFNAME: FACIL02
IXC521I REBUILD FOR STRUCTURE IGWLOCK00 HAS BEEN STOPPED
Figure 7-100 Stopping structure rebuild
7.8.3 Structure rebuild failure

There are occasions when an attempt to rebuild a structure fails. Structures that do not
support rebuild are included in this category. Refer to Appendix B, List of structures on
page 499 for a complete list of structures and their capability to rebuild.
If rebuild is not supported or fails, the connectors disconnect abnormally, indicating a
recovery bind to the structure for which connectivity was lost. The connectors enter the failed
persistent state.
Issue D XCF,STR,STRNM=strname; see Figure 7-101 on page 162.
161
STRNAME: IGWLOCK00
STATUS: ALLOCATED
TYPE: LOCK
POLICY INFORMATION:
POLICY SIZE
: 20480 K
FULLTHRESHOLD : 80
ALLOWAUTOALT : NO
DUPLEX
: DISABLED
ENFORCEORDER : NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 07/02/2007 18:49:15
CFNAME
: FACIL01
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 14336 K
ENTRIES: IN-USE:
0 TOTAL:
36507,
LOCKS:
TOTAL:
2097152
PHYSICAL VERSION: C0D5FEC2 D6736F02
LOGICAL VERSION: C0D5FEC2 D6736F02
XCF GRPNAME
: IXCLO000
DISPOSITION
: KEEP
ACCESS TIME
: 0
NUMBER OF RECORD DATA LISTS PER CONNECTION: 16
MAX CONNECTIONS: 4
# CONNECTIONS : 3
0% FULL

---------------- -- -------- -------- -------- ---- -------ZZZZZZZZ#@$1
03 00030077 #@$1
SMSVSAM 000A FAILED-PERSISTENT
STRNUM: 00000000 STRSEQ: 00000000


Figure 7-101 Displaying a structure with a failed persistent connection
Issue SETXCF FORCE,CON,CONNM=conname,STRNM=strname, as shown in Figure 7-102 on

page 163.
162
SETXCF FORCE,CON,CONNM=ZZZZZZZZ#@$1,STRNAME=IGWLOCK00
IXC354I THE SETXCF FORCE REQUEST FOR CONNECTION 638
ZZZZZZZZ#@$1 IN STRUCTURE IGWLOCK00 WAS REJECTED:
FORCE CONNECTION NOT ALLOWED FOR PERSISTENT LOCK OR SERIALIZED LIST
Figure 7-102 Removing a connection
The message in Figure 7-102 was received because the IGWLOCK00 is a LOCK type
structure, so removing a connection to it could result in undetected data loss.
7.8.4 Deleting persistent structures

After all failed persistent connectors are forced or the structure is forced, the structure may
enter an in transition state and be ready for cleanup.
IXC360I STRUCTURE IN TRANSITION
----------------------REASON IN TRANSITION: CONNECT OR DISCONNECT IN PROGRESS
CFNAME
: CFT1
NO SYSTEMS CONNECTED TO COUPLING FACILITY
ALLOCATION TIME: 05/30/1997 15:43:52
COUPLING FACILITY: 009672.IBM.02.000000040104
ACTUAL SIZE
: N/A
VERSION
: AEBCBB20 64274704
Figure 7-103 Structure in transition state
When the structure enters into this state, it means that the structure is a target for deletion.
Depending on the application that owns the structure, you may need to restart the application
for the structure to become allocated in the alternate CF. If this situation has been created by
a CF failure, then when the failed CF is eventually restored, XES resolves this condition and
this information is no longer displayed.
Another way to remove this condition is by removing the failed CF from the active CFRM
policy. IPLing the z/OS images does not clear this condition. The number of connectors in
message IXC360I (as shown in Displaying a Structure with FAILED-PERSISTENT
Connections in Figure 7-81 on page 150) must be zero (0) before proceeding. If ACTIVE
connectors exist, invoke recovery procedures for the connector, or CANCEL the connector's
address space to make the connector disconnect from a structure. Issue SETXCF FORCE to
delete the structure.
SETXCF FORCE,STR,STRNM=IGWLOCK00
IXC353I THE SETXCF FORCE REQUEST FOR STRUCTURE IGWLOCK00 WAS ACCEPTED:
REQUEST WILL BE PROCESSED ASYNCHRONOUSLY
Figure 7-104 Forcing a structure
Note: If ACTIVE connectors exist, a message similar to the one shown in Figure 7-105 will
be received.
163
IXC353I THE SETXCF FORCE REQUEST FOR STRUCTURE 677

IGWLOCK00 WAS REJECTED:
STRUCTURE CONTAINS ACTIVE CONNECTIONS
Figure 7-105 Active connections to a structure
If the structure remains, check the owner of the application and inform the appropriate
support personnel.
164
Chapter 8.
Couple Data Set management

This chapter explains how to manage a Couple Data Set (CDS). It also introduces the
commands required to manage Couple Data Sets in a Parallel Sysplex.
This chapter discusses:
What Couple Data Sets are
Couple Data Set configuration
Commands needed to manage Couple Data Sets
165
8.1 Introduction to Couple Data Set management

Couple Data Sets (CDSs) are data sets that contain status and policy information for the
sysplex. They provide a way for the systems in the sysplex to share this information so that
they can manage the sysplex environment cooperatively.
There are seven different types of Couple Data Sets that could be used in a sysplex. Each
type is associated with a different system component, such as WLM, System Logger, or XCF.
These components use the Couple Data Sets as a repository of information. For example:
Transient control information, such as the time of the latest system status update for each
system in the sysplex
More permanent control information, such as information about System Logger offload
data sets
Policy information, such as the WLM service class definitions used in the sysplex
The information held in the Couple Data Sets is critical for the continuous operation of the
sysplex. If one of the system components loses access to its Couple Data Set, that
component may fail. The impact on either a single system or the entire sysplex depends on
which component loses access to its Couple Data Set, for example:
If a system loses access to all the sysplex CDSs, it is unable to update its system status.
As a result, it will be partitioned out of the sysplex.
If a system loses access to all SFM CDSs, SFM is disabled across the entire sysplex.
When the first system is IPLed into a sysplex, it reads its Couple Data Set definition from the
COUPLExx parmlib member. This system makes sure that the Couple Data Sets are
available for use in the sysplex, and then it adds them to the sysplex. Every system that
subsequently joins the sysplex must use the same Couple Data Sets.
After the systems are active in a sysplex, it is possible to change the Couple Data Set
configuration of a sysplex dynamically. For additional information about Couple Data Sets,
refer to z/OS V1R10.0 MVS Setting Up a Sysplex, SA22-7625.
8.2 The seven Couple Data Sets

You can have seven Couple Data Sets in a sysplex environment:
Sysplex
ARM
BPXMCDS
CFRM
LOGR
SFM
WLM
Notice that the Couple Data Sets are named after the system components that use them. Not
all of these components must be active in a Parallel Sysplex, however. The following list
identifies which Couple Data Sets are mandatory and which ones are optional:
In a Parallel Sysplex, the sysplex CDS and the CFRM CDS are mandatory because they
describe the Parallel Sysplex environment you are running.
166
Although the WLM CDS is not mandatory for a sysplex, it has been a part of the z/OS
Operating System since z/OS v1.4. Most sites are now running in WLM Goal mode, so the
WLM CDS will be active in most sites.
Use of the remaining four Couple Data Sets is optional. Their use in a sysplex may vary
from site to site and will depend on which functions have been enabled in your sysplex.
Couple Data Sets contain a policy, which is a set of rules and actions that systems in the
sysplex follow. For example, the WLM policy describes the performance goals and the
importance of the different workloads running in the sysplex.
Most Couple Data Sets contain multiple policies. Only one of these policies may be active at
a time. However, a new policy can be activated dynamically by using the SETXCF command.
The seven Couple Data Sets and their contents are described briefly in Table 8-1.
Table 8-1 CDS type and description
CDS
Description
Sysplex
This is the most important CDS in the sysplex. It contains the active XCF policy, which describes the
Couple Data Set and signal connectivity configuration of the sysplex and failure-related timeout values,
such as the interval after which a system is considered to have failed.
It also holds control information about the sysplex, such as:
The system status information for every system in the sysplex.
Information about XCF groups and the members of those groups.
Information about the other Couple Data Sets defined to the sysplex.
ARM
This CDS contains the active ARM policy. This policy describes how ARM registered started tasks and
batch jobs should be restarted if they abend.
BPXMCDS
This CDS contains information that is used to support the shared HFS and zFS facility in the sysplex.
This CDS does not contain a policy.
CFRM
This CDS contains the active CFRM policy and status information about the CFs. The CFRM policy
describes the CFs that are used by the sysplex and the attributes of the CF structures that can be
allocated in them.
LOGR
This CDS contains one LOGR policy. The LOGR policy describes the structures and logstreams that
you can define. It also contains information about the Logger staging data sets and offload data sets.
You could say that this CDS is like a catalog for Logger offload data sets.
SFM
This CDS contains the SFM policy. The SFM policy describes how the systems in the sysplex will
manage a system failure, a signalling connectivity failure or a CF connectivity failure.
WLM
This CDS contains the WLM policy. The WLM policy describes the performance goals and the
importance of the different workloads running in the sysplex.
8.3 Couple Data Set configuration

It is important to develop a robust design for your Couple Data Set configuration because the
information held in the Couple Data Sets is critical for the continuous operation of the sysplex.
If one of the system components that uses a Couple Data Set loses access to its Couple Data
Set, there will be an impact to either a system or the entire sysplex. The extent of the impact
depends on which component loses access to its Couple Data Set, for example:
If a system loses access to all the sysplex CDSs, it will be unable to update its system
status. As a result, it will be partitioned out of the sysplex.
Chapter 8. Couple Data Set management
167
If a system loses access to all the SFM CDSs, SFM will be disabled across the entire
sysplex.
To avoid an outage to the sysplex, it is good practice to run with a primary and an alternate
CDS. The primary CDS is used for all read and write operations. The alternate CDS will only
be used for write operations. This is to ensure the currency of the alternate CDSs contents. If
the primary CDS fails, the sysplex will automatically switch to the alternate CDS and drop the
primary CDS from the its configuration. This leaves the sysplex running on a single CDS. If
you have a spare CDS defined, you can add this to the sysplex configuration dynamically to
ensure your sysplex continues to run with two CDSs.
To avoid contention on the Couple Data Sets during recovery processing, place the primary
sysplex CDS and primary CFRM CDS on separate volumes. Normally, these Couple Data
Sets are not busy. However, during recovery processing, they can both become very busy.
We recommend the following Couple Data Set configuration:
Define three Couple Data Sets for each component: a primary CDS, an alternate CDS,
and a spare CDS.
Run with a primary and an alternate CDS.
Place the primary sysplex CDS and primary CFRM CDS on separate volumes.
Follow the recommended CDS layout listed in Table 8-2 for a single site sysplex.
Table 8-2 CDS layout
Volume 1
Volume 2
Volume 3
Primary sysplex
Alternate sysplex
Spare sysplex
Alternate CFRM
Spare CFRM
Primary CFRM
Spare LOGR
Primary LOGR
Alternate LOGR
Primary SFM
Alternate SFM
Spare SFM
Primary ARM
Alternate ARM
Spare ARM
Alternate WLM
Spare WLM
Primary WLM
Spare BPXMCDS
Primary BPXMCDS
Alternate BPXMCDS
8.4 How the system knows which CDS to use

When the first system is IPLed into a sysplex, it reads its Couple Data Set definition from the
COUPLExx parmlib member. This system makes sure that the Couple Data Sets are
available for use in the sysplex, then it adds them to the sysplex and updates the status
information in the sysplex CDS.
Every system that subsequently joins the sysplex must use the same Couple Data Sets.
When a system is IPLed into the sysplex, if there is a mismatch between the Couple Data Set
information held in the sysplex CDS and the COUPLExx parmlib member, the system will
resolve the mismatch automatically by ignoring the COUPLExx configuration and using the
Couple Data Set configuration already in use by the other systems in the sysplex.
The name of the other Couple Data Sets and the component they are associated with is
stored in the sysplex CDS. If you remove a Couple Data Set definition from the COUPLExx
168
parmlib member, this information is not deleted from the sysplex CDS. Instead, this
information remains in the sysplex CDS until it is replaced by a new definition.
After the systems are active in a sysplex, it is possible to change the Couple Data Set
configuration dynamically by using the SETXCF COUPLE command.
8.5 Managing CDSs

The following commands can be used to manage Couple Data Sets. For additional
information about these commands, see z/OS MVS System Commands, SA22-7627.
8.5.1 Displaying CDSs

To display basic information about all the CDSs in the sysplex, use the command:
D XCF,COUPLE,TYPE=ALL
An example of the output from this command is shown in Figure 8-1 on page 170. Points of
interest in the output are:
The sysplex CDS is always displayed first and the other active CDSs are displayed in
alphabetical order of the system components that use them. There is no mandatory
naming standard for Couple Data Sets. In this example, the primary sysplex CDS is called
SYS1.XCF.CDS01.
There is a line of status information after each type of CDS. This status indicates which
systems have access to both the primary and alternate CDS. The exception is the sysplex
CDS, because all systems in the sysplex must use the same sysplex CDS. This does not
refer to the status of a policy being used in the sysplex for that type of CDS.
169
D XCF,COUPLE,TYPE=ALL
PRIMARY
DSN: SYS1.XCF.CDS01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
11/20/2002 16:27:24
4
100
(52)
203
(18)
ALL TYPES OF COUPLE DATA SETS ARE SUPPORTED
GRS STAR MODE IS SUPPORTED
ALTERNATE DSN: SYS1.XCF.CDS02
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM MAXGROUP
MAXMEMBER
11/20/2002 16:27:28
4
100
203
ALL TYPES OF COUPLE DATA SETS ARE SUPPORTED
GRS STAR MODE IS SUPPORTED
ARM COUPLE DATA SETS
PRIMARY
DSN: SYS1.XCF.ARM01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 15:08:01
4
NOT PROVIDED
ALTERNATE DSN: SYS1.XCF.ARM02
VOLSER: #@$#X2
DEVN: 1D07
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
11/20/2002 15:08:04
4
NOT PROVIDED
ARM IN USE BY ALL SYSTEMS
1
BPXMCDS COUPLE DATA SETS
PRIMARY
DSN: SYS1.XCF.OMVS01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 15:24:18
4
. . .
Figure 8-1 Displaying CDS information
If you want to display a specific CDS, use the command D XCF,COUPLE,TYPE=xxxx, where xxxx
is the component name. For a list of components, refer to Table 8-1 on page 167.
8.5.2 Displaying whether a policy is active

To display if a policy is active in the sysplex, use the following command. In this example we
are displaying the status for the SFM policy.
An example of the response to this command, when SFM is active, is shown in Figure 8-2 on
page 171.
170
You can see:

The name of the current SFM policy, SFM01 1, and when it was started.
That SFM is active from the last line of the output 2.
D XCF,POL,TYPE=SFM
TYPE: SFM
POLNAME:
SFM01 1
STARTED:
07/02/2007 20:21:59
LAST UPDATED: 05/28/2004 13:44:52
An example of the response to this command when SFM is not active is shown in Figure 8-3.
The last line of the output shows that SFM is not active 3.
TYPE: SFM
POLICY NOT STARTED 3
Figure 8-3 SFM policy display when SMF is inactive
8.5.3 Starting and stopping a policy

To start a policy, use the following command. In this example we are starting an SFM policy
called SFM01:
An example of the system response to this command is shown in Figure 8-4. The system
response shows which SFM values were taken from the SFM policy 1 and 2, and which
values are system defaults 3.
THE ACTION WAS SPECIFIED FOR THIS SYSTEM. 1
19 SPECIFIED BY SPECIFIC POLICY ENTRY 2
IXC614I SFM POLICY SFM01 INDICATES MEMSTALLTIME(NO) FOR SYSTEM #@$2 AS
SPECIFIED BY SYSTEM DEFAULT 3
IXC601I SFM POLICY SFM01 HAS BEEN STARTED BY SYSTEM #@$2
Figure 8-4 Console messages when starting SFM policy
If your system programmer asks you to stop a policy, use the following command. In this
example we are stopping an SFM policy:
An example of the system response to this command is shown in Figure 8-5 on page 172.
171
IXC607I SFM POLICY HAS BEEN STOPPED BY SYSTEM #@$2
Figure 8-5 Console messages when stopping SFM policy
8.5.4 Changing the primary CDS

The Couple Data Sets used in a sysplex are defined in the COUPLExx parmlib member. After
the systems are active in the sysplex, it is possible to change the Couple Data Set
configuration dynamically by using the SETXCF COUPLE command.
There are many reasons why you might want to replace your primary and alternate CDSs.
For example, you might want to move them to another volume. You can use the following
process to replace one set of primary and alternate CDSs with another:
1. Replace the existing alternate CDS with the replacement primary CDS by using the SETXCF
COUPLE,ACOUPLE command.
2. Remove the existing primary CDS and replace it with the replacement primary CDS by
using the SETXCF COUPLE,PSWITCH command.
3. Add the replacement alternate CDS by using the SETXCF COUPLE,ACOUPLE command.
Note that all systems must agree to swap a Couple Data Set before the swap can proceed.
Make sure all the systems in the sysplex are processing normally before you start this
procedure, because the sysplex will wait for every system to respond before completing or
rejecting the request.
CDS replacement process

In the following example, we use the SFM CDS to demonstrate how to dynamically change
the CDS configuration.
Before you modify the CDS configuration, check the current CDS configuration by issuing the
following command:
An example of the response from this command in Figure 8-6 on page 173 shows the current
SFM CDS configuration. You can see that both the primary 1 and alternate 2 SFM CDSs are
defined in this sysplex and that they are in use by all systems 3.
172
PRIMARY
DSN: SYS1.XCF.SFM01 1
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 16:08:53
4
FORMAT DATA
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
11/20/2002 16:08:53
4
FORMAT DATA
SFM IN USE BY ALL SYSTEMS 3
Figure 8-6 Current SFM CDS configuration
The first step is to replace the existing alternate CDS with the replacement primary CDS. To
do this, issue the following command:
SETXCF COUPLE,ACOUPLE=SYS1.XCF.SFM03,TYPE=SFM
Figure 8-7 shows the resulting messages that are issued.
IXC309I SETXCF COUPLE,ACOUPLE REQUEST FOR SFM WAS ACCEPTED
IXC260I ALTERNATE COUPLE DATA SET REQUEST FROM SYSTEM 792
#@$2 FOR SFM IS NOW BEING PROCESSED.
IXC253I ALTERNATE COUPLE DATA SET 794
SYS1.XCF.SFM02 FOR SFM
IS BEING REMOVED BECAUSE OF A SETXCF COUPLE,ACOUPLE OPERATOR COMMAND
DETECTED BY SYSTEM #@$2
IXC263I REMOVAL OF THE ALTERNATE COUPLE DATA SET 797
SYS1.XCF.SFM02 FOR SFM IS COMPLETE
IXC251I NEW ALTERNATE DATA SET 798
SYS1.XCF.SFM03
FOR SFM HAS BEEN MADE AVAILABLE
Figure 8-7 Replacing the alternate Couple Data Set
In Figure 8-8 on page 174, you can see the WTOR 1 that you may receive when you use the
SETXCF COUPLE,ACOUPLE command to add a new alternate CDS. This WTOR asks you to
confirm that the alternate CDS can be used. It is issued because the CDS has been used
before in the sysplex.
Attention: If this WTOR is issued, consult your system programmer before replying.
173
IXC248E COUPLE DATA SET 819
SYS1.XCF.SFM03 ON VOLSER #@$#X2
FOR SFM MAY BE IN USE BY ANOTHER SYSPLEX.
013 IXC247D REPLY U TO ACCEPT USE OR D TO DENY USE OF THE COUPLE DATA
SET FOR SFM.
1
R 13,U
IEE600I REPLY TO 013 IS;U
SYS1.XCF.SFM03
Figure 8-8 Accept or deny Couple Data Set WTOR
Display the Couple Data Set configuration again by issuing the following command:
An example of the response from this command in Figure 8-9 shows the current SFM CDS
configuration. Youll notice the alternate CDS 1 is the replacement primary CDS that was
added with the previous command.
PRIMARY
DSN: SYS1.XCF.SFM01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
11/20/2002 16:08:53
4
FORMAT DATA
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
06/27/2007 02:39:06
4
FORMAT DATA
SFM IN USE BY ALL SYSTEMS
The second step is to remove the existing primary CDS and replace it with the replacement
primary CDS by issuing the following command:
SETXCF COUPLE,PSWITCH,TYPE=SFM
An example of the response from this command in Figure 8-10 on page 175 shows the
messages that are issued. These messages contain a warning to indicate you are processing
without an alternate CDS 1.
174
SETXCF COUPLE,PSWITCH,TYPE=SFM
IXC309I SETXCF COUPLE,PSWITCH REQUEST FOR SFM WAS ACCEPTED
IXC257I PRIMARY COUPLE DATA SET 805
SYS1.XCF.SFM01 FOR SFM
IS BEING REPLACED BY
SYS1.XCF.SFM03 DUE TO OPERATOR REQUEST
IXC263I REMOVAL OF THE PRIMARY COUPLE DATA SET 808
SYS1.XCF.SFM01 FOR SFM IS COMPLETE
IXC267E PROCESSING WITHOUT AN ALTERNATE 809
COUPLE DATA SET FOR SFM.
ISSUE SETXCF COMMAND TO ACTIVATE A NEW ALTERNATE 1.
Figure 8-10 Replacing the primary Couple Data Set
configuration. Notice that there is no alternate CDS in the configuration.
PRIMARY
DSN: SYS1.XCF.SFM03
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
06/27/2007 02:39:06
4
FORMAT DATA
The final step is add the replacement alternate CDS by issuing the following command:
An example of the response from this command in Figure 8-12 shows the messages that are
issued.
SYS1.XCF.SFM04
Figure 8-12 Replacing the alternate Couple Data Set
175
configuration.
PRIMARY
DSN: SYS1.XCF.SFM03
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
06/27/2007 02:39:06
4
FORMAT DATA
ALTERNATE DSN: SYS1.XCF.SFM04
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
06/27/2007 02:39:06
4
FORMAT DATA
After you have completed changing the CDS configuration, the COUPLExx parmlib member
must be updated to reflect the new configuration.
8.5.5 IPLing a system with the wrong CDS definition

Every system that joins the sysplex must use the same Couple Data Sets. If a system that is
IPLed into the sysplex has a mismatch between the Couple Data Set information held in the
sysplex CDS and the COUPLExx member, the system will resolve the mismatch
automatically by ignoring the COUPLExx configuration and use the Couple Data Set
configuration already in use by the other systems in the sysplex.
If a mismatch occurs for the sysplex CDS, the message shown in Figure 8-14 on page 177 is
issued during the IPL.
176
IXC268I THE COUPLE DATA SETS SPECIFIED IN COUPLE00 ARE IN

INCONSISTENT STATE
IXC275I COUPLE DATA SETS SPECIFIED IN COUPLE00 ARE 098
PRIMARY:
SYS1.XCF.CDS01 1
ON VOLSER #@$#X1
ALTERNATE: SYS1.XCF.CDS02 2
ON VOLSER #@$#X2
IXC275I RESOLVED COUPLE DATA SETS ARE 100
PRIMARY:
SYS1.XCF.CDS03 3
ON VOLSER #@$#X1
ALTERNATE: SYS1.XCF.CDS04 4
ON VOLSER #@$#X2
Figure 8-14 CDS mismatch at IPL
In this example, SYS1.XCF.CDS01 1 and SYS1.XCF.CDS02 2 are specified in the

COUPLExx parmlib member. XCF replaces them with SYS1.XCF.CDS03 3 and
SYS1.XCF.CDS04 4 because these are the active sysplex CDSs.
Note: If this mismatch occurs for any other Couple Data Set, the system will resolve the
mismatch and it will not issue any messages to indicate that a mismatch occurred.
This situation and other CDS error situations that occur during an IPL are discussed in
Chapter 3, IPLing systems in a Parallel Sysplex on page 39.
8.5.6 Recovering from a CDS failure

Couple Data Sets are critical data sets required for the continuous operation of the sysplex.
As previously mentioned, if one of the system components that uses a Couple Data Set loses
access to its CDS, there will be an impact to either a system or the entire sysplex. The extent
of the impact depends on which component loses access to its CDS.
This section describes the impact on the sysplex if one or more systems lose access to either
a primary or an alternate CDS. This could be caused by:
A hardware reserve on the volume that the CDS resides on
A loss of all the paths to the volume that the CDS resides on
If your CDS configuration matches the layout recommended in Table 8-2 on page 168, then
this type of failure will affect more than one CDS at a time.
Loss of access to a primary CDS

If systems in the sysplex lose access to a primary CDS, XCF will attempt to retry the failing
I/O request for 5 minutes. If no I/O completes in that time, XCF will automatically switch to the
alternate CDS and drop the primary CDS from its configuration. This leaves the sysplex
running on a single CDS.
If you have a spare CDS defined, you should add this to the sysplex configuration dynamically
by using the SETXCF COUPLE,ACOUPLE command to ensure that your sysplex continues to run
with two CDSs.
177
An example of some of the messages you may see in this scenario are displayed in
Figure 8-15.
Messages such as 1, 2, and 3 are issued periodically to indicate that *MASTER* and
XCFAS are incurring I/O delays during the 5-minute timeout.
Messages such as 4 are issued periodically to indicate which CDS is experiencing I/O
delays and for how long.
A message such as 5 is issued to indicate that the CDS has been removed because of an
I/O error.
A message such as 6 is issued to warn you that there is no alternate for this CDS.
. . .
IOS078I 1D06,5A,XCFAS, I/O TIMEOUT INTERVAL HAS BEEN EXCEEDED 514

FOR AN ACTIVE REQUEST. THE ACTIVE REQUEST HAS BEEN TERMINATED.
QUEUED REQUESTS MAY HAVE ALSO BEEN TERMINATED. 2
. . .
FOR A QUEUED REQUEST. THE QUEUED REQUEST HAS BEEN TERMINATED.
. . .
IXC246E BPXMCDS COUPLE DATA SET 694
SYS1.XCF.OMVS01 ON VOLSER #@$#X1,
DEVN 1D06, HAS BEEN EXPERIENCING I/O DELAYS FOR 247 SECONDS.
IXC253I PRIMARY COUPLE DATA SET 987

SYS1.XCF.OMVS01 FOR BPXMCDS
IS BEING REMOVED BECAUSE OF AN I/O ERROR
ERROR CASE: UNRESOLVED I/O TIMEOUT 5
IXC263I REMOVAL OF THE PRIMARY COUPLE DATA SET 803
SYS1.XCF.OMVS01 FOR BPXMCDS IS COMPLETE
COUPLE DATA SET FOR BPXMCDS.
ISSUE SETXCF COMMAND TO ACTIVATE A NEW ALTERNATE. 6
Figure 8-15 Loss of access to a primary CDS
Loss of access to an alternate CDS

If systems in the sysplex lose access to an alternate CDS, then XCF will attempt to retry the
failing I/O request for 5 minutes. If no I/O completes in that time, XCF will automatically
remove the alternate CDS from the configuration. This leaves the sysplex running on a single
CDS.
If you have a spare CDS defined, you should add this to the sysplex configuration dynamically
by using the SETXCF COUPLE,ACOUPLE command to ensure that your sysplex continues to run
with two CDSs.
178
An example of some of the messages you may see in this scenario are shown in Figure 8-16.
These messages are almost identical to the messages you see when you lose access to a
primary CDS.
Messages such as 1, 2, and 3 are issued periodically to indicate that *MASTER* and
XCFAS are incurring I/O delays during the 5-minute timeout.
Messages such as 4 are issued periodically to indicate which CDS is experiencing I/O
delays and for how long.
A message such as 5 is issued to indicate that the CDS has been removed because of an
I/O error.
A message such as 6 is issued to warn you that there is no alternate for this CDS.
. . .

FOR AN ACTIVE REQUEST. THE ACTIVE REQUEST HAS BEEN TERMINATED.
QUEUED REQUESTS MAY HAVE ALSO BEEN TERMINATED. 2
. . .
FOR A QUEUED REQUEST. THE QUEUED REQUEST HAS BEEN TERMINATED.
. . .
IXC246E CFRM COUPLE DATA SET 800
SYS1.XCF.CFRM02 ON VOLSER #@$#X1,
DEVN 1D06, HAS BEEN EXPERIENCING I/O DELAYS FOR 249 SECONDS.
IXC253I ALTERNATE COUPLE DATA SET 982

SYS1.XCF.CFRM02 FOR CFRM
IS BEING REMOVED BECAUSE OF AN I/O ERROR
ERROR CASE: UNRESOLVED I/O TIMEOUT 5
IXC263I REMOVAL OF THE ALTERNATE COUPLE DATA SET 778
SYS1.XCF.CFRM02 FOR CFRM IS COMPLETE
COUPLE DATA SET FOR CFRM.
ISSUE SETXCF COMMAND TO ACTIVATE A NEW ALTERNATE.
Figure 8-16 Loss of access to an alternate CDS
Loss of access to both CDSs

This section examines the impact on the sysplex if a system loses access to both the primary
and alternate CDSs.
Sysplex CDS
If a system loses access to both the primary and alternate sysplex CDSs, that system would
be unable to update its system status and as a result it would be partitioned out of the
sysplex.
179
ARM CDS
If a system loses access to both the primary and alternate ARM CDSs, ARM services on that
system are disabled until a primary CDS is assigned.
BPXMCDS CDS
If a system loses access to both the primary and alternate BPXCDS CDSs, that system loses
the ability to share UNIX System Services file systems with other systems in the sysplex.
CFRM CDS
If a system loses access to both the primary and alternate CFRM CDSs, that system is
placed in a non-restartable disabled wait state X' 0A2' reason code X'9C'.
LOGR CDS
If a system loses access to both the primary and alternate LOGR CDSs, the logger loses
connectivity to its inventory data set. The logger address space on that system terminates
itself.
SFM CDS
If a system loses access to both the primary and alternate SFM CDSs, SFM is disabled
across the entire sysplex.
WLM CDS
If a system loses access to both the primary and alternate WLM CDSs, then Workload
Manager continues to run, using the policy information that was in effect at the time of the
failure. WLM is described as being in independent mode, operating only on local data, and
does not transmit data to other members of the sysplex.
8.5.7 Concurrent CDS and system failure

If a permanent error occurs on a CDS, XCF will attempt to remove the failing CDS from the
sysplex configuration. Before the removal can proceed, all systems must agree to remove the
CDS. XCF expects all systems to participate in this decision to remove the CDS, and it will
wait for every system in the sysplex to respond before completing or rejecting the request.
If a sysplex CDS is being removed and at the same time, one or more systems are
unresponsive, then the removal process will hang. This is because of a deadlock on the
sysplex CDS between permanent error processing and SFM:
Permanent error processing prohibits serialized access to the sysplex CDS while it waits
until either the unresponsive system is removed from the sysplex or it responds to the
request to remove the sysplex CDS.
SFM cannot remove the unresponsive system until it has updated the recovery record in
the sysplex CDS, but it cannot update the sysplex CDS until permanent error processing
has finished.
A concurrent failure of one or more CDSs and one or more systems requires a special
recovery procedure. When a concurrent failure occurs, if XCF does not receive a response to
its CDS removal request from all systems, XCF will issue message IXC256A, as shown in
180
IXC256A REMOVAL OF PRIMARY COUPLE DATA SET

SYS1.XCF.CDS01 FOR SYSPLEX 1
CANNOT COMPLETE UNTIL
THE FOLLOWING SYSTEM(S) ACKNOWLEDGE THE REMOVAL:
#@$3 2
Figure 8-17 Couple data set removal cannot complete
This message indicates:

Which CDS 1 is being removed by permanent error processing.
Which system or systems 2 are not responding to the request to remove the CDS.
Note: This message can roll off the screen. You can use the D R,L command to retrieve
this message from Action Message Retention Facility (AMRF), if it is enabled.
If the system or systems identified in message IXC256A are in an unrecoverable state, such
as a disabled WAIT state, those systems must be removed from the sysplex to allow CDS
recovery to proceed.
Remove each of the systems identified in message IXC256A by performing the following
steps:
1. Perform a system reset of the unresponsive system or systems.
2. Issue the command V XCF,sysname,OFFLINE from one of the active systems.
3. Issue the command V XCF,sysname,OFFLINE,FORCE from the same active system used in
step 2.
4. Repeat steps 1, 2, and 3 for every unresponsive system.
The V XCF,sysname,OFFLINE,FORCE command may not be effective. Permanent error
processing may continue to hang if the active systems do not detect that the state of the
outgoing system has transitioned from active to inactive. The active systems do not detect
this transition because permanent error processing defers serialized access to the sysplex
CDS. In this case, permanent error processing continues to wait because it still expects the
unresponsive system or systems to participate in the removal of the sysplex CDS.
If the IXC256A messages do not clear on all systems where the deadlock occurred, you will
see message IXC256A reissued on the active system where you issued the V XCF,OFFLINE
commands. This time, message IXC256A will name one or more of the active systems rather
than the unresponsive system or systems.
To break the deadlock, issue the following command for each unresponsive system from
every active system:
V XCF,sysname,OFFLINE
The system from which the partition request is issued always recognizes that the target
system will not participate in CDS removal even if the partition request cannot be processed
because of the deadlock. If every system requests partitioning, permanent error processing
runs to completion to remove the sysplex CDS and then SFM can partition out the
unresponsive system or systems normally.
181
182
Chapter 9.
XCF management
This chapter describes the Cross-System Coupling Facility (XCF) and operational aspects of
XCF including:
XCF signalling
XCF groups
XCF system monitoring
183
9.1 Introduction to XCF management

XCF is a component of z/OS that provides an interface for authorized programs to
communicate with other programs, either in the same system or a different system in the
same sysplex. This service is known as XCF signalling services.
XCF is responsible for updating the sysplex Couple Data Sets (CDS) with the system status
and XCF Groups and members in the sysplex. XCF is also responsible for managing,
accessing, and keeping track of the sysplex and function Couple Data Sets. All requests to
any of the Couple Data Sets are done through XCF services.
XCF provides notification services for all systems within the sysplex. It informs the existing
members of a group when:
A new member joins
One member updates its state information in the sysplex CDS
A member changes state or leaves the group.
Systems cannot join the sysplex unless they can establish XCF connectivity. XCF, in the
joining system, checks to ensure its system name is unique within the sysplex, and that the
sysplex name does not match that of any existing sysplex members.
Following a failure in either a system or application, XCF works with other operating
system components to decide if a system should be removed from the sysplex using
Sysplex Failure Management (SFM) or to decide if, and where, work or an application
should be restarted using Automatic Restart Manager (ARM), if these components are
enabled.
XCF is at the core of a sysplex. If one system loses all XCF connectivity, you will have a
system down. If all systems lose XCF connectivity, you will have a sysplex down. Many
core system functions, such as CONSOLE and GRS, use XCF. If XCF slows down, those
functions can be impacted, slowing down everyone using the sysplex in turn. We discuss
some operational aspects of the XCF connectivity, signalling, and things that could cause a
sysplex slowdown or a stalled member within this chapter.
9.2 XCF signalling

Signalling is the mechanism through which XCF group members communicate in a sysplex. A
communication path must exist between every member of the sysplex. XCF uses signalling
PATHIN and PATHOUT connections to allow group members to communicate with each
other.
XCF supports two mechanisms for communicating between systems:
Channel-to-Channel adapters (CTCs)
CF Structures
We recommend that you implement both mechanisms in your sysplex for availability
purposes.
184
9.2.1 XCF signalling using CTCs

CTCs have the following attributes when used by XCF for signalling:
They are not bi-directional.
For two systems to communicate using CTCs, you must have (at least) two CTC devices
between each pair of systems: one defined as PATHIN (Read), and one as PATHOUT
(Write)
They are synchronous.
There is no inherent buffering or delay within the CTC.
They are point to point.
You need at least one PATHIN and one PATHOUT for each system you want to talk to.
This makes CTCs more complex to set up and manage than CF structures. For example,
moving from a 10-way sysplex to an 11-way sysplex, with two pairs of CTC paths, would
require an additional 80 CTC devices.
CTCs used for PATHIN and PATHOUT signalling:
Each CTC device number is defined as either a PATHIN or a PATHOUT.
Each PATHOUT must communicate with a PATHIN on another sysplex member.
Each PATHIN must communicate with a PATHOUT on another sysplex member.
Both the PATHIN and PATHOUT devices may use the same physical CHPID.
Figure 9-1 shows PATHIN and PATHOUT communications between a four-system sysplex
using CTCs.
SYSA
SYSB
SYSC
SYSD
SYSA COUPLExx Member

PATHIN DEVICE(4020,4028,4030,4038)
PATHIN DEVICE(4040,4048)
PATHOUT DEVICE(5020,5028,5030,5038)
PATHOUT DEVICE(5040,5048)
Figure 9-1 XCF signalling using CTCs
PATHIN and PATHOUT device numbers are defined in the COUPLExx parmlib member.
An example of CTC addressing standards is given here.
All PATHINs begin with 40xy, where:
x is the system where the communication is being sent from.
y is the device number 0-7 on one CHPID and 8-F on another CHPID.
All PATHOUTs begin with 50xy, where:
x is the system where the communication is being sent to.
y is the device number 0-7 on one CHPID and 8-F on another CHPID.
Chapter 9. XCF management
185
9.2.2 XCF signalling using structures

XCF can use the list structure capabilities of the Coupling Facility (CF) to provide signalling
connectivity between sysplex members.
Structures have the following attributes when used by XCF:
They are bidirectional (from a definition perspective).
Each XCF structure should be defined as both PATHIN and PATHOUT to every system.
They are asynchronous.
A message is sent to CF. Then CF notifies the target system. Finally, the target system
collects the message.
They are multi-connected.
A CF structure would be used as both PATHIN and PATHOUT by every system. This
makes them much easier to define and manage than CTCs.
Figure 9-2 shows PATHIN and PATHOUT communications between a four-system sysplex
using structures.
SYSB
SYSA
SYSA COUPLExx Member

PATHIN STRNAME(IXC_DEFAULT_1)
PATHOUT STRNAME(IXC_DEFAULT_1)
SYSB COUPLExx Member
CF
SYSC
SYSC COUPLExx Member

SYSD
SYSD COUPLExx Member

Figure 9-2 XCF signalling using structures
The COUPLExx parmlib member defines the PATHOUT and PATHIN with a structure
name of IXC_xxxxx (the structure name must begin with IXC).
Multiple signalling structures can be specified in the COUPLExx member.
The structure name must be in the active CFRM policy.
During IPL, z/OS will establish a signalling path to every other image using the CF.
9.2.3 Displaying XCF PATHIN

The D XCF,PATHIN (or PI) command, as shown in Figure 9-3 on page 187, can be issued to
determine the current status of the structures and CTCs PATHIN and PATHOUT signalling
paths.
186
D XCF,PI
PATHIN FROM SYSNAME: ???????? - PATHS NOT CONNECTED TO OTHER SYSTEMS
DEVICE (LOCAL/REMOTE): 4010/???? 4018/????
1
PATHIN FROM SYSNAME: SC64
DEVICE (LOCAL/REMOTE): 4020/5010 4028/5018
2
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
3
...
Figure 9-3 D XCF,PATHIN command
Figure 9-4 shows output from a D XCF,PI command that was entered on system SC64:
1 The question marks (?) after PATHIN devices 4010 and 4018 indicate that they are not
connected to a PATHOUT device.
2 The PATHIN CTC device number 4020 on a system named SC63 (local) is connected to
PATHOUT CTC device number 5010 on the system named SC64 (remote).
3 IXC_DEFAULT_1 and IXC_DEFAULT_2 are CF structures that are being used by SC63 for
PATHIN and PATHOUT signalling to the other systems in the sysplex.
Both CTCs and CF structures are being used for signalling. We recommend that you allocate
each signalling structure in separate CFs.
9.2.4 Displaying XCF PATHOUT

The D XCF,PATHOUT (or PO) command, as shown in Figure 9-4, can be issued to determine the
current status of the structures and CTCs PATHOUT and PATHIN signalling paths. The
information displayed is the same as the response to the PATHIN command, except that the
local CTC devices reflect the PATHOUT device numbers.
D XCF,PO
PATHOUT FROM SYSNAME: ???????? - PATHS NOT CONNECTED TO OTHER SYSTEMS
DEVICE (LOCAL/REMOTE): 5010/???? 5018/????
PATHOUT FROM SYSNAME: SC64
DEVICE (LOCAL/REMOTE): 5020/4010 5028/4018
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
...
Figure 9-4 D XCF,PATHOUT command
9.2.5 Displaying XCF PATHIN - CTCs

This display command, as shown in Figure 9-5 on page 188, provides additional information
about the PATHIN and PATHOUT CTCs.
The status of the PATHIN devices can be:
Working
The signalling path is operational.
Linking
The signalling path is in the process of establishing signalling links

(CTC state when a system is being IPLed).
187
Restarting
XCF is restarting a failed path (state a CTC is left in if a system is

removed from the sysplex).
Inoperative
Hardware or definition error, but will be the state of a system that

shares a common COUPLExx member with other systems in the
sysplex.
Stopping
A CTC is in the process of stopping (SETXCF STOP command).
D XCF,PI,DEV=ALL
LOCAL DEVICE
REMOTE
PATHIN
PATHIN
SYSTEM
STATUS
4010
???????? INOPERATIVE
4018
4020
SC64
WORKING
4028
SC64
WORKING
...
REMOTE
PATHOUT RETRY
????
100
????
100
5010
100
5018
100
LAST MXFER
MAXMSG RECVD TIME
750
1
750
1
750 65393
274
750 39713
591
Figure 9-5 D XCF,PI,DEV=ALL
9.2.6 Displaying XCF PATHOUT - CTCs

This display command, as shown in Figure 9-6, provides additional information about the
PATHIN and PATHOUT CTCs. The status of the PATHOUT devices is the same as the
PATHIN devices; see 9.2.5, Displaying XCF PATHIN - CTCs on page 187 for more detail.
The D XCF,PATHOUT (or PO) command can be issued to determine the current status of the
structures and CTCs PATHOUT and PATHIN signalling paths. The information displayed is
the same as the response to the PATHIN command, except that the local CTC devices reflect
the PATHOUT device numbers.
D XCF,PO,DEV=ALL
LOCAL DEVICE
REMOTE
PATHOUT
PATHOUT
SYSTEM
STATUS
5010
5018
5020
SC64
WORKING
5028
SC64
WORKING
...
REMOTE
PATHIN
????
????
4010
4018
RETRY
100
100
100
100
LAST MXFER
MAXMSG RECVD TIME
750
1
750
1
750 65393
274
750 39713
591
Figure 9-6 D XCF,PO,DEV=ALL
9.2.7 Displaying XCF PATHIN - structures

This display command, as shown in Figure 9-7 on page 189, provides PATHIN information
about ALL signalling structures. It can also be used to display individual structure names
(with wildcards).
The status of the PATHIN structures can be:
188
Working
The signalling path is operational.
Starting
Verifying that the signalling path is suitable for XCF.
Linking
The signalling path is in the process of establishing signalling links

(state when a system is being IPLed).
Restarting
XCF is restarting a failed path (state a structure is left in if a system is

removed from the sysplex).
Inoperative
The signalling path is defined to XCF but is not usable until hardware
or definition problems are resolved.
Stopping
A structure is in the process of stopping (SETXCF STOP command).
D XCF,PI,STRNM=ALL
STRNAME
REMOTE PATHIN
PATHIN
SYSTEM STATUS
IXC_DEFAULT_1
WORKING
#@$1
WORKING
#@$2
WORKING
IXC_DEFAULT_2
WORKING
#@$1
WORKING
#@$2
WORKING
STRNAME
PATHIN
LIST
IXC_DEFAULT_1
9
11
IXC_DEFAULT_2
9
11
UNUSED 1
PATHS RETRY
6
10
10
LAST MXFER
MAXMSG RECVD TIME
2000
18369 1911
55620 2332
2000
66492 1936
74970 2116
REMOTE
SYSTEM
PATHIN
STATUS
DELIVRY BUFFER
PENDING LENGTH
MSGBUF SIGNL
IN USE NUMBR NOBUF
#@$1
#@$2
WORKING
WORKING
0
0
956
956
36 18369
0 55620
0
0
#@$1
#@$2
WORKING
WORKING
0
0
956
956
50 66492
12 74970
0
0
Figure 9-7 D XCF,PI,STRNAME=ALL
1 Unused paths - the values shown in this column indicate the number of lists in the list
structure that are available for use as signalling paths.
9.2.8 Displaying XCF PATHOUT - structures

The display command in Figure 9-8 on page 190 provides PATHIN information about ALL
signalling structures. It can also be used to display individual structure names (with
wildcards).
The status of the PATHOUT devices is the same as the PATHIN devices; see Figure 9-7 for
the details. The information displayed is the same as the response to the PATHIN.
189
D XCF,PO,STRNAME=ALL
STRNAME
REMOTE PATHOUT
PATHOUT
SYSTEM STATUS
IXC_DEFAULT_1
WORKING
#@$1
WORKING
#@$2
WORKING
IXC_DEFAULT_2
WORKING
#@$1
WORKING
#@$2
WORKING
STRNAME
REMOTE PATHOUT
PATHOUT
LIST SYSTEM STATUS
IXC_DEFAULT_1
8 #@$1
WORKING
10 #@$2
WORKING
IXC_DEFAULT_2
8 #@$1
WORKING
10 #@$2
WORKING
UNUSED
PATHS RETRY
6
10
10
TRANSFR BUFFER
PENDING LENGTH
TRANSPORT
MAXMSG CLASS
2000 DEFAULT
2000 DEFAULT
MSGBUF SIGNL MXFER

IN USE NUMBR TIME
0
0
956
956
24 62772
18 62749
1480
2189
0
0
956
956
16 21848
28 70706
783
4394
Figure 9-8 D XCF,PO,STRNAME=ALL
9.2.9 Starting and stopping signalling paths

To start or stop an inbound or outbound signalling path you can use the SETXCF START or
SETXCF STOP command, as shown in Figure 9-9. These commands can be used to start or
stop CTC devices or CF structures as signalling paths to and from specific systems.
SETXCF STOP,{PATHIN,{DEVICE=([/]indevnum[,[/]indevnum]...)}
{STRNAME=(strname[,strname]...)
}
[,UNCOND=NO|YES]
{PATHOUT,{DEVICE=([/]outdevnum[,[/]outdevnum]...)} }
{STRNAME=(strname[,strname]...)
}
[,UNCOND=NO|YES]
Figure 9-9 Syntax of SETXCF STOP command
The syntax for the SETXCF START command is the same as the SETCXCF STOP command shown
in Figure 9-9.
An example of the output of a SETXCF STOP command of an inbound signalling structure path
is shown in Figure 9-10 on page 191. The command was issued on system name #@$1.
190
SETXCF STOP,PI,STRNAME=IXC_DEFAULT_2
IXC467I STOPPING PATHIN STRUCTURE IXC_DEFAULT_2 032
RSN: OPERATOR REQUEST
RSN: PROPAGATING STOP OF STRUCTURE
DIAG073: 08690003 0E011000 01000000 00000000 00000000
RSN: PROPAGATING STOP OF STRUCTURE
DIAG073: 08690003 0E011000 01000000 00000000 00000000
...
IXC467I STOPPING PATHOUT STRUCTURE IXC_DEFAULT_2 LIST 9 756
RSN: OTHER SYSTEM STOPPING ITS SIDE OF PATH
...
SUCCESSFULLY: PROPAGATING STOP OF STRUCTURE
SUCCESSFULLY: OTHER SYSTEM STOPPING ITS SIDE OF PATH
...
Figure 9-10 SETXCF STOP,PI,STRNAME=strname
1 The command output shows the signalling structure (IXC_DEFAULT_2), which was being
used as an inbound path (PATHIN) to system #@$1, being stopped.
2 It further shows the outbound paths (PATHOUT), which were being used on the other
systems to communicate with the inbound path (PATHIN) on #@$1, also being stopped. As a
result, the other systems in the sysplex can no longer communicate with #@$1 via the
IXC_DEFAULT_2 signalling structure.
After stopping the inbound (PATHIN) path from the IXC_DEFAULT_2 signalling structure on
#@$1, the PATHIN and PATHOUTs on #@$1 and PATHOUTs on both #@$2 and #@$3
were displayed, with the output shown in Figure 9-11 on page 192.
191
D XCF,PI
STRNAME:
IXC_DEFAULT_1
STRNAME:
IXC_DEFAULT_1
D XCF,PO
PATHOUT TO SYSNAME:
#@$2
STRNAME:
IXC_DEFAULT_1
PATHOUT TO SYSNAME:
#@$3
STRNAME:
IXC_DEFAULT_1
RO #@$2,D XCF,PO
RESPONSE=#@$2
PATHOUT TO SYSNAME:
#@$1
STRNAME:
IXC_DEFAULT_1
PATHOUT TO SYSNAME:
#@$3
STRNAME:
IXC_DEFAULT_1
RO #@$3,D XCF,PO
RESPONSE=#@$3
PATHOUT TO SYSNAME:
#@$1
STRNAME:
IXC_DEFAULT_1
PATHOUT TO SYSNAME:
#@$2
STRNAME:
IXC_DEFAULT_1
IXC_DEFAULT_2
IXC_DEFAULT_2
3
IXC_DEFAULT_2
4
IXC_DEFAULT_2
Figure 9-11 Display paths after a SETXCF STOP,PI command
1 The display of the PATHIN on #@$1 shows the inbound path is only using the
IXC_DEFAULT_1 structure.
2 The display of the PATHOUT on #@$1 shows the outbound path using both structures,
IXC_DEFAULT_1 and IXC_DEFAULT_2.
3 The display of the PATHOUT on #@$2 shows the outbound path is only using the
4 the display of the PATHOUT on #@$3 shows the outbound path is only using the
9.2.10 Transport classes

As XCF messages are generated, they are assigned to a transport class based on group
name or message size. The messages are copied into a signal buffer from the XCF buffer
pool. The messages are sent over outbound paths (PATHOUT) defined for the appropriate
transport class. Inbound paths are not directly assigned transport classes.
You can use the operator command D XCF,CLASSDEF,CLASS=ALL, shown in Figure 9-12 on
page 193, to obtain information about the current behavior of the XCF transport classes. The
command, which has a single image scope, returns information regarding message traffic
192
throughout the sysplex. The command returns information regarding the size of messages
being sent through the transport class to all members of the sysplex, and it identifies current
buffer usage needed to support the load.
D XCF,CD,CLASS=ALL
TRANSPORT
CLASS
DEFAULT
CLASS
LENGTH
MAXMSG
DEFAULT
956
2000
ASSIGNED
GROUPS
UNDESIG
DEFAULT TRANSPORT CLASS

SUM MAXMSG:
6000
SEND CNT:
44728
SEND CNT:
311
SEND CNT:
10481
SEND CNT:
1
SEND CNT:
138
USAGE FOR SYSTEM #@$1

IN USE:
22 NOBUFF:
BUFFLEN (FIT):
956
BUFFLEN (BIG): 4028
BUFFLEN (BIG): 8124
BUFFLEN (BIG): 16316

SUM MAXMSG:
6000
SEND CNT:
39092
SEND CNT:
314
SEND CNT:
10495
SEND CNT:
2
SEND CNT:
139

IN USE:
22 NOBUFF:
BUFFLEN (FIT):
956
BUFFLEN (BIG): 4028
BUFFLEN (BIG): 8124

SUM MAXMSG:
2768
SEND CNT:
39146
SEND CNT:
1072
SEND CNT:
9

IN USE:
2 NOBUFF:
BUFFLEN (FIT):
956
BUFFLEN (BIG): 4028
BUFFLEN (BIG): 8124
0
2
3
0
2
3
0
2
Figure 9-12 Display transport class information
With this information you can determine how the transport classes are being used and,
potentially, whether additional transport classes are needed to help signalling performance. In
Figure 9-12, most of the messages being sent fit into the DEFAULT buffer 2, which is the only
transport class defined (956 bytes) 1. There is other traffic which could use a larger buffer
(8 k), and which would benefit from having an additional transport class defined 3.
For more detailed information about transport classes and their performance considerations,
refer to the white paper Parallel Sysplex Performance: XCF Performance Considerations
authored by Joan Kelley and Kathy Walsh, which is available at:
http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100743/
9.2.11 Signalling problems

If the Parallel Sysplex has CTCs and structures defined, the loss of one signalling path does
not affect the availability of the members. After the problem that caused the loss has been
resolved, XCF will automatically start using the signalling path again when it is available.
Loss of one CTC signalling path

For example, if a permanent I/O error occurs on a CTC, you receive the messages displayed
in Figure 9-13 on page 194.
193
IOS102I DEVICE 4020 BOXED, PERMANENT I/O ERROR

IEE793I 4020
PENDING OFFLINE AND BOXED
IXC467I RESTARTING PATHIN DEVICE 4020 506
USED TO COMMUNICATE WITH SYSTEM SC64
RSN: I/O ERROR WHILE WORKING
IXC467I STOPPING PATHIN DEVICE 4020 508
USED TO COMMUNICATE WITH SYSTEM SC64
RSN: HALT I/O FAILED
DIAG073:08220003 0000000C 0000000C 00000001 00000000
IXC307I STOP PATHIN REQUEST FOR DEVICE 4020 COMPLETED 509
SUCCESSFULLY: HALT I/O FAILED
Figure 9-13 Loss of a CTC signalling path
Problem diagnosis and recovery should be performed on the failing device. After the problem
has been resolved and the device has been successfully varied online, XCF will start using it
again, as shown in Figure 9-14.
V 4020,ONLINE,UNCOND
IEE302I 4020
ONLINE
IXC306I START PATHIN REQUEST FOR DEVICE 4020 COMPLETED 339
SUCCESSFULLY: DEVICE CAME ONLINE
IXC466I INBOUND SIGNAL CONNECTIVITY ESTABLISHED WITH SYSTEM SC64 340
VIA DEVICE 4020 WHICH IS CONNECTED TO DEVICE 5010
Figure 9-14 CTC signalling path recovery
For examples of signalling problems during IPLs, refer to 3.6.6, Unable to establish XCF
connectivity on page 56.
9.3 XCF groups

Before an application can use XCF signalling services, it must join the same XCF group that
its peers are in, using an agreed group name. To obtain a list of all groups in the sysplex, as
well as the number of members in each group, use the D XCF,GROUP command as shown in
Figure 9-15.
D XCF,G
GROUPS(SIZE):
ATRRRS(3)
EZBTCPCS(3)
...
SYSMCS2(8)
SYSWLM(3)
COFVLFNO(3)
IDAVQUI0(3)
CSQGPSMG(3)
IGWXSGIS(6)
SYSRMF(3)
SYSXCF(2)
SYSTTRC(3)
XCFJES2A(3)
Figure 9-15 XCF group list
Group names are important because they are one way of segregating XCF message traffic.
Table 9-1 on page 195 lists common IBM XCF groups. Other vendor products may have other
XCF groups, which are not listed in the table.
194
Table 9-1 IBM XCF groups

Exploiter
Group Name
Exploiter
Group Name
Any lock structure user
IXCLOxxx
Object Access Method
Name not fixed
APPC, ASCH
SYSATBxx
OMVS
SYSBPX
CICS MRO
DFHIR000
RACF
IRRXCF00
CICSplex System Mgr
Name not fixed
RMF
SYSRMF
CICSVR
DWWCVRCM
RRS
ATRRRS
Console Services
SYSMCS, SYSMCS2
SA for z/OS
INGXSGxx
DAE
SYSDAE
TCP/IP
EZBTCPCS
DB2
Names not fixed (3)
Tivoli Workload Scheduler
Name not fixed
DFSMS and PDSE sharing
SYSIGW00, SYSIGW01
Trace
SYSTTRC
ENF
SYSENF
TSO Broadcast
SYSIKJBC
GRS
SYSGRS, SYSGRS2
VLF
COFVLFNO
HSM
ARCxxxxxx
VSAM/RLS
IDAVQUI0, IGWXSGIS,
SYSIGW01, SYSIGW02,
SYSIGW03
I/O Ops (part of SA/390)
ESCM
VTAM, TCP/IP
ISTXCF, ISTCFS01
IMS
Names not fixed

(numerous groups)
WebSphere MQ
CSQGxxxx
IOS
SYSIOSxx
WLM
SYSWLM
IRLM
DXRxxx
XES
IXCLOxxx
JES2 MAS
Name not fixed
zFS
IOEZFS
JES3 complex
Name not fixed
You can use the D XCF,G,<groupname> command, as shown in Figure 9-16, to find out what
members are in a given group.
D XCF,G,ISTXCF
IXC332I 02.16.52
GROUP ISTXCF:
DISPLAY XCF 628

#@$1M$$$USIBMSC
#@$3M$$$USIBMSC
#@$2M$$$USIBMSC
Figure 9-16 XCF groupname list
1 Member names connected to the ISTXCF group.

However, the member names are not always very informative, so you may need to issue the
D XCF,G,<groupname>,ALL command to display additional information about a specific
<groupname>. An example of the output of the command is shown in Figure 9-17 on
page 196. The command was issued on system #@$1 to list the ISTXCF group that is the
VTAM generic resource group.
195
D XCF,G,ISTXCF,ALL
INFORMATION FOR GROUP ISTXCF
MEMBER NAME:
SYSTEM:
#@$1M$$$USIBMSC
#@$1
#@$2M$$$USIBMSC
#@$2
#@$3M$$$USIBMSC
#@$3
JOB ID:
NET
NET
NET
STATUS:
ACTIVE
ACTIVE
ACTIVE
INFO FOR GROUP ISTXCF MEMBER #@$1M$$$USIBMSC ON SYSTEM #@$1

MEMTOKEN: 030000B6 001C0003
ASID: N/A
SYSID: 030002B2
INFO: ONLY AVAILABLE ON SYSTEM #@$1

MEMTOKEN: 01000120 001C0001
ASID: 001B
SYSID: 010002AF
INFO: CURRENT
COLLECTED: 06/27/2007 02:38:41.760953
SIGNALLING SERVICE
MSGO ACCEPTED:
MSGO XFER CNT:
MSGO XFER CNT:
MESSAGE TABLE:
3
2298
1973
325
SENDPND
0
NOBUFFER:
LCL CNT:
LCL CNT:
RESPPND
0
0
0
0
BUFF LEN: 956

BUFF LEN: 8124
COMPLTD MOSAVED MISAVED

0
0
0
MSGI RECEIVED:
MSGI XFER CNT:
2153
2259
PENDINGQ:
XFERTIME:
0
2408
MSGI PENDINGQ:
SYMPATHY SICK:
IO BUFFERS
0
0
DREF
0
PAGEABLE
0
EXIT 04C08100: 06/27/2007 02:38:41.718997 ME 00:00:00.000017

GROUP SERVICE
EVNT RECEIVED:
PENDINGQ:
EXIT 01E51CA0: 06/27/2007 01:31:30.723220 01 00:00:00.000027

MEMTOKEN: 020000B2 001C0002
ASID: 001B
SYSID: 020002B1
INFO: CURRENT
COLLECTED: 06/27/2007 02:38:41.866796
...
Figure 9-17 XCF groupname ALL list
1 Summary information at the top of the output of the command shows the member names,
the system name and the job name of the task associated with the member, and the status of
the member.
2 The system (#@$3) was unable to obtain the most current data for system #@$1 from the
sysplex Couple Data Set. To obtain the latest information, the command would need to be
issued from the #@$1 system.
3 The signalling service data describes the use of the XCF Signalling Service by the member.
One line will appear for each different signal size used by the member.
196
9.3.1 XCF stalled member detection

There may be times when an associated task of an XCF groups member experiences
problems that could delay the processing of incoming XCF messages. This is known as a
stalled member.
So how do you know when an XCF member is not collecting its messages? Some symptoms
are:
The application is unresponsive.
There are console messages from the exploiter.
There are XCF stalled member console messages.
Some common causes of these problems are:
The applications WLM dispatch priority is set too low.

A single CPU is dominated by higher priority work.
There is a tight loop in an unrelated work unit.
The LPAR weight is set too low.
There are two ways to address stalled members and the problems they can cause:
Let SFM automatically address the situation for you. This requires a recent enhancement
to SFM related to sympathy sickness, as explained in Chapter 5, Sysplex Failure
Management on page 73.
Take some manual action to identify the culprit and resolve the situation before other
users start being impacted. We will discuss this further in this section.
To identify XCF members that are not collecting their messages quickly enough, XCF notes
the time every time it schedules a member's message exit. If the exit has not completed
processing in four minutes, XCF issues a Stalled Member message (IXC431I), as shown in
Figure 9-18.
10:59:09.06 IXC431I GROUP B0000002 MEMBER M1 JOB MAINASID ASID 0023
STALLED AT 02/06/2007 10:53:57.823698 ID: 0.2
LAST MSGX: 02/06/2007 10:58:13.112304 12 STALLED 0 PENDINGQ
LAST GRPX: 02/06/2007 10:53:53.922204 0 STALLED 0 PENDINGQ
11:00:17.23 *IXC430E SYSTEM SC04 HAS STALLED XCF GROUP MEMBERS
Figure 9-18 IXC431I stalled member message
The drawback of relying on these messages is that a member can be stalled for around four
minutes before IXC431I is issued. If you have an idea that you have a stalled member, you
can issue the D XCF,GROUP command to look for stalled members, as shown in Figure 9-19.
D XCF,G
IXC331I 11.00.31 DISPLAY XCF
GROUPS(SIZE):1*B0000002(3)
ISTCFS01(3)
SYSGRS(3)
SYSIGW01(3)
SYSIOS02(1)
. . .
COFVLFNO(3)
SYSDAE(4)
SYSIEFTS(3)
SYSIKJBC(3)
SYSIOS03(1)
CTTXGRP(3)
SYSENF(3)
SYSIGW00(3)
SYSIOS01(1)
SYSJES(3)
Figure 9-19 XCF stalled member display
1 The asterisk (*) indicates stalls.
197
The D XCF,G command will indicate the group with the stalled member after 30 seconds. An
asterisk (*) indicates the stall. However (prior to z/OS 1.8 or APAR OA09194), you must issue
the command on every system. You only receive the indication if the stalled member is on the
system where the command was issued.
Issue the D XCF,G,<groupname> command to learn which members are in a given group, as
D XCF,G,B0000002
IXC332I 02.16.52
GROUP B0000002:
DISPLAY XCF 628

1*M1
M2
M3
Figure 9-20 XCF groupname list with stall
1 Member names connected to the B0000002 group, with an asterisk (*) indicating the stalled
member.
Issue the D XCF,G,<groupname>,<member> command to display additional information for a
given member, as shown in Figure 9-21.
D XCF,G,B0000002,M1
INFORMATION FOR GROUP SYSMCS2
MEMBER NAME:
SYSTEM:
JOB ID:
STATUS:
#@$1
#@$1
TEST01
ACTIVE
. . .
SIGNALLING SERVICE
MSGO ACCEPTED: 2401 NOBUFFER: 0
MSGO XFER CNT: 0 LCL CNT: 2401 BUFF LEN: 956
MSGI RECEIVED: 844 PENDINGQ: 0
MSGI XFER CNT: 4001 XFERTIME: N/A
EXIT 01FB9300: 02/06/2007 10:57:15.939863 ME 00:00:00.001107
*EXIT 01FB9500: 02/06/2007 10:53:58.181009 ME RUNNING
. . .
Figure 9-21 XCF member list with stall
Addressing the situation:

If you have a stalled member that is processing messages slowly, move the task
associated with the job ID to a better WLM service class, to give it more system resources
(CPU).
If you have stalled member that is not processing messages at all, and you cannot resolve
the hang, that address space is probably hung and will need to be cancelled.
After the address space goes away, all the buffers holding messages destined for the
address space will be freed up.
9.4 XCF system monitoring

XCF monitors the health of the sysplex. Every few seconds, XCF updates its entry in the
sysplex Couple Data Set (CDS), inserting the current time (using GMT format). You can
display the local system's view of the last time this was done for all systems by executing the
D XCF,SYSPLEX,ALL command, as shown in Figure 9-22 on page 199.
198
D XCF,S,ALL
SYSTEM
#@$2
2084 6A3A
N/A 06/27/2007 21:03:28
#@$3
2084 6A3A
N/A 06/27/2007 21:03:32
#@$1
2084 6A3A
N/A 06/27/2007 21:03:29
SYSTEM STATUS
ACTIVE
ACTIVE
ACTIVE
TM=SIMETR
TM=SIMETR
TM=SIMETR
Figure 9-22 XCF display of active systems
XCF also checks the time stamp for all the other ACTIVE members. If any systems time
stamp is older than the current time, minus that systems INTERVAL value from the
COUPLExx parmlib member, then XCF suspects that the system may be dead and flags the
status of the system with a non-active status, as shown in Figure 9-23.
D XCF,S,ALL
SYSTEM
#@$2
2084 6A3A
N/A 06/27/2007 21:15:08
#@$3
2084 6A3A
N/A 06/27/2007 21:19:45
#@$1
2084 6A3A
N/A 06/27/2007 21:19:44
SYSTEM STATUS
MONITOR-DETECTED STOP
ACTIVE
TM=SIMETR
ACTIVE
TM=SIMETR
Figure 9-23 XCF display of active systems with a failed system
1 MONITOR-DETECTED STOP means the system has not updated its status on the CDS within the
time interval specified on that system's COUPLExx parmlib member. This can mean that:
The system is going through reconfiguration.

A spin loop is occurring.
The operator pressed stop.
The system is in a restartable wait state.
The system lost access to the couple data set.
The next step depends on whether there is an active SFM policy:

If there is an active SFM policy in your sysplex, XCF checks to see if the dead system is
issuing any XCF signals. If it is still issuing signals, message IXC426D is issued. If it is not,
issuing signals and you specify ISOLATETIME(0), SFM will automatically partition the
system from the sysplex.
If there is no active SFM policy in your sysplex, message IXC402D is issued, asking you to
remove the sick system, or wait sssss seconds, as shown in Figure 9-24, and check again
for the heartbeat (checking for XCF signals is not performed in this case).
*026 IXC402D #@$2 LAST OPERATIVE AT 21:15:08. REPLY DOWN AFTER SYSTEM
RESET, OR INTERVAL=SSSSS TO SET A REPROMPT TIME.
Figure 9-24 IXC402D message
199
200
10
Chapter 10.
Managing JES2 in a Parallel

Sysplex
JES2 is able to exploit many of the functions provided by a Parallel Sysplex. This chapter
covers some operational scenarios for using JES2 in a Parallel Sysplex. JES3 is covered in
Chapter 13, Managing JES3 in a Parallel Sysplex on page 271.
The following topics are covered in this chapter:
MAS and JESXCF management
JES2 checkpoint management
JES2 subsystem restart
JES2 subsystem shutdown
JES2 batch management
JES2 input and output management in a MAS
JES2 and WLM management
201
10.1 Introduction to managing JES2 in a Parallel Sysplex

z/OS uses a job entry subsystem (JES) to receive jobs into the operating system, to schedule
them for processing, and to control their output processing. JES is that component that
provides the necessary functions to get jobs into, and output from, the system. JES receives
jobs into the system and processes all output data produced by the job. It is designed to
provide efficient spooling, scheduling, and management facilities for z/OS.
Why does z/OS need a JES? By separating job processing into a number of tasks, z/OS
operates more efficiently. At any point in time, the computer system resources are busy
processing the tasks for individual jobs, while other tasks are waiting for those resources to
become available. In its simplest view, z/OS divides the management of jobs and resources
between the JES and the base control program of z/OS. In this manner, JES manages jobs
before and after running the program; the base control program manages them during
processing.
JES also provides a spool that can be used to store some input and some output for the jobs.
Historically, output was stored on the spool before it was printed. However, with more modern
tools such as SDSF, it is possible to view the output on the spool and thus reduce the need
for printing. See Chapter 11, System Display and Search Facility and OPERLOG on
page 231 for more information about this topic.
z/OS has two versions of job entry systems: JES2 and JES3. JES2 is the most common and
is the JES referred to in this chapter. As mentioned, JES3 is covered in Chapter 13,
Managing JES3 in a Parallel Sysplex on page 271.
10.2 JES2 multi-access spool support

JES2 uses multi-access spool (MAS) support to share spool volume or volumes across many
systems. A MAS is required to support the Parallel Sysplex Automatic Restart Manager
(ARM) function of restarting subsystems on different z/OS images. A base sysplex is a
requirement for a JES2 MAS. If a shared JES2 environment is to be implemented on more
than one system, then a MAS configuration is required.
There is no requirement for the JES2 MAS, or as it is commonly called, the JESPLEX, to
match the sysplex. It is not uncommon to have multiple JESPLEXes within a single sysplex,
as shown in Figure 10-1 on page 203.
202
JESPlex Systems 1, 2, 3, 4
JESPlex
Systems
A, B, C
Stand alone
JES2 Systems
Figure 10-1 Multiple JESPLEXes in a single sysplex
JESXCF is a system address space that contains functions and data areas used by JES2 to
send messages to other members in the MAS. It provides a common set of messaging
services that guarantee delivery and first-in first-out (FIFO) queuing of messages between
JES2 subsystems. When a JES2 member fails or a participating z/OS system fails, other
JES2 members are notified through JESXCF.
The JESXCF address space is created as soon as JES2 first starts on any system in the
Parallel Sysplex. Only the first system up in the Parallel Sysplex creates structures. The other
systems will find these on connection to the CF and use them.
JES2 has two assisting address spaces, JES2MON and JES2AUX. These address spaces
provide support services for JES2. The name is derived from the subsystem definition in
PARMLIB(IEFSSNxx), as is the JES2 PROC. Thus, if the entry SUBSYS SUBNAME(JES2)
is replaced with SUBSYS SUBNAME(FRED), then the system would have started tasks
FRED, FREDMON and FREDAUX.
JES2AUX is an auxiliary address space used by JES2 to hold spool buffers. These are user
address space buffers going to the spool.
JES2MON is the JES2 health monitor that is intended to address situations where JES2 is
not responding to commands and where it is not possible to determine the issue. Operations
can communicate directly with JES2MON.
Resources such as those listed are monitored across the MAS:
JNUM
JQEs
JOEs
TGs
Job Numbers
Job Queue Elements
Job Output Elements
Spool Space and Track Groups
If you want to cancel a job, restart a job, or send a message, you can do so from any member
of the MAS. It is not necessary to know which member the job is running on. In addition, the
$C command can also be used to cancel an active time-sharing user (TSO user ID) from any
member of the MAS.
Chapter 10. Managing JES2 in a Parallel Sysplex
203
JES2 thresholds
The first member of the MAS that detects a JES2 threshold issues a message to the console.
There are three ways in which threshold values can be set:
In the JES2 PARMLIB, by the system programmer
Modified by a JES2 command, such as $TSPOOLDEF,TGSPACE=WARN=85
Not set, in which case the default value is used
Other members still issue the messages, but rather than sending them to the console, they
are written to the hardcopy log. This means that you must have an effective method for
monitoring consoles on a sysplex-wide basis to avoid missing critical messages. This could
be done by having automation tool that issues an alert to bring the problem to the attention of
an operator.
10.3 JES2 checkpoint management

The JES2 checkpoint is the general term used to describe the checkpoint data set that JES2
maintains on either direct access storage devices (DASD) or on a Coupling Facility (CF). The
checkpoint data set (regardless of where it resides) contains a backup copy of the JOB and
OUTPUT queues. These queues contain information about what work is yet to be processed
and how far along that work has progressed. The checkpoint locking mechanism is used to
serialize access when any change is required to the JOB or OUTPUT queues (for example, a
change in phase or processing status, such as purging an output).
Similar to the spool data sets, the checkpoint data set is accessible by all members of the
multi-access spool (MAS) complex, but only one member has control (access) of the
checkpoint data set at a time. Furthermore, the checkpoint data set provides information for
all members of the MAS about jobs and the output from those jobs.
Checkpoint data in a CF structure is suitable for MAS configurations of all sizes (2 to 32
members). The CF serialized list removes the need for both the software lock and the
hardware RESERVE logic. JES2 use of the CF provides for more equitable sharing of the
checkpoint among all members. On DASD, a faster processor can monopolize the
checkpoint. When the checkpoint uses a FIFO method of queuing, all MAS members can
have equal access to the data.
Placing the checkpoint in the CF allows automatic switching of the checkpoint to an alternate
checkpoint without incurring an outage to the JES2 subsystem. When a problem with the
primary checkpoint is encountered, automatic switching can be achieved by defining a
CKPT1 structure in a CF and a NEWCKPT1 on DASD or in another CF. The secondary
checkpoint CKPT2 should be active and CKPT2 should be on DASD.
JES2 checkpoint duplexing is required to allow auto checkpoint recovery action to proceed.
We recommend that you have duplexing always enabled.
Note: It is not recommended to place both checkpoints on Coupling Facility structures, or
to place the primary checkpoint on DASD and the secondary checkpoint on a Coupling
Facility structure. If both checkpoints reside on Coupling Facilities that become volatile (a
condition where, if power to the Coupling Facility device is lost, the data is lost), then your
data is less secure than when a checkpoint data set resides on a DASD. If no other active
MAS member exists, you can lose all checkpoint data and require a JES2 cold start.
Placing the primary checkpoint on a DASD while the secondary checkpoint resides on a
Coupling Facility provides no benefit to an installation.
204
For more information about placement of the checkpoint, see JES2 Initialization and Tuning
Guide, SA22-7532.
10.3.1 JES2 checkpoint reconfiguration

The JES2 checkpoint can be reconfigured in a controlled manner via the checkpoint
reconfiguration dialog. This dialog can also be used to change from a structure to a data set,
or to implement a new checkpoint data set.
For example, if there is a disk subsystem upgrade and the volumes containing the JES2
checkpoint are to be moved, then the following process might be followed.
1.
2.
3.
4.
Suspend the usage of checkpoint.

Move the checkpoint to a new structure, with a different name.
Move the checkpoint to a new volume.
Use checkpoint reconfiguration to resume use of the checkpoint.
This section illustrates suspending a checkpoint and then reactivating it. The starting
configuration can be seen in Figure 10-2.
$D CKPTDEF
$HASP829 CKPTDEF
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
CKPT1=(STRNAME=JES2CKPT_1,INUSE=YES,
VOLATILE=YES),CKPT2=(DSNAME=SYS1.JES2.CKPT2,
Figure 10-2 $D CKPTDEF - display current checkpoint configuration
1 The current value of INUSE=YES will change.

The reconfiguration dialog is initiated as seen in Figure 10-3 on page 206.
205
$T CKPTDEF,RECONFIG=YES
$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTING
$HASP233 REASON FOR JES2 CHECKPOINT
* RECONFIGURATION IS OPERATOR
*
REQUEST
$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTED
* - DRIVEN BY
*
MEMBER #@$2
*$HASP271 CHECKPOINT RECONFIGURATION OPTIONS
*
*
VALID RESPONSES ARE:
*
*
'1' - FORWARD CKPT1 TO NEWCKPT1
*
*
'5' - SUSPEND THE USE OF CKPT1
*
'6' - SUSPEND THE USE OF CKPT2
*
'CANCEL' - EXIT FROM RECONFIGURATION
*
CKPTDEF (NO OPERANDS)
- DISPLAY MODIFIABLE
*
SPECIFICATIONS
*
CKPTDEF (WITH OPERANDS) - ALTER MODIFIABLE
*
SPECIFICATIONS
*189 $HASP272 ENTER RESPONSE (ISSUE D R,
* MSG=$HASP271 FOR RELATED MSG)
R 189,6
IEE600I REPLY TO 189 IS;6
$HASP280 JES2 CKPT2 DATA SET (SYS1.JES2.CKPT2 ON
IS NO LONGER IN USE
$HASP255 JES2 CHECKPOINT RECONFIGURATION COMPLETE
Figure 10-3 $T CKPTDEF,RECONFIG=YES to remove checkpoint2
1 The reconfiguration options available.

2 We choose to suspend CKPT2.
3 Reconfiguration is complete.
After a successful reconfiguration, we are no longer using CKPT2, as shown in Figure 10-4.
$D CKPTDEF
$HASP829 CKPTDEF
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
VOLSER=#@$#M1,INUSE=NO),NEWCKPT1=(DSNAME=,
VOLSER=),NEWCKPT2=(DSNAME=,VOLSER=),
MODE=DUPLEX,DUPLEX=ON,LOGSIZE=7,
4
5
Figure 10-4 $D CKPTDEF - CKPT2 not active
4 CKPT2 has the value INUSE=NO, indicating that it is not active.

5 Even though it says MODE=DUPLEX and DUPLEX=ON, we are not in duplex mode
because we have suspended the use of CKPT2.
206
Now we can move the volume to the new DASD subsystem and resume the use of CKPT2,
as shown in Figure 10-4 on page 206.
$T CKPTDEF,RECONFIG=YES
*$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTING
*$HASP233 REASON FOR JES2 CHECKPOINT
* RECONFIGURATION IS OPERATOR
*
REQUEST
*$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTED
* - DRIVEN BY MEMBER #@$2
*$HASP271 CHECKPOINT RECONFIGURATION OPTIONS
*
*
*
*
*
'8' - UPDATE AND START USING CKPT2
*
*
*
SPECIFICATIONS
*
*
SPECIFICATIONS
* MSG=$HASP271 FOR RELATED MSG) SPECIFICATIONS
R190,8
IEE600I REPLY TO 190 IS;8
*$HASP273 JES2 CKPT2 DATA SET WILL BE ASSIGNED TO
*
*
SYS1.JES2.CKPT2 ON #@$#M1
*
*
*
*
'CONT'
- PROCEED WITH ASSIGNMENT
*
*
*
SPECIFICATIONS
*
*
SPECIFICATIONS
* MSG=$HASP273 FOR RELATED MSG)
R 191,CONT
IEE600I REPLY TO 191 IS;CONT
$HASP280 JES2 CKPT2 DATA SET (SYS1.JES2.CKPT2 ON
#@$#M1) IS NOW IN USE
10
11
Figure 10-5 $T CKPTDEF,RECONFIG=YES to resume using CKPT2
6 The available options.

7 This time option 8 was chosen, resume CKPT2.
8 A confirmation message is issued.
9 The reply CONT confirms this is the correct configuration.
10 Message indicating we are using CKPT2.
11 Checkpoint reconfiguration is complete.
207
10.3.2 JES2 loss of CF checkpoint reconfiguration

The operator may experience times when the JES2 checkpoint fails and needs to be
reconfigured. The following examples illustrate the message flow in a Parallel Sysplex
environment when the CF containing the JES2 checkpoint structure is lost. Note that it is
likely that the problem with the JES2 checkpoint structure will be symptomatic of a larger
problem, most likely the failure of the entire CF.
Example 1 and Example 2 show a difference between automatic and operator dialog
switching with the CKPT1 and CKPT2 configuration defined. Example 3 illustrates a flow
when NEWCKPT1 is not defined.
The JES2 checkpoint structure held in the CF has a default disposition of keep. This means
that when the last JES2 MAS member is closed, the structure remains in the CF. It only
disappears when the CF is deactivated or fails. When the first JES2 MAS member is
restarted, the structure is reallocated.
Example 1: JES2 reconfigure CKPT1 to NEWCKPT1 with OPVERIFY=NO

The starting checkpoint configuration for this example is shown in Figure 10-6. We simulated
a CF failure by deactivating the CF LPAR. Figure 10-7 on page 209 shows JES2
automatically moving to the new checkpoint. This configuration enables JES2 to continue
processing immediately and thus has lesser impact. It is recommended that you configure
automation to bring this to the operators attention.
$D CKPTDEF
$HASP829 CKPTDEF
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
NEWCKPT1=(DSNAME=SYS1.JES2.CKPT1,
VOLSER=#@$#J1),NEWCKPT2=(DSNAME=,VOLSER=),
Figure 10-6 Checkpoint configuration - OPVERIFY=NO
1
2
3
4
5
208
The primary checkpoint is a structure.

The duplex checkpoint is active and on DASD.
NewCKPT1 is defined.
The mode is duplex.
Opverify is NO.
1
2
3
4
*IXL158I PATH 10 IS NOW NOT-OPERATIONAL TO CUID: 030F 435 1

*$HASP275 MEMBER #@$2 -- JES2 CKPT1 DATA SET - I/O ERROR - REASON CODE 2
*$HASP275 MEMBER #@$3 -- JES2 CKPT1 DATA SET - I/O ERROR - REASON CODE 3
. . .
$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTED - DRIVEN BY 454 4
MEMBER #@$1
$HASP290 MEMBER #@$2 -- JES2 CKPT1 IXLLIST LOCK REQUEST FAILURE 455 5
*** CHECKPOINT DATA SET NOT DAMAGED BY THIS MEMBER ***
RETURN CODE = 0000000C
REASON CODE = 0C080C06
RECORD
= UNKNOWN
. . .
$HASP280 JES2 CKPT1 DATA SET (SYS1.JES2.CKPT1 ON #@$#J1) IS NOW IN USE 6
. . .
*094 $HASP294 WAITING FOR RESERVE (VOL #@$#J1). 7
REPLY REPLY 'CANCEL' TO END WAIT
*$HASP256 FUTURE AUTOMATIC FORWARDING OF CKPT1 IS SUSPENDED UNTIL 8
NEWCKPT1 IS RESPECIFIED.
ISSUE $T CKPTDEF,NEWCKPT1=(...) TO RESPECIFY
$HASP255 JES2 CHECKPOINT RECONFIGURATION COMPLETE 9
Figure 10-7 CF Failure - JES2 checkpoint OPVERIFY=NO
1
2
3
4
5
6
7
8
9
An indication that the CF is failing.

System #@$2 detects an error on the checkpoint structure.
Checkpoint reconfiguration dialog is being initiated.
Checkpoint recovery messages.
System has started using the new CKPT1 data set.
System is waiting for reserve to be obtained for new checkpoint data set.
As a side affect of moving the checkpoint, we no longer have a NEWCKPT1 defined.
Checkpoint reconfiguration has completed.
Example 2: JES2 Reconfigure CKPT1 to NEWCKPT1 with OPVERIFY=YES

The only difference in Figure 10-8 on page 210 from the previous example is
OPVERIFY=YES 5. This time, when the CF failure is simulated, JES2 prompts for an operator
intervention before continuing to use the new checkpoint configuration. This results in the
JES2 WTOR 4 in Figure 10-9 on page 210 that is requesting the operator to verify or modify
the proposed new checkpoint.
209
$D CKPTDEF
$HASP829 CKPTDEF
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
NEWCKPT1=(DSNAME=SYS1.JES2.CKPT1,
VOLSER=#@$#J1),NEWCKPT2=(DSNAME=,VOLSER=),
ALLCKPT=WTOR),OPVERIFY=YES
1
2
3
4
Figure 10-8 Checkpoint configuration - OPVERIFY=YES
1
2
3
4
5
The primary checkpoint is a structure.

The duplex checkpoint is active and on DASD.
NEWCKPT1 is defined.
The mode is duplex.
Opverify is YES.
IXL158I PATH 0F IS NOW NOT-OPERATIONAL TO CUID: 030F 153
. . .
$HASP290 MEMBER #@$1 -- JES2 CKPT1 IXLLIST LOCK REQUEST FAILURE
REASON CODE = 0C1C0C06
RECORD
= UNKNOWN
. . .
$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTED - DRIVEN BY
MEMBER #@$3
$HASP273 JES2 CKPT1 DATA SET WILL BE ASSIGNED TO NEWCKPT1 548
SYS1.JES2.CKPT1 ON #@$#J1
3
4

'CONT'
'TERM'
'DELETE'
CKPTDEF
- PROCEED WITH ASSIGNMENT

- TERMINATE MEMBERS WITH I/O ERROR ON CKPT1
- DISCONTINUE USING CKPT1
(NO OPERANDS) - DISPLAY MODIFIABLE
SPECIFICATIONS
SPECIFICATIONS
*146 $HASP272 ENTER RESPONSE (ISSUE D R,MSG=$HASP273 FOR RELATED MSG)
. . .
R 146,CONT
. . .
147 $HASP294 WAITING FOR RESERVE (VOL #@$#J1). REPLY 'CANCEL' TO END
WAIT
IEE400I THESE MESSAGES CANCELLED - 147.
$HASP280 JES2 CKPT1 DATA SET (SYS1.JES2.CKPT1 ON #@$#J1) IS NOW IN USE
$HASP256 FUTURE AUTOMATIC FORWARDING OF CKPT1 IS SUSPENDED UNTIL 564
NEWCKPT1 IS RESPECIFIED.
ISSUE $T CKPTDEF,NEWCKPT1=(...) TO RESPECIFY
Figure 10-9 CF Failure - JES2 checkpoint OPVERIFY=YES
210
1
2
3
4
5
6
An indication that the CF is failing.

Checkpoint reconfiguration dialog is being initiated.
The new checkpoint data set, depending on the operator response to this message.
Operator response CONT confirms the suggested new checkpoint.
Checkpoint reconfiguration has completed.
Example 3: JES2 Reconfigure CKPT1 to CKPT2 with no NEWCKPT1 or

NEWCKPT2
This example illustrates a situation when there is no NEWCKPT1 value coded, as shown in
Figure 10-10.
$D CKPTDEF
$HASP829 CKPTDEF
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
$HASP829
1
2
Figure 10-10 Checkpoint configuration - NEWCKPT1 and NEWCKPT2 not defined
1 CKPT1 is an active structure.

2 Neither NEWCKPT1 nor NEWCKPT2 is defined.
After creating a CF failure, JES2 initiates the checkpoint reconfiguration dialog shown in
211
IXL158I PATH 0F IS NOW NOT-OPERATIONAL TO CUID: 030F 453

$HASP275 MEMBER #@$3 -- JES2 CKPT1 DATA SET - I/O ERROR
$HASP290 MEMBER #@$3 -- JES2 CKPT1 IXLLIST LOCK REQUEST FAILURE 861
REASON CODE = 0C1C0C06
RECORD
= UNKNOWN
IXC518I SYSTEM #@$3 NOT USING 862
. . .
$HASP233 REASON FOR JES2 CHECKPOINT RECONFIGURATION IS CKPT1 I/O 505
ERROR(S) ON 3 MEMBER(S)
$HASP285 JES2 CHECKPOINT RECONFIGURATION STARTED - DRIVEN BY 506
MEMBER #@$2
$HASP282 NEWCKPT1 DSNAME, VOLUME AND STRNAME ARE NULL 507

'TERM'
- TERMINATE MEMBERS WITH I/O ERROR ON CKPT1
'DELETE' - DISCONTINUE USING CKPT1
SPECIFICATIONS
SPECIFICATIONS
153 $HASP272 ENTER RESPONSE (ISSUE D R,MSG=$HASP282 FOR RELATED MSG)
. . .
R 153,DELETE
IEE600I REPLY TO 153 IS;DELETE
. . .
$HASP280 JES2 CKPT1 DATA SET (STRNAME JES2CKPT_1) IS NO LONGER IN USE
154 $HASP294 WAITING FOR RESERVE (VOL #@$#M1). REPLY 'CANCEL' TO END
WAIT
$HASP280 JES2 CKPT1 DATA SET (STRNAME JES2CKPT_1) IS NO LONGER IN USE
IEE400I THESE MESSAGES CANCELLED - 154.
3
4
5
Figure 10-11 CF failure - JES2 checkpoint no NEWCKPT1 or NEWCKPT2 defined
1 An indication that the CF is failing.

2 JES2 tells operator there is no NEWCKPT1 defined.
3 The response that will terminate all JES2 systems unable to access the checkpoint; in this
case, that would be all systems.
4 The option that stops using the checkpoint with a problem, which is the option selected in
this example.
5 If desired, a new checkpoint data set or structure could be defined. If selected, a new
WTOR would be provided after the definition is complete and new options would allow the
new checkpoint to be used.
6 Following the reply DELETE, JES2 indicates that the primary checkpoint is no longer in use.
7 Checkpoint reconfiguration is complete.
212
10.3.3 JES2 checkpoint parmlib mismatch

The checkpoint reconfiguration allows the dynamic changing of the checkpoint configuration.
This means that it is possible for the JES2 PARMLIB definition and the active definition to get
out of sync. When this happens, JES2 detects the condition during JES2 startup and asks for
confirmation of which checkpoint should be used. This dialog can be seen in Figure 10-12. In
this example, parmlib specifies that CKPT1 is a data set, but JES2 has determined it was last
using a structure.
Because JES2 does not start until this reply is answered, any such misconfigurations will
increase the JES2 outage and should be avoided.
*$HASP416 VERIFY CHECKPOINT DATA SET INFORMATION 229
*
VALUES FROM CKPTDEF
CKPT1=(DSNAME=SYS1.JES2.CKPT1,VOLSER=#@$#J1,INUSE=YES),
CKPT2=(DSNAME=SYS1.JES2.CKPT2,VOLSER=#@$#M1,INUSE=YES)
*
VALUES JES2 WILL USE
CKPT1=(STRNAME=JES2CKPT_1,INUSE=YES),
CKPT2=(DSNAME=SYS1.JES2.CKPT2,VOLSER=#@$#M1,INUSE=YES),
LAST WRITTEN WEDNESDAY, 4 JUL 2007 AT 23:33:50 (GMT)
*194 $HASP417 ARE THE VALUES JES2 WILL USE CORRECT? ('Y' OR 'N')
R 194,Y
IEE600I REPLY TO 194 IS;Y
$HASP478 INITIAL CHECKPOINT READ IS FROM CKPT1 234
(STRNAME JES2CKPT_1)
LAST WRITTEN WEDNESDAY, 4 JUL 2007 AT 23:33:50 (GMT)
$HASP493 JES2 MEMBER-#@$1 QUICK START IS IN PROGRESS
Figure 10-12 Checkpoint mismatch during JES2 startup
1 The values in PARMLIB.

2 The values that were last active.
3 The operator confirms to continue using the last active configuration.
4 JES2 is starting.
10.4 JES2 restart

When JES2 is started, or restarted, specific messages are generated. There are three JES2
start options:
JES2 cold start
A cold start affects all JES2 members in the MAS. It formats the spool, which results in
deleting all existing queues. A cold start can only be performed when the first JES2 starts
in the MAS.
JES2 hot start
A hot start occurs when JES2 restarts after a JES2 abend.
JES2 warm start
Normally, a JES2 warm start is done. This occurs when JES2 is shut down cleanly and a
cold start is not performed.
213
The messages generated in a Parallel Sysplex differ, depending upon whether the restart is
on the first JES2 MAS member or on an additional JES2.
The following examples are discussed in this chapter:
Cold start on the first JES2 in a Parallel Sysplex MAS

Cold start on an additional JES2 in a Parallel Sysplex
Warm start on the first JES2 in a Parallel Sysplex
Warm start on an additional JES2 in a Parallel Sysplex MAS
Hot start on the first JES2 in a Parallel Sysplex MAS
Hot start on an additional JES2 in a Parallel Sysplex MAS
The configuration used is the recommended Parallel Sysplex configuration with two duplexed
checkpoints: CKPT1 located in the CF, and CKPT2 on DASD.
Note: The message flow assumes NOREQ is specified in the options. Otherwise, a
$HASP400 message is also issued, requiring a $S to start JES2 processing.
10.4.1 JES2 cold start

This section describes two situations that can occur with JES2 cold start in Parallel Sysplex
MAS.
Cold start on the first JES2 in a Parallel Sysplex MAS

A cold start on the first JES2 subsystem after a clean shutdown produces the usual cold start
dialog, which the operator responds to. In Figure 10-13 on page 215, the structures exist prior
to the JES2 cold start. If the structures were not allocated in the CF, there would be messages
indicating the successful allocation of the checkpoint structures.
A cold start formats spool and the checkpoint data sets 6, 7, 8 and 9, which results in all input,
output, and held queues being purged. If a cold start is required, consideration should be
given to using the JES2 SPOOL offload and reload process to save the required queues.
Note: If your sites automation product starts JES2, it may interfere by replying to the JES2
startup messages when a cold start is attempted.
214
S JES2
IEF677I WARNING MESSAGE(S) FOR JOB JES2 ISSUE
017 $HASP426 SPECIFY OPTIONS - JES2 z/OS 1.8 SSNAME=JES2
. . .
17COLD,NOREQ
. . .
IXZ0001I CONNECTION TO JESXCF COMPONENT
ESTABLISHED GROUP XCFJES2A MEMBER N1$#@$2
$HASP9084 JES2 MONITOR ADDRESS SPACE STARTED FOR JES2
$HASP537 THE CURRENT CHECKPOINT USES 2946 4K RECORDS
IXL014I IXLCONN REQUEST FOR STRUCTURE JES2CKPT_#@$1_1 080
WAS SUCCESSFUL. JOBNAME: JES2 ASID: 001
CONNECTOR NAME: JES2_#@$2 CFNAME: FACIL01
IXL015I STRUCTURE ALLOCATION INFORMATION FOR
STRUCTURE JES2CKPT_#@$1_1, CONNECTOR NAME JES2_#@$2
CFNAME
---------------------------------------FACIL01
FACIL02
PREFERRED CF ALREADY SELECTED AC001800
$HASP436 CONFIRM COLD START ON 083
CKPT1 - STRNAME=JES2CKPT_#@$1_1
CKPT2 - VOLSER=#@$#Q1 DSN=SYS1.#@$2.CKPT2
SPOOL - PREFIX=#@$#Q DSN=SYS1.#@$2.HASPACE
018 $HASP441 REPLY 'Y' TO CONTINUE INITIALIZATION OR 'N' TO TERMINATE
IN RESPONSE TO MESSAGE HASP436
. . .
R 018,Y
. . .
(SYS1.#@$2.CKPT1 ON #@$#Q1)
$HASP405 JES2 IS UNABLE TO DETERMINE IF OTHER MEMBERS ARE ACTIVE
019 $HASP420 REPLY 'Y' IF ALL MEMBERS ARE DOWN (IPL REQUIRED),
'N' IF NOT
REPLY 19,Y
$HASP478 INITIAL CHECKPOINT READ IS FROM CKPT1
(SYS1.#@$2.CKPT1 ON #@$#Q1)
$HASP405 JES2 IS UNABLE TO DETERMINE IF OTHER
MEMBERS ARE ACTIVE
$HASP266 JES2 CKPT2 DATA SET IS BEING FORMATTED
$HASP267 JES2 CKPT2 DATA SET HAS BEEN
SUCCESSFULLY FORMATTED
$HASP266 JES2 CKPT1 DATA SET IS BEING FORMATTED
$HASP267 JES2 CKPT1 DATA SET HAS BEEN
SUCCESSFULLY FORMATTED
$HASP492 JES2 COLD START HAS COMPLETED
1
2
3
6
7
8
9
10
Figure 10-13 JES2 cold start
1
2
3
4
5
6
Reply indicating a cold start is to be performed.

Connection to the XCF group is established.
Connection to the checkpoint structure is established.
Verification that the parameters are correct.
Verification that no other systems are active to this MAS.
CKPT1 is being formatted.
215
7 CKPT2 has been formatted.

8 CKPT1 is being formatted.
9 CKPT2 has been formatted.
10 Cold start has completed.
Cold start on an additional JES2 in a Parallel Sysplex MAS

An attempt to perform a cold start of JES2 when there is an active JES2 system in the MAS,
even after a clean shutdown, is not possible. Figure 10-14 shows JES2 terminating because it
has attempted to initialize while there were active members in the XCF group.
S JES2
039 $HASP426 SPECIFY OPTIONS - JES2 z/OS 1.7
SSNAME=JES2
R 39,COLD,NOREQ
IEE600I REPLY TO 039 IS;COLD,NOREQ
WAS SUCCESSFUL. JOBNAME: JES2 ASID: 001C
$HASP436 CONFIRM COLD START ON 388
CKPT1 - STRNAME=JES2CKPT_#@$1_1
CKPT2 - VOLSER=#@$#Q1 DSN=SYS1.#@$2.CKPT2
SPOOL - PREFIX=#@$#Q DSN=SYS1.#@$2.HASPACE
040 $HASP441 REPLY 'Y' TO CONTINUE INITIALIZATION OR 'N' TO TERMINATE
IN RESPONSE TO MESSAGE HASP436
. . .
REPLY 40,Y
. . .
(STRNAME JES2CKPT_#@$1_1)
LAST WRITTEN THURSDAY, 5 JUL 2007 AT 22:28:31 (GMT)
$HASP792 JES2 HAS JOINED XCF GROUP XCFJES2B THAT INCLUDES ACTIVE
MEMBERS THAT ARE NOT PART OF THIS MAS
MEMBER=N1$#@$2,REASON=DIFFERENT COLD START TIME
$HASP428 CORRECT THE ABOVE PROBLEMS AND RESTART JES2
IXZ0002I CONNECTION TO JESXCF COMPONENT DISABLED,
GROUP XCFJES2B MEMBER N1$#@$1
2
3
Figure 10-14 JES2 cold start - second system
1 Reply indicating a cold start is to be performed.

2 Connection to the checkpoint structure is established.
3 Connector name.
4 JES2 rejects the second JES2 from doing a cold start.
5 JESXCF disconnects the system, thus preventing any JESXCF communication until JES2
is correctly restarted.
10.4.2 JES2 warm start

After JES2 has shut down cleanly, with the primary checkpoint in a CF and the secondary on
DASD, there are three restart options:
This is the first JES2 system started, and the checkpoint structure still exists.
216
This is the first JES2 system started and the checkpoint structure has been deleted; for
example, the CF has been powered off.
There is an existing active JES2 system.
In a MAS configuration, this would be the normal JES2 start configuration.
Warm start on the first JES2 with an active structure

If the CF has not been deactivated and the CKPT1 is accessible at restart, there are no
unusual operator dialogs. In Figure 10-15, notice how JES2 connects to the JESXCF group
and to the checkpoint structure.
S JES2,PARM='WARM,NOREQ
. . .
IXZ0001I CONNECTION TO JESXCF COMPONENT ESTABLISHED, 294
GROUP XCFJES2A MEMBER N1$#@$1
. . .
LAST WRITTEN FRIDAY, 6 JUL 2007 AT 00:54:10 (GMT)
IEF196I IEF237I 1D0B ALLOCATED TO $#@$#Q1
$HASP492 JES2 MEMBER-#@$1 QUICK START HAS COMPLETED
1
2
3
Figure 10-15 JES2 warm start - first system to join
1 JES2 is connecting to XCF.

2 JES2 connects to the CF checkpoint structure.
3 The time the last checkpoint update occurred.
Warm start on a second JES2 system with an active structure

The normal JES2 configuration is for multiple JES2 systems sharing a single MAS. In most
cases when JES2 is started, there will already be an active JES2 system in the JESplex.
Thus, the messages in Figure 10-16 on page 218 would be considered the normal JES2
restart sequence.
217
S JES2,PARM='WARM,NOREQ'
. . .
. . .
IEF196I IEF237I 1D0B ALLOCATED TO $#@$#Q1
$HASP492 JES2 MEMBER-#@$1 QUICK START HAS COMPLETED
1
2
3
Figure 10-16 JES2 warm start - second system to join

2 JES2 connects to the CF checkpoint structure.
3 JES2 reads the checkpoint data.
Warm start on the first JES2 with no active structure

This scenario is more complex. The checkpoint data in the CF is not available, for example
following a power outage in which all CF structures were lost. In this situation, CKPT1 is lost
and only the DASD copy (CKPT2) is available.
Note: If this were a scheduled outage, it would be good practice to use the checkpoint
reconfiguration dialog to suspend the use of the CF checkpoint structure.
In Figure 10-17 on page 219, JES2 discovers that the CKPT1 structure is empty. z/OS then
reallocates an empty structure, which JES2 attempts to connect to. JES2 asks the operator to
confirm the parameters in use or to change them. The values are correct and the operator
continues with initialization. Because JES2 cannot read CKPT1, it reads CKPT2 and the
warm start continues normally because CKPT2 is a duplexed copy of CKPT1.
218
. . .
< IXC582I Messages indicating a new structure has been allocated >
. . .
$HASP290 MEMBER #@$2 -- JES2 CKPT1 IXLLIST READ_LIST REQUEST FAILURE
RETURN CODE = 00000008
REASON CODE = 0C1C0825
RECORD
= CHECK
$HASP460 UNABLE TO CONFIRM THAT CKPT1 IS A VALID CHECKPOINT 585
DATA SET DUE TO AN I/O ERROR READING THE LOCK RECORD.
VERIFY THE SPECIFIED CHECKPOINT DATA SETS ARE CORRECT:
2
3
VALUES FROM CKPTDEF

CKPT1=(STRNAME=JES2CKPT_#@$1_1,INUSE=YES),
CKPT2=(DSNAME=SYS1.#@$2.CKPT2,VOLSER=#@$#Q1,INUSE=YES)
VALUES JES2 WILL USE
CKPT1=(STRNAME=JES2CKPT_#@$1_1,INUSE=YES),
CKPT2=(DSNAME=SYS1.#@$2.CKPT2,VOLSER=#@$#Q1,INUSE=YES)
DATA SET JES2 COULD NOT CONFIRM
CKPT1=(STRNAME=JES2CKPT_#@$1_1,INUSE=YES)
076 $HASP417 ARE THE VALUES JES2 WILL USE CORRECT? ('Y' OR 'N')
R 76,Y
(SYS1.#@$2.CKPT2 ON #@$#Q1)
$HASP493 JES2 ALL-MEMBER WARM START IS IN PROGRESS
$HASP850 1500 TRACK GROUPS ON #@$#Q1
$HASP851
96224 TOTAL TRACK GROUPS MAY BE ADDED
$HASP492 JES2 ALL-MEMBER WARM START HAS COMPLETED
5
6
Figure 10-17 JES2 warm start missing CF structure

2 The messages indicating a new CF structure has been allocated has been removed for
clarity.
3 JES2 has an error attempting to read the checkpoint from the CF structure.
4 JES2 is unable to confirm that the definitions are correct because it cannot read the primary
checkpoint.
5 JES2 prompts an operator to confirm the values are correct.
6 JES2 reads the checkpoint information from the secondary checkpoint on DASD.
10.4.3 JES2 hot start

A JES2 hot start is the result of following the warm start process after an abnormal JES2
shutdown ($PJES2,ABEND). Refer to 10.5.3, Abend shutdown on any JES2 in a Parallel
Sysplex MAS on page 222 for more information about JES2 abnormal shutdowns.
219
GROUP XCFJES2A MEMBER TRAINER$#@$2
IXL014I IXLCONN REQUEST FOR STRUCTURE JES2CKPT_1 442
(STRNAME JES2CKPT_1)
$HASP493 JES2 MEMBER-#@$2 HOT START IS IN PROGRESS
1
2
Figure 10-18 JES2 hot start
1 JES2 connects to XCF.

2 JES2 connects to the checkpoint structure.
2 This message indicates this is a hot start.
10.5 JES2 subsystem shutdown

Regardless of whether it is a single system MAS or a multi-system MAS, and whether the
JES2 checkpoint is on a DASD volume or in a structure, JES2 is brought down the same way.
The messages are only slightly different, indicative of where the checkpoint is located. In a
multi-system MAS, JES2 will continue to process work and update the spool until all JES2
systems in the MAS are down.
The JESXCF messages that are issued when JES2 is abended are different from the
messages that are issued when it is cleanly stopped. However, as shown in the following
examples, the messages are the same whether or not it is the last JES2 MAS member being
closed.
The recommended setup for a Parallel Sysplex, as used in the following examples, is CKPT1
defined as a structure in CF and duplexed to CKPT2, which is defined on DASD.
10.5.1 Clean shutdown on any JES2 in a Parallel Sysplex

A clean shutdown on any but the last JES2 in a MAS produces the following message flow in
a Parallel Sysplex.
Prior to z/OS 1.8, when attempting to stop JES2, it was not uncommon for the operator to
receive messages like those shown in Figure 10-19. The operator would then need to
determine the non-drained resource or abend JES2.
$PJES2
$HASP623 MEMBER DRAINING
$HASP607 JES2 NOT DORMANT -- MEMBER DRAINING 39 (GMT) 1
Figure 10-19 JES2 draining - z/OS 1.7
With z/OS 1.8 and later, however, IBM added an enhancement that can be seen in
220
The improved display shows which address spaces are holding out JES2. The expected
behavior of JES2 for a clean shutdown can be seen in Figure 10-21.
$PJES2
$HASP608
$HASP608
$HASP608
$HASP608
$HASP608
$HASP623
$HASP607
$PJES2 906
ACTIVE ADDRESS SPACES
ASID
JOBNAME JOBID
-------- -------- -------0028
ZFS
STC10221
MEMBER DRAINING
JES2 NOT DORMANT -- MEMBER DRAINING,
RC=10 ACTIVE ADDRESS SPACES
Figure 10-20 JES2 draining - z/OS 1.8
1 JES2 still has some active processes.

2 JES2 still has some active address spaces.
3 A list of active processes that need to be terminated is supplied.
Message IXZ0002I indicates that the JESXCF address space connection has been disabled.
This means JES2 has left the JESPLEX. Message $HASP9085 indicates that the JES2MON
function has also terminated.
$HASP099 ALL AVAILABLE FUNCTIONS COMPLETE
$PJES2
$HASP608 $PJES2 COMMAND ACCEPTED
IXZ0002I CONNECTION TO JESXCF COMPONENT DISABLED,
$HASP9085 JES2 MONITOR ADDRESS SPACE STOPPED FOR JES2
$HASP085 JES2 TERMINATION COMPLETE
1
2
Figure 10-21 JES2 - clean shutdown
1 The JES2 connection to XCF is disabled.

2 The JES2 Monitor task finishes.
3 JES2 terminates.
10.5.2 Clean shutdown of the last JES2 in a Parallel Sysplex

A clean shutdown of the last JES2 in a MAS produces the same message flow as a clean
stop of any JES2 MAS member in a Parallel Sysplex. The significant message is again
IXZ0002I, as shown in Figure 10-22 on page 222. It indicates the JESXCF address space
connection is disabled.
When JES2 abnormally terminates, as seen in Figure 10-23 on page 223, the JESXCF
component is broken.
221
04.17.25 #@$2
$PJES2
04.17.25 #@$2
$HASP608 $PJES2 COMMAND ACCEPTED
04.17.25 #@$2
*CNZ4201E SYSLOG HAS FAILED
04.17.25 #@$2
IEE043I A SYSTEM LOG DATA SET HAS BEEN QUEUED TO
SYSOUT CLASS L
04.17.25 #@$2
*IEE037D LOG NOT ACTIVE
04.17.26 #@$2
IXZ0002I CONNECTION TO JESXCF COMPONENT DISABLED, 1
04.17.26 #@$2
$HASP9085 JES2 MONITOR ADDRESS SPACE STOPPED FOR
JES2
04.17.31 #@$2
Figure 10-22 $PJES2, last system in the jesplex
1 The JESXCF is connection is stopped.
10.5.3 Abend shutdown on any JES2 in a Parallel Sysplex MAS

An abend shutdown on any JES2 system, or a catastrophic failure of JES2 in a MAS,
produces the following message flow in a Parallel Sysplex.
The significant message is IXZ0003I 1, as seen in Figure 10-23 on page 223, indicating the
connection to the JESXCF address space is broken.
222
01.20.17 #@$2
$PJES2,ABEND
*01.20.17 #@$2
*$HASP095 JES2 CATASTROPHIC ERROR. CODE = $PJ2
01.20.18 #@$2
$HASP088 JES2 ABEND ANALYSIS
$HASP088 -----------------------------------------------------$HASP088 FMID = HJE7730
LOAD MODULE = HASJES20
$HASP088 SUBSYS = JES2 z/OS 1.8
$HASP088 DATE = 2007.186
TIME
= 1.20.18
$HASP088 DESC = OPERATOR ISSUED $PJES2, ABEND
$HASP088 MODULE
MODULE
OFFSET SERVICE ROUTINE
EXIT
$HASP088 NAME
BASE
+ OF CALL LEVEL
CALLED
##
$HASP088 -------- -------------- ------- ---------- ---$HASP088 HASPCOMM 000127D8 + 0081C8 OA18916 *ERROR $PJ2
$HASP088 PCE = COMM
(0C9B55E0)
$HASP088 R0 = 0001A642 00C7E518 00006F16 00016FC8
$HASP088 R4 = 00000000 0C9B5D84 00000004 0C9B5D88
$HASP088 R8 = 0001A642 00000000 00000000 00007000
$HASP088 R12 = 00012828 0C9B55E0 0C9750B8 0002D380
$HASP088 -----------------------------------------------------*01.20.18 #@$2
*$HASP198 REPLY TO $HASP098 WITH ONE OF THE
* FOLLOWING:
* END
- STANDARD ABNORMAL END
* END,DUMP
- END JES2 WITH A DUMP (WITH AN OPTIONAL TITLE)
* END,NOHOTSTART - ABBREVIATED ABNORMAL END (HOT-START IS AT RISK)
* SNAP
- RE-DISPLAY $HASP088
* DUMP
- REQUEST SYSTEM DUMP (WITH AN OPTIONAL TITLE)
*01.20.18 #@$2
*009 $HASP098 ENTER TERMINATION OPTION
IST314I END
01.21.30 #@$2
01.21.30 #@$2
01.21.31 #@$2
01.21.31 #@$2
REASON=D7D1F240
01.21.31 #@$2
GROUP XCFJES2A
9end
IEE600I REPLY TO 009 IS;END
IEF450I JES2 JES2 - ABEND=S02D U0000
IXZ0003I CONNECTION TO JESXCF COMPONENT BROKEN
MEMBER TRAINER$#@$2
Figure 10-23 JES2 abend - shut down any system in the sysplex
1 The JES2 connection to XCF is broken and stopped.
10.6 JES2 batch management in a MAS

Batch workloads generally run under the control of the job entry subsystem, either JES2 or
JES3. The way that jobs are submitted can vary greatly, but after a job has started to run
under the control of an initiator, it cannot be moved from one image to another. Because of
this, any workload-balancing actions have to be taken before the job starts to run.
There are two types of JES2 initiators, JES controlled initiators and WLM initiators. If JES2 is
not configured to use WLM initiators, there are two ways of controlling where a job runs:
Job class
Either through coding on the JCL or through the use of JES exits, a job will be assigned a
class in which it will run. (It is possible to change the assigned class via operator
command.)
223
For a job to start execution, there must be an initiator available that has been started and
can accept work in that class. The initiator does not need to be on the system where the
job was first read onto the JES queue. Furthermore, if more than one system has initiators
started for that class, you cannot control which system will execute the job through the use
of class alone.
System affinity
Again, either through explicit coding in JCL or through a JES exit, it is possible to assign a
specific system affinity to a job. Additionally, it is possible to assign an affinity to the JES2
internal reader. This affinity can be altered by using the command $TINTRDR,SYSAFF=. For
example, when we issue $TINTRDR,SYSAFF=#@$2 on system #@$3, then all jobs submitted
on #@$3 ran on system #@$2.
The same technique can also be applied to local readers by using the
$T RDR(nn),SYSAFF= command. The affinity will ensure that the job will only be executed
on a specific system. This does not guarantee, however, that an initiator is available to
process the job in its assigned class.
Through the controlled use of classes and system affinity, you can determine where a job will
be executed. You can let JES2 manage where the job will run by having all initiator classes
started on all members of the MAS. The scheduling environment, which is discussed in
11.11, Using the SCHEDULING ENVIRONMENT (SE) command on page 248, can also be
used to control where jobs run.
If you want to cancel a job, restart a job, or send a message, you can do so from any member
of the MAS. It is not necessary to know which member the job is running on. We see this in
Figure 10-24.
#@$3
#@$2
-$CJ(12645)
$HASP890 JOB(TESTJOB1)
$HASP890 JOB(TESTJOB1)
$HASP890
$HASP890
1
STATUS=(EXECUTING/#@$2),CLASS=A,
PRIORITY=9,SYSAFF=(#@$2),HOLD=(NONE),
CANCEL=YES
CANCEL
TESTJOB1,A=002B
IEE301I TESTJOB1
CANCEL COMMAND ACCEPTED
IEA989I SLIP TRAP ID=X222 MATCHED. JOBNAME=TESTJOB1, ASID=002B
IEF450I TESTJOB1 STEP3 - ABEND=S222 U0000 REASON=00000000
3
4
Figure 10-24 JES2 cancel a job on another system
1
2
3
4
The $CJ is issued on system #@$3.

The job, TESTJOB1, is executing on system #@$2.
System #@$2 issues a cancel command on behalf of JES2 on system #@$3.
The job TESTJOB1 finishes with a system abend S522.
The $C command for a TSO user is converted into a C U=xxxx command. It does not matter
which system the TSO user is logged onto; the cancel command is routed to the appropriate
system. JES2 cannot cancel an STC. As a result, it still does not matter which system the
STC is running on, because any attempt to issue a JES2 cancel will fail.
For more information about batch management, refer to Getting the Most Out of a Parallel
Sysplex, SG24-2073. For more information about JES2 commands, refer to z/OS JES2
Commands, SA22-7526.
224
10.7 JES2 and Workload Manager

There are two types of initiators:
JES2-managed initiators
WLM-managed initiators
Who controls the initiators is determined by the job class through the MODE=type parameter
on the JES2 JOBCLASS initialization statement.
Note: Batch jobs that are part of a critical path should remain in JES-managed job classes.
This gives more control of the batch jobs.
10.7.1 WLM batch initiators

WLM initiators are controlled dynamically by Workload Manager (WLM). They run under the
master subsystem. WLM adjusts the number of initiators on each system based on:
The queue of jobs awaiting execution in WLM-managed classes

The performance goals and relative importance of the work
The success of meeting these performance goals
The capacity of each system to do the work
You can switch initiators from one mode to another by using the $TJOBCLASS command with
the MODE= parameter. However, ensure that all jobs with the same service class are
managed by the same type of initiator. For example, assume that job classes A and B are
assigned to the HOTBATCH service class. If JOBCLASS(A) is controlled by WLM, and
JOBCLASS(B) is controlled by JES2, then WLM will find it difficult to manage the
HOTBATCH goals without managing class B jobs.
Unlike JES2 initiators, WLM initiators do not share classes. Also, the number of WLM
initiators is not limited. Using $TJOBCLASS, however, you can limit the number of concurrent
WLM jobs in a particular class.
10.7.2 Displaying batch initiators

To display the batch initiators, enter the $DJOBCLASS(*) command as shown in Figure 10-25.
You can also issue $DJOBCLASS(*),TYPE=WLM or $DJOBCLASS(*),TYPE=JES. As an alternative,
you can use the SDSF JC command to display job classes.
$DJOBCLASS(*)
. . .
$HASP837 JOBCLASS(K)
$HASP837 JOBCLASS(K)
$HASP837
$HASP837 JOBCLASS(L)
$HASP837
. . .
MODE=JES,QHELD=NO,SCHENV=,
XEQCOUNT=(MAXIMUM=*,CURRENT=0)
1
2
MODE=WLM,QHELD=NO,SCHENV=,
XEQCOUNT=(MAXIMUM=*,CURRENT=5)
3
4
Figure 10-25 $DJOBCLASS(*)
1 K class initiators are JES2-managed.

2 There are no K class jobs currently running.
225
3 L class initiators are WLM-managed.

4 There are five L class jobs running.
10.7.3 Controlling WLM batch initiators

WLM automatically controls its initiators, but there are JES2 commands that can be used to
manage them:
Limit the number of jobs in each job class

Stop or start initiators on an individual system
Control the system affinity of a job
Immediately start a job
Limit the number of jobs in each job class

To limit the number of jobs in each class that can run at the same time in a MAS, use the
$TJOBCLASS command. In Figure 10-26 we set the limit to five. This parameter is not valid for a
JES2-managed initiator class.
$T JOBCLASS(L),XEQC=(MAX=5)
$HASP837
MODE=WLM,QHELD=NO,SCHENV=,
XEQCOUNT=(MAXIMUM=5,CURRENT=0) 1
Figure 10-26 $T JOBCLASS - set maximum WLM initiators
Stop or start initiators on an individual system ($PXEQ $SXEQ)

The $PXEQ command will stop new work from being selected. The related command $SXEQ
enables new work to be selected.
Note: Both $SXEQ and $PXEQ commands affect both JES2 and WLM initiators, but they only
affect a single system.
$PXEQ
$HASP000 OK
$HASP222 XEQ DRAINING
$HASP000 OK
$S XEQ
$HASP000 OK
Figure 10-27 $PXEQ and $SXEQ
Control the system affinity of a job

To control the system affinity of a job, use the $TJOBnnn,SYSAFF=member command.
Figure 10-28 on page 227 shows a job without any SYSAFF. We then assign a SYSAF and
finally show that it now has a SYSAFF assigned. SDSF can also be used to assign a SYSAFF
to a job.
226
$DJ 12789
$HASP890 JOB(TE$TJOB1)
$HASP890
. . .
$TJ(12789),S=#@$1
$HASP890
. . .
$DJ 12789
$HASP890
STATUS=(AWAITING EXECUTION),CLASS=L,
PRIORITY=9,SYSAFF=(ANY),HOLD=(JOB)
PRIORITY=9,SYSAFF=(#@$1),HOLD=(JOB)
PRIORITY=9,SYSAFF=(#@$1),HOLD=(JOB)
Figure 10-28 $TJ to assign SYSAFF
Immediately start a job

To immediately start the WLM job, use the $SJnnnn command. Figure 10-29 shows a job
being released and immediately started. There is no corresponding command for a
JES2-managed jobclass.
$SJ(12789)
$HASP890 JOB(TE$TJOB1) STATUS=(AWAITING EXECUTION),CLASS=L,
$HASP890
PRIORITY=9,SYSAFF=(#@$1),HOLD=(NONE)
IWM034I PROCEDURE INIT STARTED FOR SUBSYSTEM JES2
APPLICATION ENVIRONMENT SYSBATCH
PARAMETERS SUB=MSTR
Figure 10-29 $SJ immediately start a WLM-managed job
10.8 JES2 monitor

This JES2 monitor is not intended to be a performance monitor. However, it does provide
useful information that you can use in performance analysis. All health monitor commands
have the JES2 command prefix, which by default is the dollar sign ($) character followed by
the letter J.
All commands that have the JES2 command prefix followed by a J are sent to the monitor
command subtask. If the monitor does not recognize the command, it is routed to the JES2
address space for normal command processing.
The available commands are:
$JDSTATUS
Displays current status of JES2; see Figure 10-30 on page 228.
$JDJES
Displays information about JES2; see Figure 10-31 on page 228.
$JDMONITOR
Displays monitor task and module status information; see

$JDDETAILS
Displays detailed information about JES2 resources, sampling, and

MVS waits; see Figure 10-33 on page 229.
227
$JDHISTORY
Displays history information. Use caution with this command because

it provides a significant amount of spooled output; see Figure 10-34 on
page 229.
$JSTOP
Stops the monitor (JES2 restarts it automatically within few minutes);

see Figure 10-35 on page 230.
$JDSTATUS
$HASP9120 D STATUS
$HASP9121 NO OUTSTANDING ALERTS
$HASP9150 NO JES2 NOTICES
1
2
3
Figure 10-30 $JDSTATUS
1 Issue a command to the JES2 Monitor to display the information about JES2.
2 No outstanding JES2 alerts.
3 No outstanding JES2 notices.
$JDJES
$HASP9120
$HASP9121
$HASP9122
$HASP9150
1
D JES
NO OUTSTANDING ALERTS
NO INCIDENTS BEING TRACKED
NO JES2 NOTICES
Figure 10-31 $JDJES
1 Issue a command to the JES2 Monitor to display the status of JES.

2 No alerts, incidents being tracked, or JES2 notices.
$JDMONITOR
$HASP9100 D MONITOR
NAME
STATUS
ALERTS
-------- ------------ -----------------------MAINTASK ACTIVE
SAMPLER ACTIVE
COMMANDS ACTIVE
PROBE
ACTIVE
$HASP9102 MONITOR MODULE INFORMATION
NAME
ADDRESS LENGTH ASSEMBLY DATE LASTAPAR
-------- -------- -------- -------------- -------HASJMON 0C965000 000010B8 03/29/06 14.57 NONE
HASJSPLR 0C96A388 00002DA0 03/29/06 14.53 NONE
HASJCMDS 0C9660B8 000042D0 03/19/07 21.47 OA20195
1
2
LASTPTF
-------NONE
NONE
UA33267
Figure 10-32 $JDMONITOR
1 Issue the Display Monitor command.

2 List each of the processes and its status.
3 List the major programs and their maintenance level.
228
$JDDETAILS
$HASP9103 D DETAIL
$HASP9104 JES2 RESOURCE USAGE SINCE 2007.183 23:00:01
RESOURCE
LIMIT
USAGE
LOW
HIGH AVERAGE
-------- -------- -------- -------- -------- -------BERT
65620
341
341
345
342
BSCB
0
0 7483647
0
0
BUFX
89
0
0
2
0
CKVR
2
0
0
1
0
CMBS
201
0
0
0
0
CMDS
200
0
0
0
0
ICES
33
0
0
0
0
JNUM
32760
2025
2016
2025
2020
JOES
20000
2179
2166
2179
2172
JQES
32760
2025
2016
2025
2020
LBUF
23
0
0
0
0
NHBS
53
0
0
0
0
SMFB
52
0
0
0
0
TBUF
104
0
0
0
0
TGS
9911
6062
6052
6063
6056
TTAB
3
0
0
0
0
VTMB
10
0
0
0
0
$HASP9105 JES2 SAMPLING STATISTICS SINCE 2007.183 23:00:01
TYPE
COUNT PERCENT
---------------- ------ ------ACTIVE
493
0.83
IDLE
58740
99.16
LOCAL LOCK
0
0.00
NON-DISPATCHABLE
0
0.00
PAGING
0
0.00
OTHER WAITS
0
0.00
TOTAL SAMPLES
59233
. . .
Figure 10-33 $JDDETAILS
The $JDDETAILS command causes the monitor to display all the JES2 resources and their
limits. This is similar to the information seen with SDSF using the RM feature. Refer to 11.7,
Resource monitor (RM) command on page 246, for more detailed information.
$JDHISTORY
$HASP9130 D HISTORY
$HASP9131 JES2 BERT
USAGE HISTORY
DATE
TIME
LIMIT
USAGE
LOW
HIGH AVERAGE
-------- -------- -------- -------- -------- -------- -------2007.184 0:00:00
65620
342
341
342
341
2007.183 23:00:01
65620
341
341
345
342
2007.183 22:00:00
65620
343
343
344
343
. . .
Figure 10-34 $JDHISTORY
The $JDHISTORY command displays a history for all JES2 control blocks, that is, BERTs,
BSCBs, and so on.
229
$JSTOP
$HASP9101 MONITOR STOPPING 753
IEA989I SLIP TRAP ID=X13E MATCHED. JOBNAME=JES2MON , ASID=0020.
$HASP9085 JES2 MONITOR ADDRESS SPACE STOPPED FOR JES2
. . .
IEF196I IEF375I JOB/JES2MON /START 2007182.1936
IEF196I IEF376I JOB/JES2MON /STOP 2007184.0035 CPU
0MIN 47.69SEC
IEF196I SRB
0MIN 33.21SEC
IRR812I PROFILE ** (G) IN THE STARTED CLASS WAS USED 771
TO START JES2MON WITH JOBNAME JES2MON.
. . .
Figure 10-35 $JSTOP - JES2MON restarts
1 Stop the JES2 monitor.

2 Message indicating JES2 monitor has stopped.
3 The JES2 monitor gets automatically restarted by JES2.
230
1
2
3
11
Chapter 11.
System Display and Search

Facility and OPERLOG
This chapter discusses the IBM System Display and Search Facility (SDSF) and OPERLOG.
SDSF, which is an optional feature of z/OS, provides you with information that enables you to
monitor, manage, and control a z/OS JES2 system.
This chapter discusses how to use SDSF to perform the following tasks:
View the system log and operlog
Issue system commands view the result
Print or save job output and the system log
Look at the running jobs
Display JES2 resources
For more information about SDSF, refer to SDSF Operation and Customization, SA22-7670.
231
11.1 Introduction to System Display and Search Facility

The IBM System Display and Search Facility (SDSF) provides you with an easy and efficient
way to monitor, manage, and control your z/OS JES2 system. You can:
Control job processing (hold, release, cancel, requeue, and purge jobs)
Save and print all or part of the syslog or a jobs output
Control devices (printers, lines, and initiators) across the Multi-Access Spool (MAS)
Browse the syslog
Edit the JCL and resubmit a job without needing access to the source JCL
Manage system resources, such as members of the MAS, job classes, and WLM enclaves
Monitor and control the IBM Health Checker for z/OS checker
With SDSF panels, there is no need to learn or remember complex command syntax. SDSF's
action characters, overtype able fields, action bar, pull-downs, and pop-up windows allow you
to select available functions.
Figure 11-1 shows the primary SDSF panel.
Display Filter View Print Options Help
------------------------------------------------------------------------------HQX7730 ----------------- SDSF PRIMARY OPTION MENU -------------------------COMMAND INPUT ===>
SCROLL ===> CSR
DA
I
O
H
ST
Active users
Input queue
Output queue
Held output queue
Status of jobs
LOG
SR
MAS
JC
SE
RES
ENC
PS
System log
System requests
Members in the MAS
Job classes
Scheduling environments
WLM resources
Enclaves
Processes
END
Exit SDSF
INIT
PR
PUN
RDR
LINE
NODE
SO
SP
Initiators
Printers
Punches
Reader
Lines
Nodes
Spool offload
Spool volumes
RM
CK
Resource monitor
Health checker
ULOG
User session log
Licensed Materials - Property of IBM

5694-A01 (C) Copyright IBM Corp. 1981, 2006. All rights reserved.
US Government Users Restricted Rights - Use, duplication or
disclosure restricted by GSA ADP Schedule Contract with IBM Corp.
Figure 11-1 Primary SDSF panel
232
11.2 Using the LOG command

You can use the LOG command to access either SYSLOG or OPERLOG panels to view the
z/OS system log in chronological order.
SYSLOG displays the z/OS system log data sorted by date and time. OPERLOG displays the
merged, sysplex-wide system message log managed by the System Logger that is an
alternative to the JES2 spool used for the system log. The OPERLOG is also sorted by date
and time.
You can see outstanding write-to-operator-with-reply messages (WTORs) at the bottom of
both logs.
LOG command options:
LOG SYSLOG (or LOG S) displays the SYSLOG panel.
LOG OPER (or LOG O) displays the OPERLOG panel.
LOG with no parameters displays the default log panel for an individual system.
Use the SET LOG command to specify the default panel that is displayed when you enter the
LOG command with no parameters. You can use this command in any SDSF panel (except
help and tutorial panels).
SET LOG command options:
OPERACT (or A) specifies that the OPERLOG panel is displayed if OPERLOG is active on
the system you are logged on to; otherwise, the SYSLOG panel is displayed.
OPERLOG (or O) specifies that the OPERLOG panel is displayed.
SYSLOG (or S) specifies that the SYSLOG panel is displayed.
? displays the current setting for SET LOG.
SET LOG with no parameters is the same as SET LOG OPERACT.
Tip: The BOTTOM, TOP, NEXT and PREV commands run faster when preceded by the FINDLIM,
LOGLIM and FILTER commands.
Use FILTER to limit the data displayed on the OPERLOG panel.
Use LOGLIM to limit the amount of OPERLOG data that SDSF will search for records
that meet filter criteria. The limit is between 0 (no limit) and 999 hours.
Use FINDLIM to limit the amount of OPERLOG, SYSLOG and ULOG data that SDSF
will search for records that meet filter criteria. The limit is any number between 1000
and 9999999.
For more information about FILTER, LOGLIM and FINDLIM commands, refer to SDSF
Operation and Customization, SA22-7670.
11.2.1 Example of the SYSLOG panel

If you enter LOG S in a command line, the SYSLOG panel is displayed, as shown in
Chapter 11. System Display and Search Facility and OPERLOG
233

------------------------------------------------------------------------------1
2
3
SDSF SYSLOG 7609.101 #@$2 #@$3 06/27/2007 0W
7889
COLUMNS
1 80
COMMAND INPUT ===>
SCROLL ===> CSR
4
5
6
7
8
4000000 #@$2
2007178 02:29:02.17
00000080 IEE366I NO SMF DATA SE
0020000 #@$2
2007178 02:29:02.60 STC07872 00000290 IEF695I START SMFCLR
, GROUP SYS1
4000000 #@$2
2007178 02:29:02.60 STC07872 00000090 $HASP373 SMFCLR STAR
0000000 #@$2
2007178 02:29:11.91
00000280 IEA989I SLIP TRAP ID=X
0004000 #@$2
2007178 02:29:25.58 STC07872 00000290 . . .
Figure 11-2 Example SYSLOG Panel
Useful fields in the syslog panel include:

1 JES2 sysid of the current log.
2 JES2 sysid of the system you are logged onto.
3 Date of SYSLOG data set.
4 Originating system name.
5 Date when message was logged.
6 Time when message was logged.
7 Job identifier, console name, or multiline ID.
8 Text of message.
In our case, although we were logged onto system #@$3, by issuing the command SYSID
#@$2 we were able to view the syslog from system #@$2.
Note: Using the status command with a job prefix of SYSLOG* displays all the active
syslog and any syslog that has been spun off using the WRITE LOG command but has not
yet been archived. Figure 11-3 illustrates such an example.

------------------------------------------------------------------------------ISFPCU41 US DISPLAY ALL CLASSES
LINE 1-38 (48)
COMMAND INPUT ===>
SCROLL ===> CSR
NP JOBNAME JobID
Owner
Prty Queue
C Pos ASys ISys PrtDest
SYSLOG STC07609 +MASTER+ 15 EXECUTION
#@$2 #@$2 LOCAL
#@$1 #@$1 LOCAL
#@$3 #@$3 LOCAL
SYSLOG STC01121 +MASTER+
1 PRINT
2
#@$3 LOCAL
1 PRINT
3
#@$3 LOCAL
1 PRINT
4
#@$3 LOCAL
1 PRINT
5
#@$2 LOCAL
. . .
Figure 11-3 SYSLOG* ST Output
If syslog is not being regularly offloaded, or if a runaway task creates excessive syslog
messages, then SDSF may not be able to view the log. This will be indicated by the error
message ISF002I MASTER SDSF SYSLOG INDEX FULL when the SDSF LOG command is issued.
234
There are two causes of this message:

An error in your ISPF profile - this can be corrected by deleting the ISFPROF member of
your ISPF profile.
Excessive SYSLOG entries being kept in the spool - in this case, your systems
programmer will need to archive some of the syslog entries.
11.2.2 Example of the OPERLOG panel

If you enter LOG O in a command line, the OPERLOG panel is displayed, as shown in
Figure 11-4. Notice that there are messages from each system in the combined operlog
output.
------------------------------------------------------------------------------1
2
SDSF OPERLOG DATE 06/27/2007
0 WTORS
COLUMNS 02- 81
COMMAND INPUT ===>
SCROLL ===> CSR
3
4
5
6
7
8 9
NC0000000 #@$1
2007178 02:50:37.26 MACNIV
00000290 D XCF,S,ALL
MR0000000 #@$1
2007178 02:50:37.30 MACNIV
00000080 IXC335I 02.50.37 DIS
LR
253 00000080 SYSTEM
TYPE SERIAL L
DR
253 00000080 #@$2
2084 6A3A
N
DR
253 00000080 #@$3
2084 6A3A
N
ER
253 00000080 #@$1
2084 6A3A
N
N 4000000 #@$2
2007178 02:51:01.27 STC07638 00000090 ERB101I ZZ : REPORT AV
N 4000000 #@$3
N 4000000 #@$1
. . .
Figure 11-4 Example OPERLOG Panel
1 Date of OPERLOG data set.

2 The number of outstanding WTORs.
3 Record type and request type.
4 First 28 routing codes.
5 Originating system name.
6 Date when message was logged.
7 Job identifier, console name, or multiline ID.
8 User exit flags.
9 Text of message.
11.3 Using the ULOG command

Use the ULOG command to display all z/OS and JES2 commands and responses (including
commands generated by SDSF) that you issued during your session. The log is deleted when
you end the SDSF session or when you issue a ULOG CLOSE command. If you have issued the
ULOG CLOSE command, you will need to issue ULOG to re-enable the ULOG feature.
11.3.1 Example of the ULOG panel

If you enter ULOG in a command line, the ULOG panel is displayed, as shown in Figure 11-5 on
page 236.
235

------------------------------------------------------------------------------1
SDSF ULOG CONSOLE COTTRELC
LINE 0
COLUMNS 02- 81
COMMAND INPUT ===>
SCROLL ===> CSR
********************************* TOP OF DATA **********************************
2
3
4
5
#@$2
2007178 02:16:03.95
ISF031I CONSOLE COTTREL ACTIVATED
#@$2
2007178 02:16:09.90
-D XCF,STR
#@$2
2007178 02:16:10.13
STRNAME
ALLOCATION TIME
CIC_DFHLOG_001
06/27/2007 01:08:52
CIC_DFHSHUNT_001 06/27/2007 01:08:53
CIC_GENERAL_001
--. . .
Figure 11-5 Example ULOG
1 Extended console name. This will be NOT ACTIVE if the console was turned off with the ULOG
CLOSE command, or you are not authorized to use it.
2 System name on which the command was issued, or from which the response originated.
3 Date when the message was logged.
4 Job ID applying to the message, if available.
5 Command text or message response. If it is echoed by SDSF, it is preceded by a
hyphen (-).
Use the PRINT command to save the ULOG data. You can route it to a printer or save it in a
data set. You will need to save the ULOG data before exiting SDSF. The PRINT command is
described in 11.5, Printing and saving output in SDSF on page 239.
The ULOG command creates an EMCS console, and the default name for the console is your
user ID. Each EMCS console in a sysplex requires a unique name. Thus, if you attempt to
open a second ULOG screen, for example by having two SDSF sessions within a single ISPF
session you receive the message ISF031I, as seen in 2 Figure 11-6.
------------------------------------------------------------------------------1
SDSF ULOG CONSOLE COTTRELC (SHARED)
LINE 0
COLUMNS 42- 121
COMMAND INPUT ===>
SCROLL ===> CSR
********************************* TOP OF DATA **********************************
ISF031I CONSOLE COTTRELC ACTIVATED (SHARED)
2
******************************** BOTTOM OF DATA ********************************
Figure 11-6 Second ULOG session started
When this happens, SDSF will share an EMCS console 1. However, responses to commands
are always sent to the first SDSF console, not to the second (shared) console.
Figure 11-7 on page 237 shows a command issued. However, the output is displayed in the
primary SDSF ULOG panel, as seen in Figure 11-8 on page 237.
236

------------------------------------------------------------------------------SDSF ULOG CONSOLE COTTRELC (SHARED)
LINE 0
COLUMNS 42- 121
COMMAND INPUT ===>
SCROLL ===> CSR
********************************* TOP OF DATA **********************************
ISF031I CONSOLE COTTRELC ACTIVATED (SHARED)
-D T
1
******************************** BOTTOM OF DATA ********************************
Figure 11-7 Command issued on shared ULOG console
1 The command is issued on the shared console but no response is received.

------------------------------------------------------------------------------SDSF ULOG CONSOLE COTTRELC
LINE 0
COLUMNS 02- 81
COMMAND INPUT ===>
SCROLL ===> CSR
********************************* TOP OF DATA **********************************
#@$3
2007189 20:03:16.57
ISF031I CONSOLE COTTRELC ACTIVATED
#@$3
2007189 20:03:54.14 TSU11621 IEE136I LOCAL: TIME=20.03.54 DATE=200
******************************** BOTTOM OF DATA ********************************
Figure 11-8 Command output displayed on primary ULOG console
1 The response from the command is returned to this screen, even though the command was
not issued on this screen.
IBM has removed the restriction against having the same user ID logged onto two different
JES2 systems, although other restrictions remain, such as TSO enqueue and ISPF issues. If
you log on onto two systems and attempt to use the same EMCS console ID for both, then
the attempt to open the second ULOG will fail, as seen in 1 in Figure 11-9.
------------------------------------------------------------------------------SDSF ULOG CONSOLE NOT ACTIVE
LINE 0
COLUMNS 42- 121
COMMAND INPUT ===>
SCROLL ===> CSR
********************************* TOP OF DATA **********************************
ISF032I CONSOLE COTTRELC ACTIVATE FAILED, RETURN CODE 0004, REASON CODE 0000 1
******************************** BOTTOM OF DATA ********************************
Figure 11-9 ULOG on a second system
Commands can be issued from the second system but the output is only visible in the system
log. That is, you can view the results using the LOG command but not the ULOG command.
11.4 Using the DISPLAY ACTIVE (DA) command

The SDSF DA command shows information about z/OS active address spaces (jobs, started
tasks, initiators and TSO users) that are running in the sysplex. SDSF obtains the information
from the Resource Measurement Facility (RMF). If RMF Monitor I is not active, the DA
command will produce no output.
237
11.4.1 Example of the DISPLAY ACTIVE panel

If you enter DA on a command line, the DISPLAY ACTIVE panel is displayed, as shown in
Figure 11-10.
------------------------------------------------------------------------------1
2
3
4
ISFPCU41 @$2 (ALL)
PAG
0 CPU
9
LINE 23-60 (184)
COMMAND INPUT ===>
SCROLL ===> CSR
5
6
7
8
9
10
11
12
13
NP JOBNAME StepName CPU%
CPU-Time
SIO EXCP-Cnt Real Paging SysName
CONSOLE CONSOLE
0.00
11.34 0.00
5,101 2949 0.00 #@$2
CONSOLE CONSOLE
0.00
7.11 0.00
3,293 3309 0.00 #@$3
COTTREL IKJACCNT 0.19
4.15 0.00
21,801 1388 0.00 #@$2
D#$1DBM1 D#$1DBM1 0.00
5.34 0.00
5,196 8009 0.00 #@$1
D#$1IRLM D#$1IRLM 0.00
23.16 0.00
69 5956 0.00 #@$1
D#$1MSTR D#$1MSTR 0.00
62.20 0.24
24,616 2196 0.00 #@$1
D#$1SPAS D#$1SPAS 0.00
0.36 0.00
148 669 0.00 #@$1
D#$2DBM1 D#$2DBM1 0.00
3.25 0.00
7,678 2062 0.00 #@$2
D#$2DIST D#$2DIST 0.00
0.35 0.00
1,802 220 0.00 #@$2
D#$2IRLM D#$2IRLM 0.19
23.52 0.00
69 4953 0.00 #@$2
D#$2MSTR D#$2MSTR 0.32
65.28 0.24
22,272 512 0.00 #@$2
D#$2SPAS D#$2SPAS 0.00
0.40 0.00
148 131 0.00 #@$2
DELIGNY IKJACCNT 0.00
26.27 0.00
975 1938 0.00 #@$1
DEVMAN DEVMAN
0.00
0.12 0.00
53
70 0.00 #@$1
. . .
Figure 11-10 DISPLAY ACTIVE panel example
There are many different columns available and the actual display can be modified using the
ARRANGE command. Some of the more useful fields are:
1 System ID of system you are logged on to.
2 Systems displayed (z/OS value or SYSNAME value).
3 Total demand paging rate.
4 Percent of time that the CPU is busy (z/OS/LPAR/zAAP Views).
5 Where JES2 commands such as C are issued.
6 Jobname.
7 CPU% usage by each job.
8 Total CPU used by each job.
9 Current I/O Rate for each job.
10 Total IOs performed by each job.
11 Real memory used by each job.
12 Paging rate for each job.
13 System where this job is running.
Use the SYSNAME command to restrict the display to a particular system or to view all the
systems. SYSNAME <sysid> will restrict the display to one system. SYSNAME ALL will display the
active tasks on all systems.
To restrict the display to only batch jobs, use DA OJOB. To restrict the display to only STCs, use
DA OSTC. To restrict the display to only TSO user IDs, use DA OTSU.
The arrange command allows you reorder the columns displayed. Thus, to move the SIO and
EXCP-Cnt columns before the CPU% and CPU-TOTAL, issue the commands seen in Figures
Figure 11-11 on page 239 and Figure 11-12 on page 239.
238

------------------------------------------------------------------------------ISFPCU41 @$2 (ALL)
PAG
0 CPU
7
LINE 23-60 (184)
COMMAND INPUT ===> arrange SIO A STEPNAME
SCROLL ===> CSR 1
NP JOBNAME StepName CPU%
CPU-Time
SIO EXCP-Cnt Real Paging SysName
CONSOLE CONSOLE
0.00
12.12 0.00
5,427 3173 0.00 #@$2
CONSOLE CONSOLE
0.00
7.75 0.00
3,550 3435 0.00 #@$3
4.30 0.00
22,085 1268 0.00 #@$2
D#$1DBM1 D#$1DBM1 0.00
5.53 0.00
5,196 8012 0.00 #@$1
. . .
Figure 11-11 Arrange After command
1 Move the SIO column after the StepName column.

------------------------------------------------------------------------------ISFPCU41 @$2 (ALL)
PAG
0 CPU 77
LINE 23-60 (184)
COMMAND INPUT ===> arrange excp-cnt b cpu%
SCROLL ===> CSR
NP JOBNAME StepName
SIO
CPU%
CPU-Time EXCP-Cnt Real Paging SysName
CONSOLE CONSOLE
0.00
0.08
12.13
5,427 3174 0.00 #@$2
CONSOLE CONSOLE
0.05
0.05
7.76
3,554 3435 0.00 #@$3
0.08
4.31
22,085 1275 0.00 #@$2
D#$1DBM1 D#$1DBM1 0.00
0.12
5.54
5,196 8012 0.00 #@$1
. . .
Figure 11-12 Arrange Before command
1 Move the EXCP-Cnt column before the CPU% column.

------------------------------------------------------------------------------ISFPCU41 @$2 (ALL)
PAG
0 CPU
9
LINE 23-60 (184)
COMMAND INPUT ===>
SCROLL ===> CSR
NP JOBNAME StepName
SIO EXCP-Cnt CPU%
CPU-Time Real Paging SysName
CONSOLE CONSOLE
0.00
5,427 0.03
12.13 3175 0.00 #@$2
CONSOLE CONSOLE
0.00
3,554 0.01
7.76 3436 0.00 #@$3
22,085 0.03
4.31 1275 0.00 #@$2
D#$1DBM1 D#$1DBM1 0.00
5,196 0.08
5.54 8012 0.00 #@$1
. . .
Figure 11-13 Reordered DA display
The resulting of reordering the display is seen Figure 11-13.
11.5 Printing and saving output in SDSF

There are a number of ways to save or print output in SDSF including using:
PRINT MENU
PRINT Command
XDC command
239
11.5.1 Print menu

Using the PRINT menu is the easiest method to print or save output. Figure 11-14 shows the
initial screen, which you reach by putting the cursor on Print and pressing Enter.
Filter View Print Options Help
---------------------- -------------------------------ISFPCU41 OG 7252.101 |
1. Print open sysout... |
COMMAND INPUT ===>
|
2. Print open data set... |
N 0000000 #@$1
2007 |
3. Print open file...
|
NR4000000 #@$1
2007 |
4. Print...
|
N 0000000 #@$1
2007 |
*. Print close
|
N 0000000 #@$1
2007 |
6. Print screen with ISPF |
NR4000000 #@$1
2007 -------------------------------NR4000000 #@$1
2007177 22:16:17.76 STC07280 00000090
N 0004000 #@$1
2007177 22:16:17.79 STC07280 00000290
S
N 0004000 #@$1
2007177 22:16:17.79 STC07280 00000290
S
. . .
Display
---------------------COLUMNS
1 80
SCROLL ===> CSR
IEF196I IEF237I A943 A
ERB102I ZZ : TERMINATE
IEF196I IEF285I SYS1
IEF196I IEF285I VOL
ERB451I RMF: SMF DATA
ERB102I RMF: TERMINATE
--------P
-JOBNAME STEPNAME PRO
CLOCK SERV PG
PAG
Figure 11-14 SDSF PRINT menu screen
You can use the different options to print to a data set, to a pre-allocated DD statement, and
then specify which lines are to be printed.
11.5.2 Print command

Although the SDSF PRINT menu is simpler to use, the PRINT command can be quicker, and
the XDC command offers more options. The PRINT command can be used to save or to print,
all or part of the output currently being viewed, which could be the SYSLOG, ULOG, or a
JOBLOG. Using the PRINT command is a three-step process:
1. PRINT <select target>
PRINT ODSN data set name * NEW for a new data set
PRINT ODSN data set name * OLD | MOD for an existing data set
PRINT OPEN <?> to print to a JES2 sysout class.
2. PRINT <range>
PRINT
PRINT <starting line> <number of lines>
3. PRINT CLOSE
For more information about the PRINT command, refer to SDSF Operation and Customization,
SA22-7670.
Figure 11-15 on page 241 illustrates using the PRINT command to open a data set so output
can be appended to it.
240

------------------------------------------------------------------------------1
ISFPCU41 OG 6972.101 #@$1 #@$2 06/26/2007 0W
10693
COLUMNS
1 80
COMMAND INPUT ===> PRINT ODSN COTTREL.XX * MOD
SCROLL ===> CSR
S
--------P
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -JOBNAME STEPNAME PRO
S
CLOCK SERV PG
PAG
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -D#$1IRLM STARTING
S
885.4 354K 0
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -D#$1IRLM ENDED. NAME
S
TOTAL ELAPSED TIME= 88
N 4020000 #@$1
2007177 18:59:50.38 STC07173 00000090 IEF352I ADDRESS SPACE
N 4000000 #@$1
2007177 18:59:50.38 STC07173 00000090 $HASP395 D#$1IRLM ENDE
N 0000000 #@$1
2007177 18:59:50.41
N 4000000 #@$1
NC0000000 #@$1
2007177 19:00:12.06 #@$1M01 00000290 K S
NC0000000 #@$1
2007177 19:00:13.87 #@$1M01 00000290 K S,DEL=R,SEG=28,CON=N
NC0000000 #@$1
2007177 19:00:18.49 #@$1M01 00000290 D A,L
MR0000000 #@$1
2007177 19:00:18.49 #@$1M01 00000080 IEE114I 19.00.18 2007.
LR
797 00000080
JOBS
M/S
TS US
LR
797 00000080 00001
00018
0000
DR
797 00000080
LLA
LLA
LLA
Figure 11-15 Print OPEN
1 The row of the first line of the syslog shown on the screen.
In Figure 11-16 we print 7500 lines, line numbers 10703 through 18203, to the currently open
print output data set. We choose line 10703 because we want to start the print at the
time19:00.
------------------------------------------------------------------------------ISFPCU41 OG 6972.101 #@$1 #@$2 06/26/2007 0W
10693 PRINT OPENED
COMMAND INPUT ===> print 10703 18203
SCROLL ===> CSR
S
--------P
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -JOBNAME STEPNAME PRO
S
CLOCK SERV PG
PAG
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -D#$1IRLM STARTING
S
885.4 354K 0
N 0004000 #@$1
2007177 18:59:50.37 STC07173 00000290 -D#$1IRLM ENDED. NAME
S
TOTAL ELAPSED TIME= 88
N 4020000 #@$1
2007177 18:59:50.38 STC07173 00000090 IEF352I ADDRESS SPACE
N 4000000 #@$1
2007177 18:59:50.38 STC07173 00000090 $HASP395 D#$1IRLM ENDE
N 0000000 #@$1
2007177 18:59:50.41
N 4000000 #@$1
NC0000000 #@$1
2007177 19:00:12.06 #@$1M01 00000290 K S
NC0000000 #@$1
2007177 19:00:13.87 #@$1M01 00000290 K S,DEL=R,SEG=28,CON=N
NC0000000 #@$1
2007177 19:00:18.49 #@$1M01 00000290 D A,L
Figure 11-16 Print range
Finally, as shown in Figure 11-17 on page 242, we close the output data set. Therefore, the
print range is from line 10703 through 18203.
241

------------------------------------------------------------------------------1
ISFPCU41 OG 6972.101 #@$1 #@$2 06/26/2007 0W
10703
142 PAGES PRINTED
COMMAND INPUT ===> print close
SCROLL ===> CSR
N 4000000 #@$1
NC0000000 #@$1
2007177 19:00:12.06 #@$1M01 00000290 K S
NC0000000 #@$1
2007177 19:00:13.87 #@$1M01 00000290 K S,DEL=R,SEG=28,CON=N
NC0000000 #@$1
2007177 19:00:18.49 #@$1M01 00000290 D A,L
MR0000000 #@$1
2007177 19:00:18.49 #@$1M01 00000080 IEE114I 19.00.18 2007.
LR
797 00000080
JOBS
M/S
TS US
LR
797 00000080 00001
00018
0000
. . .
Figure 11-17 Print Close
Note: In Figure 11-17 on page 242, 1, the top line of the screen has moved to line 10703,
which is first line written to the output data set.
11.5.3 XDC command

The X action characters from most tabular displays enable printing of output, as explained
here:
X prints output data sets.

XD displays a panel for opening a print data set, then performs the print.
XS displays a panel for opening sysout, then performs the print.
XF displays a panel for opening a print file, then performs the print.
Add the c option to any of the x action characters to close the print file when printing is
complete. For example, XDC displays a panel for opening a print data set when the data set
information is provided SDSF prints to the data set, and then closes the print data set.
Use the following steps to have XDC print one or more JES2-managed DD statements for a
job.
1. Expand the job number JOB06301 into its separate output data sets
by using the ? command.
2. Use the XDC command against the SYSPRINT DD statement.
3. Supply the attributes of a the data set that you want the output copied to.
4. Finally, XDC automatically closes the print output file.
These steps are illustrated in the following figures.
Figure 11-18 on page 243 shows using ? to expand the output available from JOB06301.
242

------------------------------------------------------------------------------ISFPCU41 OUTPUT DISPLAY ALL CLASSES LINES 5,906
LINE 1-15 (15)
COMMAND INPUT ===>
SCROLL ===> CSR
NP JOBNAME JobID
Owner
Prty C ODisp Dest
Tot-Rec Tot?
COTTREL# JOB06301 COTTREL 144 T HOLD LOCAL
225
340
403
Figure 11-18 OUTPUT queue with ? command
Figure 11-19 shows using the XDC command 1 to print the output of the SYSPRINT DD output
to a data set.
------------------------------------------------------------------------------ISFPCU41 DATA SET DISPLAY - JOB COTTREL# (JOB06301)
LINE 1-4 (4)
COMMAND INPUT ===>
SCROLL ===> CSR
NP DDNAME StepName ProcStep DSID Owner
C Dest
Rec-Cnt Page
JESMSGLG JES2
2 COTTREL T LOCAL
18
JESJCL JES2
3 COTTREL T LOCAL
45
JESYSMSG JES2
4 COTTREL T LOCAL
13
xdc SYSPRINT STEP1
102 COTTREL T LOCAL
149
Figure 11-19 Invoking the XDC command
Figure 11-20 shows defining the data set to be printed to. In this example, we create a new
PDS with a member.
ISFPNO41
COMMAND INPUT ===>
SDSF Open Print Data Set

SCROLL ===> CSR
Data set name ===> 'COTTREL.TEST.OUTPUT'

Member to use ===> fred
Disposition
===> new
(OLD, NEW, SHR, MOD)
If the data set is to be created, specify the following.
Volume serial will be used to locate existing data sets if specified.
Management class
Storage class
Volume serial
Device type
Data class
Space units
Primary quantity
Secondary quantity
Directory blocks
Record format
Record length
Block size
===>
===>
===>
===>
===>
===>
===>
===>
===>
===>
===>
===>
CYLS
1
1
10
VBA
240
3120
(Blank for default management class)

(Blank for default storage class)
(Blank for authorized default volume)
(Generic unit or device address)
(Blank for default data class)
(BLKS, TRKS, CYLS, BY, KB, or MB)
(In above units)
(In above units)
(Zero for sequential data set)
Figure 11-20 XDC data set information panel
Finally, in Figure 11-21 on page 244, notice that the data set has been closed automatically, 1
243

------------------------------------------------------------------------------ISFPCU41 DATA SET DISPLAY - JOB COTTREL# (JOB06301)
PRINT CLOSED
149 LINE 1
COMMAND INPUT ===>
SCROLL ===> CSR
NP DDNAME StepName ProcStep DSID Owner
C Dest
Rec-Cnt Page
JESMSGLG JES2
2 COTTREL T LOCAL
18
JESJCL JES2
3 COTTREL T LOCAL
45
JESYSMSG JES2
4 COTTREL T LOCAL
13
SYSPRINT STEP1
102 COTTREL T LOCAL
149
Figure 11-21 XDC automatically closes the print file
1 Data set was closed, as seen in the PRINT CLOSED message.
11.6 Using the STATUS (ST) command

The STATUS panel allows authorized users to display information about jobs, started tasks,
and TSO users on all the JES2 queues, as shown in Figure 11-22.
-----------------------------------------------------------------------------SDSF STATUS DISPLAY ALL CLASSES
LINE 1-38 (1878)
COMMAND INPUT ===>
SCROLL ===> CSR
NP JOBNAME JobID
Owner
Prty Queue
SUHOOD TSU08004 SUHOOD
15 EXECUTION
#@$2 #@$2 LOCAL
RESERVE JOB07999 COTTREL
1 PRINT
A 1739
#@$2 LOCAL
COTTRELY JOB08003 COTTREL
1 PRINT
A 1740
#@$2 LOCAL
COTTRELX JOB08002 COTTREL
1 PRINT
A 1741
#@$2 LOCAL
SMFCLR STC08005 STC
1 PRINT
1742
#@$3 LOCAL
?I#$1DBR JOB08006 HAIN
1 PRINT
A 1743
#@$1 LOCAL
HAINDBRC JOB08007 HAIN
1 PRINT
A 1744
#@$1 LOCAL
SMFCLR STC08008 STC
1 PRINT
1745
#@$1 LOCAL
SMFCLR STC08022 STC
1 PRINT
1759
#@$2 LOCAL
COTTREL TSU07941 COTTREL
15 EXECUTION
#@$2 #@$2 LOCAL
HAIN
TSU07943 HAIN
15 EXECUTION
#@$1 #@$1 LOCAL
Figure 11-22 STATUS panel example
11.6.1 Using the I action on STATUS panel

The I action on the STATUS panel, as shown in Figure 11-23 on page 245 at 1, allows you to
display information about jobs, started tasks, and TSO users on the JES2 queues. You can
process a job from this panel even if it has been printed or processed (and not yet purged).
Active jobs are highlighted on the panel.
244

-----------------------------------------------------------------------------SDSF STATUS DISPLAY ALL CLASSES
LINE 1-38 (1878)
COMMAND INPUT ===>
SCROLL ===> CSR
NP JOBNAME JobID
Owner
Prty Queue
SUHOOD TSU08004 SUHOOD
15 EXECUTION
#@$2 #@$2 LOCAL
RESERVE JOB07999 COTTREL
1 PRINT
A 1739
#@$2 LOCAL
I
COTTRELY JOB08003 COTTREL
1 PRINT
A 1740
#@$2 LOCAL
1
COTTRELX JOB08002 COTTREL
1 PRINT
A 1741
#@$2 LOCAL
SMFCLR STC08005 STC
1 PRINT
1742
#@$3 LOCAL
?I#$1DBR JOB08006 HAIN
1 PRINT
A 1743
#@$1 LOCAL
HAINDBRC JOB08007 HAIN
1 PRINT
A 1744
#@$1 LOCAL
SMFCLR STC08008 STC
1 PRINT
1745
#@$1 LOCAL
SMFCLR STC08022 STC
1 PRINT
1759
#@$2 LOCAL
COTTREL TSU07941 COTTREL
15 EXECUTION
#@$2 #@$2 LOCAL
HAIN
TSU07943 HAIN
15 EXECUTION
#@$1 #@$1 LOCAL
Figure 11-23 The I command
1 Selected job to display information about.

The result of the I action is shown in Figure 11-24.
------------------------------------------------------------------------------SDSF STATUS DISPLAY ALL CLASSES
LINE 1-38 (1878)
COMMAND INPUT ===>
SCROLL ===> CSR
NP ---------------------------------------------------------------------------|
Job Information
|
|
|
| Job name
COTTRELY Job class limit exceeded?
|
| Job ID
JOB08003 Duplicate job name wait?
|
| Job schedulable? N/A
Time in queue
|
| Job class mode
Average time in queue
|
I | Job class held?
Position in queue
of
|
|
Active jobs in queue
|
|
|
| Scheduling environment
available on these systems:
|
|
|
|
|
|
|
|
|
|
|
|
|
| F1=Help F12=Cancel
|
---------------------------------------------------------------------------LOWE
TSU08017 LOWE
1 PRINT
1754
#@$1 LOCAL
Figure 11-24 STATUS PANEL I option
For more information about the ST command, refer to SDSF Operation and Customization,
SA22-7670.
245
11.7 Resource monitor (RM) command

SDSF interacts with JES2. It can display the input, output, and held queues. If authorized, the
SDSF user can display and modify from the RM panel many of these JES2 resources. The
result of issuing the RM command is seen in Figure 11-25. The values displayed are normally
configured by your sites system programmer. For an explanation of these values, refer to the
SDSF help panel or z/OS JES2 Initialization and Tuning Reference, SA22-7533.
------------------------------------------------------------------------------SDSF RESOURCE MONITOR DISPLAY #@$3
LINE 1-17 (17)
COMMAND INPUT ===>
SCROLL ===> CSR
PREFIX=CICS* DEST=(ALL) OWNER=* SYSNAME=*
NP RESOURCE SysId Status
Limit InUse InUse% Warn% IntAvg IntHigh IntLow
BERT
#@$3
65620
322
0.49
80
322
322
322
BSCB
#@$3
0
0
0.00
0
0
0 2B
BUFX
#@$3
89
0
0.00
80
0
0
0
CKVR
#@$3
2
0
0.00
80
0
0
0
CMBS
#@$3
201
0
0.00
80
0
0
0
CMDS
#@$3
200
0
0.00
80
0
0
0
ICES
#@$3
33
0
0.00
80
0
0
0
JNUM
#@$3
32760
1342
4.09
80
1342
1342 1342
JOES
#@$3
20000
1506
7.53
80
1506
1506 1506
JQES
#@$3
32760
1342
4.09
80
1342
1342 1342
LBUF
#@$3
23
0
0.00
80
0
0
0
NHBS
#@$3
53
0
0.00
80
0
0
0
SMFB
#@$3
52
0
0.00
80
0
0
0
TBUF
#@$3
104
0
0.00
0
0
0
0
TGS
#@$3
9911
7853 79.23
80
7853
7853 7853
TTAB
#@$3
3
0
0.00
80
0
0
0
VTMB
#@$3
10
0
0.00
80
0
0
0
Figure 11-25 SDSF RM Panel
11.8 SDSF and MAS

JES2 supports a multi-access spool (MAS) configuration, that is, multiple systems sharing
JES2 input, job, and output queues. The JES2 spool is where the job input and output queues
are stored. A copy of the JES2 queues and other status information (for example, spool
space allocation maps) is written to the checkpoint data set to facilitate a warm start.
SDSF, which is an optional feature of z/OS, can monitor, manage, and control your z/OS
JES2 system. Although there are many benefits to having a MAS to match the scope of the
sysplex, it is not required. There are some situations in which it is beneficial to have the JES2
MAS smaller than the sysplex. As seen in Figure 11-26 on page 247, it is possible to have
many JESPLEXes within a single sysplex.
SDSF only works within a single MAS. Thus, if system #@$1 has a separate JES2 MAS from
systems #@$2 and #@$3, then it would not be possible to view the JES2 environment of
system #@$1 while using SDSF on #@$2 or #@$3. The systems that share a MAS are
commonly called a JESPLEX.
Figure 11-26 on page 247 shows a single sysplex, with nine z/OS systems. JESPLEX1 has
four z/OS images, JESPLEX2 has three z/OS images. Systems Y and Z each are
246
stand-alone systems. In this configuration, SDSF running on system 1 could only manage the
JES2 MAS for the systems sharing that MAS, that is, systems 1, 2, 3, and 4.
For more detailed information about JES2 MAS, refer to 10.2, JES2 multi-access spool
support on page 202.
Note: By using the Operlog facility, when it is configured appropriately, you can view all the
syslog for the sysplex from any of the JESPLEXs.
JESPlex Systems 1, 2, 3, 4
JESPlex
Systems
A, B, C
Stand alone
JES2 Systems
Figure 11-26 Multiple JESPLEXs in a single sysplex
11.9 Multi-Access Spool (MAS) command

Issue the MAS command from any command line while in SDSF to display the MAS panel, as
------------------------------------------------------------------------------1
2
ISFPCU41 DISPLAY #@$2 XCFJES2A 78% SPOOL
LINE 1-3 (3)
COMMAND INPUT ===>
SCROLL ===> CSR
NP NAME Status
SID PrevCkpt
Hold ActHold Dormancy
ActDorm SyncT
#@$1 ACTIVE
1
0.62
0
0 (0,100)
101
1
#@$2 ACTIVE
2
0.59
0
0 (0,100)
101
1
#@$3 ACTIVE
3
0.65
0
0 (0,100)
101
1
. . .
Figure 11-27 MAS display
There are many fields displayed on the MAS panel. Some of the more useful values are:
1 System logged onto.
2 Spool utilization.
247
11.10 Using the JOB CLASS (JC) command

The Job Class (JC) command allows authorized users to display and control the job classes in
the MAS. It shows both JES2 and WLM-managed job classes. If you enter JC in a command
line, the JOB CLASS panel is displayed as shown in Figure 11-28.
------------------------------------------------------------------------------ISFPCU41 CLASS DISPLAY ALL CLASSES
LINE 1-38 (38)
COMMAND INPUT ===>
SCROLL ===> CSR
NP CLASS
Status Mode Wait-Cnt Xeq-Cnt Hold-Cnt ODisp
QHld Hold
A
NOTHELD JES
(,)
NO NO
B
NOTHELD WLM
(,)
NO NO
S
NOTHELD JES
(,)
NO NO
STC
109
(,)
T
NOTHELD WLM
(,)
NO NO
TSU
11
(,)
U
NOTHELD JES
(,)
NO NO
. . .
Figure 11-28 JOB CLASS panel example
For more information about the JC command, refer to SDSF Operation and Customization,
SA22-7670.
11.11 Using the SCHEDULING ENVIRONMENT (SE) command

A scheduling environment is a list of resource names along with their required states. It allows
you to manage the scheduling of work in an asymmetrical sysplex where the systems differ. If
an MVS image satisfies all of the requirements in the scheduling environment associated with
a given unit of work, then that unit of work can be assigned to that MVS image. If any of the
resource requirements are not satisfied, then that unit of work cannot be assigned to that
MVS image.
Note: The scheduling environment is not the same as a job scheduling package such as
TWS (OPCA). A job scheduling package is responsible for keeping track of which jobs
have been submitted, tracking whether they have completed, and ensuring job
dependencies are maintained.
In contrast, the scheduling environment is used to ensure that jobs do not start when
resources such as DB2 or IMS are not available.
Figure 11-29 on page 249 shows four systems. MVSA and MVSB have a DB2 data sharing
environment. MVSC and MVSD have an IMS data sharing environment. We can set up WLM
resource variables, such as DB2DBP0 and IMS0. Jobs can be then set up so they will only be
scheduled if these WLM-managed variables match the required value.
248
Figure 11-29 JESPlex different workloads
Figure 11-29 shows a DB2 datasharing region DBP0 on systems MVSA and MVSB. JCL
such as shown in Figure 11-30 will only run on systems for which the condition 1
SCHENV=DB2DBP0 is true, such as systems MVSA or MVSB and when DBP0 is active.
WLM manages DB2DBP0.
//DB2LOAD
//
//
//
//
. . .
JOB (C003,6363),'DB2LOAD',
REGION=0M,
CLASS=A,
SCHENV=DB2DBP0,
MSGCLASS=O
Figure 11-30 SCHENV=DB2DBP0-this JCL will not run on systems where this resource is not available
SDSF allows authorized users to display the scheduling environment by using the
SE command. If an authorized user enters SE on a command line, the SCHEDULING
ENVIRONMENT panel is displayed, as shown in Figure 11-31 on page 250.
249

------------------------------------------------------------------------------ISFPCU41 DULING ENVIRONMENT DISPLAY MAS SYSTEMS
LINE 1-13 (13)
COMMAND INPUT ===>
SCROLL ===> CSR
NP SCHEDULING-ENV
Description
Systems
BATCHUPDATESE
off shift batch updates to DB
CB390SE
S/390 Component Broker SE
DB_REORGSE
reorganization of DB timeframe
ONLINEPRODSE
production online timeframe
AFTERMIDNIGHT
After Midnight Processing
AFTER6PM
After 6PM Processing
FRIDAY
Friday Processing
MONDAY
Monday Processing
SATURDAY
Saturday Processing
SUNDAY
Sunday Processing
THURSDAY
Thursday Processing
#@$1,#@$2,#@$3
TUESDAY
Tuesday Processing
WEDNESDAY
Wednesday Processing
WEEKEND
Weekend Processing
Figure 11-31 SCHEDULING ENVIRONMENT panel example
To display resources for a scheduling environment, access the panel with the R action
character, as seen in Figure 11-32.
------------------------------------------------------------------------------ISFPCU41 DULING ENVIRONMENT DISPLAY MAS SYSTEMS
LINE 1-13 (13)
COMMAND INPUT ===>
SCROLL ===> CSR
NP SCHEDULING-ENV
Description
Systems
R
BATCHUPDATESE
off shift batch updates to DB
CB390SE
S/390 Component Broker SE
DB_REORGSE
reorganization of DB timeframe
. . .
Figure 11-32 SDSF Select select scheduling resource
R To find the resources that are required to be available when work under this environment
can be run.
Figure 11-33 on page 251 shows that for work scheduled as BATCHUPDATES to run, then
two resources need to resolved. DB2_PROD has to be ON and PRIME_SHIFT has to be
OFF. A resource can have three values, ON, OFF, and RESET.
For more detailed information about the topic of resources, refer to 11.12, Using the
RESOURCE (RES) command on page 251.
250

------------------------------------------------------------------------------SDSF RESOURCE DISPLAY MAS SYSTEMS BATCHUPDATESE
LINE 1-2 (2)
COMMAND INPUT ===>
SCROLL ===> CSR
PREFIX=DEF* DEST=(ALL) OWNER=* SYSNAME=*
1
2
3
4
NP RESOURCE
ReqState #@$1
#@$2
#@$3
DB2_PROD
ON
RESET
RESET
RESET
PRIME_SHIFT
OFF
RESET
RESET
RESET
Figure 11-33 SCHEDULING ENVIRONMENT panel ReqState ON
1 ReqState is the Required state of the resource.

2 The State of this resource on system #@$1.
11.12 Using the RESOURCE (RES) command

Figure 11-33 showed resources and their required state for a particular scheduling
environment. SDSF can display all resources, and their states, via the RES command. The
result is the RES panel seen in Figure 11-35 on page 252.
A resource can have three values, ON, OFF and RESET. Work can be scheduled when a
required a resource is either ON or OFF. When a resource is in a RESET status, it matches
neither ON nor OFF. Thus, work requiring a resource to be either ON or OFF will not run if the
resource is in a RESET state.
------------------------------------------------------------------------------SDSF RESOURCE DISPLAY MAS SYSTEMS
LINE 1-13 (13)
COMMAND INPUT ===>
SCROLL ===> CSR
NP RESOURCE
#@$1
#@$2
#@$3
CB390ELEM
RESET
RESET
RESET
DB2_PROD
RESET
RESET
RESET
PRIME_SHIFT
RESET
RESET
RESET
AFTERMIDNIGHT
OFF
OFF
OFF
AFTER6PM
RESET
RESET
RESET
FRIDAY
OFF
OFF
OFF
MONDAY
OFF
OFF
OFF
SATURDAY
OFF
OFF
OFF
SUNDAY
OFF
OFF
OFF
THURSDAY
ON
ON
ON
TUESDAY
OFF
OFF
OFF
WEDNESDAY
OFF
OFF
OFF
WEEKEND
OFF
OFF
OFF
Figure 11-34 RESOURCE panel example
Using SDSF, an authorized user can reset the state of these values by overtyping the field, as
shown in Figure 11-35 on page 252. When the SDSF panel is used in this manner to issue an
MVS, or JES2 command for you, the command can be seen by viewing the LOG or the ULOG
panels.
251
In this case the commands issued were:

RO #@$1,F WLM,RESOURCE=CB390ELEM,ON
RO #@$2,F WLM,RESOURCE=CB390ELEM,ON
The ability to reset the WLM resources can, and should, be restricted by a security product
such as RACF.
------------------------------------------------------------------------------SDSF RESOURCE DISPLAY MAS SYSTEMS
LINE 1-3 (3)
COMMAND INPUT ===>
SCROLL ===> CSR
NP RESOURCE
#@$1
#@$2
#@$3
CB390ELEM
on
on
RESET
DB2_PROD
RESET
RESET
RESET
PRIME_SHIFT
RESET
RESET
RESET
Figure 11-35 Change Resource state
For more information about the RES command, refer to SDSF Operation and Customization,
SA22-7670.
11.13 SDSF and ARM

SDSF runs a batch SDSF server task. If this task is not active, then the LOG and OPERLOG
may not work, as can be seen in Figure 11-36. The server task can be defined to ARM for
automatic restart. Refer to SDSF Operation and Customization, SA22-7670, for a detailed
description of ARM configuration. You can also refer to Chapter 6, Automatic Restart
Manager on page 83, to see an example of configuring the SDSF server task to be restarted
if it abends.
------------------------------------------------------------------------------HQX7720 ----------------- SDSF PRIMARY OPTION MENU --LOG FUNCTION INOPERATIVE
COMMAND INPUT ===>
SCROLL ===> CSR
DA
Active users
INIT
Initiators
Figure 11-36 SDSF LOG function inoperative
11.14 SDSF and the system IBM Health Checker

The IBM Health Checker for z/OS provides a foundation to help simplify and automate the
identification of potential configuration problems before they impact system availability.
Individual products, z/OS components, or ISV software can provide checks that take
advantage of the IBM Health Checker for z/OS framework. The system Health Checker is a
started task named HZSPROC. If the STC is not active, then the system health checker is not
active.
SDSF has built-in support for the Health Checker. This is accessed by using the CK option.
The output can be seen in Figure 11-37 on page 253. Normally you would only be interested
in exception conditions; therefore, a SORT STATUS on the display is recommended. For more
252
information about the system Health Checker, refer to Chapter 12, IBM z/OS Health
Checker on page 257.
------------------------------------------------------------------------------SDSF HEALTH CHECKER DISPLAY #@$2
LINE 1-38 (81)
COMMAND INPUT ===>
SCROLL ===> CSR
NP NAME
CheckOwner
State
Status
ASM_LOCAL_SLOT_USAGE
IBMASM
ACTIVE(ENABLED)
EXCEPT
ASM_NUMBER_LOCAL_DATASETS
IBMASM
ACTIVE(ENABLED)
EXCEPT
ASM_PAGE_ADD
IBMASM
ACTIVE(ENABLED)
EXCEPT
ASM_PLPA_COMMON_SIZE
IBMASM
ACTIVE(ENABLED)
EXCEPT
ASM_PLPA_COMMON_USAGE
IBMASM
ACTIVE(ENABLED)
SUCCES
CNZ_AMRF_EVENTUAL_ACTION_MSGS
IBMCNZ
ACTIVE(ENABLED)
SUCCES
CNZ_CONSOLE_MASTERAUTH_CMDSYS
IBMCNZ
ACTIVE(ENABLED)
SUCCES
CNZ_CONSOLE_MSCOPE_AND_ROUTCODE IBMCNZ
ACTIVE(ENABLED)
EXCEPT
CNZ_CONSOLE_ROUTCODE_11
IBMCNZ
ACTIVE(ENABLED)
EXCEPT
CNZ_EMCS_HARDCOPY_MSCOPE
IBMCNZ
ACTIVE(ENABLED)
SUCCES
CNZ_EMCS_INACTIVE_CONSOLES
IBMCNZ
ACTIVE(ENABLED)
SUCCES
CNZ_SYSCONS_MSCOPE
IBMCNZ
ACTIVE(ENABLED)
SUCCES
. . .
Figure 11-37 SDSF CK panel
11.15 Enclaves
An enclave is a piece of work that can span multiple dispatchable units (SRBs and tasks) in
one or more address spaces, and is reported on and managed as a unit. It is managed
separately from the address space it runs in. CPU and I/O resources associated with
processing the transaction are managed by the transactions performance goal and reported
to the transaction.
A classical example of an enclave is DB2 work. The DB2 work tasks run under the DB2
STCs, not the jobs that make the DB2 SQL call. However, the CPU is reported against the
enclave. This allows the performance team to assign different priorities to different DB2 work.
Without enclaves, all the DB2 work runs as in the DB2 STCs and has the same priority.
The SDSF ENC command allows authorized personnel to view the current active work and
which enclave the work is active in; Figure 11-38 on page 254 shows SDSF ENC output.
253

------------------------------------------------------------------------------SDSF ENCLAVE DISPLAY #@$1
ALL
LINE 1-33 (33)
COMMAND INPUT ===>
SCROLL ===> CSR
1
2
3
4
5
PREFIX=COTTREL* DEST=(ALL) OWNER=* SYSNAME=*
NP TOKEN
SSType Status
SrvClass Subsys
OwnerJob Per PGN ResGro
60003D4328
DDF
INACTIVE DDF_PBPP D#$#
D#$2DIST
2
DDF_PB
200005DD1A1
DDF
INACTIVE DDF_LO
D#$#
D#$2DIST
1
DDF_PB
2B4005CD227
DDF
INACTIVE DDF_PBPP D#$#
D#$3DIST
1
DDF_PB
. . .
2FC005DD217
DDF
INACTIVE DDF_LO
D#$#
D#$1DIST
1
DDF_PB
2DC005DE4AC
JES
ACTIVE
CICS_LO JES2
#@$C1A1A
1
Figure 11-38 SDSF ENC output
1 When some work is placed in an enclave, a token is created for WLM to manage this piece
of work. Each piece of work has a unique token.
2 The work can be of various types; DDF means it is from a remote system (for example,
AIX). JES means it is from within the sysplex.
3 ACTIVE means doing work. INACTIVE means waiting for a resource.
4 Subsys indicates where the work came from.
5 OWNER indicates which job or STC made the DB2 call. D#$2DIST means it was a distributed
call, which can be from outside of the sysplex.
------------------------------------------------------------------------------SDSF ENCLAVE DISPLAY #@$1
ALL
LINE 1-33 (33)
COMMAND INPUT ===>
SCROLL ===> CSR
1
2
3
4
5
PREFIX=COTTREL* DEST=(ALL) OWNER=* SYSNAME=*
NP TOKEN
zAAP-Time zACP-Time zIIP-Time zICP-Time
60003D4328
0.00
0.00
0.00
0.00
200005DD1A1
0.00
0.00
0.00
0.00
2B4005CD227
0.00
0.00
0.00
0.00
. . .
2FC005DD217
0.00
0.00
0.00
0.00
2DC005DE4AC
0.00
0.00
0.00
0.00
Figure 11-39 SDSF ENC <scroll right>
Figure 11-39 shows the same token, 1. Notice which zAAP 2 and 3 and zIIP 4 and 5
resources have been used by these processes. However, in our test system, we have neither
of these specialty engines, so the value is 0.
Figure 11-40 on page 255 shows the SDSF DA panel with two columns, ECPU% and
ECPU-TIME. These columns display the CPU enclave usage for different jobs. A typical user of
enclave work (and thus, ECPU-TIME) is DB2 DDF. This is DB2 distributed work where the
DB2 query has come from another DB2 system. This can be from a different MVS image or
another platform, such as AIX.
Note: zIIP and zAAP are specialty processors provided by IBM. Contact IBM, or see the
IBM Web site at http://www.ibm.com, for further information about these processors.
254

------------------------------------------------------------------------------SDSF DA MVSA (ALL)
PAG
0 CPU/L
37/ 21
LINE 1-37 (935)
COMMAND INPUT ===>
SCROLL ===> CSR
1
2
3
4
NP JOBNAME Group Server
Quiesce
CPU-Time ECPU-Time
CPU% ECPU% CPUCri
DB2ADIST
NO
36467.92 510975.02
3.65 22.21 YES
OMVS
NO
147344.68 147344.68
0.43
0.31 NO
XCFAS
NO
117377.29 117377.29
3.08
2.25 NO
DB2ADBM1
NO
96534.18 100879.94
3.92
2.86 YES
TCPIP
NO
66798.66 66798.66
1.93 1.41 NO
Figure 11-40 DA command, with Enclave CPU
1 CPU-Time that has been consumed by this address space, and has been charged to this
address space.
2 ECPU-Time that been consumed by this address space. This includes CPU-TIME for work
performed on behalf of another address space. The difference between ECPU-TIME and
CPU-TIME is work that has run in this address space but was scheduled by another task.
This extra time is charged to the requesting address space.
3 Current interval percentage of CPU-TIME.
4 Current interval percentage of ECPU-TIME.
11.16 SDSF and REXX

With IBM z/OS, you can harness the versatility of REXX to interface and interact with the
power of SDSF. A function called REXX with SDSF is available that provides access to SDSF
functions through the use of the REXX programming language. This REXX support provides
a simple and powerful alternative to using SDSF batch. REXX with SDSF integrates with your
REXX executable by executing commands and returning the results in REXX variables.
For more information about this feature, refer to SDSF Operation and Customization,
SA22-7670 and Implementing REXX Support in SDSF, SG24-7419.
255
256
12
Chapter 12.
IBM z/OS Health Checker

This chapter provides details of operational considerations for the IBM z/OS Health Checker
environment on a Parallel Sysplex environment. It includes:
Introduction to z/OS Health Checker
List of checks available by component
Useful commands
257
12.1 Introduction to z/OS Health Checker

The z/OS Health Checker provides a foundation to help simplify and automate the
identification of potential configuration problems before they impact system availability. It
achieves this by comparing active values and settings to those suggested by IBM or defined
by the installation.
The z/OS Health Checker comprises two main components:
The framework, which manages functions such as check registration, messaging,
scheduling, command processing, logging, and reporting. The framework is provided as
an open architecture in support of check writing. The IBM Health Checker for z/OS
framework is available as a base function.
The Checks, which evaluate settings and definitions specific to products, elements, or
components. Checks are provided separately and are independent of the framework. The
architecture of the framework supports checks written by IBM, independent software
vendors (ISVs), and users. You can manage checks and define overrides to defaults using
the MODIFY command or the HZSPRMxx PARMLIB member.
12.2 Invoking z/OS Health Checker

z/OS Health Checker is a z/OS started task, usually named HZSPROC. To start the IBM
Health Checker for z/OS, issue the command shown in Figure 12-1.
S HZSPROC
Figure 12-1 Starting HZSPROC
Tip: The task of starting HZSPROC should be included in any automation package or in
the parmlib COMMNDxx member, so that it is automatically started after a system restart.
Each check has a set of predefined values including:
How often the check will run
Severity of the check which will then influence how the check output will be issued
Routing and descriptor code for the check
Some check values can be overridden by using SDSF, statements in the HZSPRMxx
member, or the MODIFY command; overrides are usually performed when some check values
are not suitable for your environment or configuration.
Note: Before changing any check values, consult your system programmer.
The HZSPROC started task reads parameters, if coded, from parmlib member HZSPRMxx.
After HZSPROC is active on your z/OS system images, you can invoke the Health Checker
application using option 1 CK from the SDSF primary option menu, as shown in Figure 12-2
on page 259.
258
HQX7730 ----------------COMMAND INPUT ===>
SDSF PRIMARY OPTION MENU
DA
I
O
H
ST
Active users
Input queue
Output queue
Held output queue
Status of jobs
LOG
SR
MAS
JC
SE
RES
System log
System requests
Members in the MAS
Job classes
Scheduling environments
WLM resources
-----------------------SCROLL ===> CSR
INIT
PR
PUN
RDR
LINE
NODE
SO
SP
Initiators
Printers
Punches
Readers
Lines
Nodes
Spool offload
Spool volumes
RM
CK
Resource monitor
Health checker 1
Licensed Materials - Property of IBM

Figure 12-2 CK option from SDSF menu
12.3 Checks available for z/OS Health Checker

Checks for the z/OS Health Checker are delivered both as an integrated part of a z/OS
release, or separately as PTFs. Many new and updated checks are distributed as PTFs.
The URL http://www.ibm.com/servers/eserver/zseries/zos/hchecker/check_table.html
provides an up-to-date list of checks. Table 12-1 lists the Health Checker checks and other
information.
Table 12-1 z/OS Health Checker checks
Check owner
Check name
APAR number or z/OS release
IBMASM
ASM
ASM_PAGE_ADD
Integrated in z/OS V1R8.
IBMCNZ
Consoles
CNZ_CONSOLE_MSCOPE_AND_ROUTCOD
CNZ_SYSCONS_MSCOPE
CNZ_SYSCONS_PD_MODE
CNZ_SYSCONS_ROUTCODE
CNZ_TASK_TABLE
CNZ_SYSCONS_MASTER (z/OS V1R4-V1R7 only)
OA09095 contains checks for z/OS

V1R4-V1R7 and is integrated in z/OS
V1R8.
IBMCSV
Contents
Supervision
CSV_APF_EXISTS
CSV_LNKLST_SPACE
CSV_LNKLST_NEWEXTENTS

V1R4-V1R8.
Chapter 12. IBM z/OS Health Checker
259
Check owner
Check name
IBMCS
Communications
Server
CSTCP_SYSTCPIP_CTRACE_TCPIPstackname
CSTCP_TCPMAXRCVBUFRSIZE_TCPIPstackname
CSVTAM_CSM_STG_LIMIT
IBMGRS
GRS
GRS_CONVERT_RESERVES
GRS_EXIT_PERFORMANCE
GRS_MODE
GRS_SYNCHRES

V1R8.
OA08397 supports APAR OA10830 for
z/OS V1R4-V1R6 and is integrated in
z/OS V1R7.
GRS_GRSQ_SETTING
GRS_RNL_IGNORED_CONV
IBMIXGLOGR
System logger
IXGLOGR_ENTRYTHRESHOLD
IXGLOGR_STAGINGDSFULL
IXGLOGR_STRUCTUREFULL

V1R4-V1R8.
IBMRACF
RACF
RACF_SENSITIVE_RESOURCES
RACF_GRS_RNL

V1R8.
OA10774 contains a fix for z/OS
V1R4-V1R7 in support of RACF class
XFACILIT for IBM Health Checker for
z/OS.
OA15290 contains a fix for
RACF_SENSITIVE_RESOURCES for
z/OS V1R4-V1R8.
RACF_IBMUSER_REVOKED
RACF_TEMPDSN_ACTIVE
RACF_FACILITY_ACTIVE
RACF_OPERCMDS_ACTIVE
RACF_TAPEVOL_ACTIVE
RACF_TSOAUTH_ACTIVE
RACF_UNIXPRIV_ACTIVE

V1R8.
RRS_DUROFFLOADSIZE
RRS_MUROFFLOADSIZE
RRS_RMDATALOGDUPLEXMODE
RRS_RMDOFFLOADSIZE
RRS_RSTOFFLOADSIZE

V1R8.
RRS_ARCHIVECFSTRUCTURE
IBMRSM
RSM
RSM_AFQ
RSM_HVSHARE
RSM_MAXCADS
RSM_MEMLIMIT
RSM_REAL
RSM_RSU

V1R8.
IBMSDUMP
SDUMP
SDUMP_AUTO_ALLOCATION
SDUMP_AVAILABLE

V1R8.
IBMRRS
RRS
260
Check owner
Check name
IBMUSS
z/OS UNIX
USS_AUTOMOUNT_DELAY
USS_FILESYS_CONFIG
USS_MAXSOCKETS_MAXFILEPROC

V1R4 - V1R7 and is integrated in z/OS
V1R8.
OA14022 and OA14576 contain fixes for
the USS_FILESYS_CONFIG check for
z/OS V1R4-V1R7.
IBMVSAM
VSAM
IBMVSM
VSM
IBMXCF
XCF
VSAM_SINGLE_POINT_FAILURE

V1R8.
VSAMRLS_DIAG_CONTENTION

V1R8.
VSAM_ INDEX_TRAP
OA15539 contains this check for z/OS

V1R8.
VSM_CSA_CHANGE
VSM_CSA_LIMIT
VSM_CSA_THRESHOLD
VSM_PVT_LIMIT
VSM_SQA_LIMIT
VSM_SQA_THRESHOLD

V1R8.
VSM_ALLOWUSERKEYCSA
XCF_CF_STR_PREFLIST
XCF_CDS_SEPARATION
XCF_CF_CONNECTIVITY
XCF_CF_STR_EXCLLIST
XCF_CLEANUP_VALUE
XCF_DEFAULT_MAXMSG
XCF_FDI
XCF_MAXMSG_NUMBUF_RATIO
XCF_SFM_ACTIVE
XCF_SIG_CONNECTIVITY
XCF_SIG_PATH_SEPARATION
XCF_SIG_STR_SIZE
XCF_SYSPLEX_CDS_CAPACITY
XCF_TCLASS_CLASSLEN
XCF_TCLASS_CONNECTIVITY
XCF_TCLASS_HAS_UNDESIG

V1R8.
OA14637 contains a fix for the XCF
checks for z/OS V1R4-V1R7.
12.4 Working with check output

After z/OS Health Checker has been configured and started on your z/OS system images,
output from the checks is in the form of messages issued by check routines as either:
Exception messages issued when a check detects a potential problem or a deviation from
a suggested setting
Information messages issued to the message buffer to indicate either a clean run (no
exceptions found) or that a check is not appropriate in the current environment and will not
run
Reports issued to the message buffer, often as supplementary information for an
exception message
261
You can view complete check output messages in the message buffer using the following:
HZSPRINT utility
SDSF
Health Checker log stream (in our example, HZS.HEALTH.CHECKER.HISTORY) for
historical data using the HZSPRINT utility
When a check exception is detected, a WTO is issued to the syslog. Figure 12-3 is a sample
of a check message.
HZS0002E CHECK(IBMASM,ASM_LOCAL_SLOT_USAGE): 535
ILRH0107E Page data set slot usage threshold met or exceeded
Figure 12-3 WTO message issued by check ASM_LOCAL_SLOT_USAGE
Any check exception messages are issued both as WTOs and to the message buffer. The
WTO version contains only the message text in Figure 12-3. The exception message in the
message buffer, shown in Figure 12-4 on page 263, includes both the text and an explanation
of the potential problem, including severity. It also displays information about what actions
might fix the potential problem.
Tip: To obtain the best results from IBM Health Checker for z/OS, let it run continuously on
your system so that you will know when your system has changed. When you get an
exception, resolve it using the information in the check exception message or overriding
check values, so that you do not receive the same exceptions over and over.
Also consider configuring your automation software to trigger on some of the WTOs that
are issued.
262
CHECK(IBMASM,ASM_LOCAL_SLOT_USAGE)
START TIME: 07/09/2007 00:25:51.131746
CHECK DATE: 20041006 CHECK SEVERITY: MEDIUM
CHECK PARM: THRESHOLD(30%)
* Medium Severity Exception *

ILRH0107E Page data set slot usage threshold met or exceeded
Explanation: The slot usage on 1 or more local page data sets meets
or exceeds the check warning threshold of 30%.
System Action: The system continues processing.
Operator Response:
N/A
System Programmer Response: Consider adding additional page data sets

if slot utilization remains at a high level. This can be done
dynamically via the PAGEADD command, or during the next IPL by
specifying additional data sets in the IEASYSxx parmlib member.
Problem Determination: Message ILRH0108I in the message buffer
displays the status of the local page data sets that meet or exceed
the usage warning value.
Source: Aux Storage Manager
Reference Documentation:
"Auxiliary Storage Management Initialization" in z/OS MVS
Initialization and Tuning Guide
"Statements/parameters for IEASYSxx - PAGE " in z/OS MVS
Initialization and Tuning Reference
"PAGEADD Command" in z/OS MVS System Commands
Automation: N/A
Check Reason: To check on the local page data set utilization
ILRH0108I Page Data Set Detail Report

Type
---LOCAL
Status
-----OK
Usage
----44%
Dataset Name
-----------PAGE.#@$3.LOCAL1
END TIME: 07/09/2007 00:25:51.140624
STATUS: EXCEPTION-MED
Figure 12-4 Exception message issued by check ASM_LOCAL_SLOT_USAGE
Figure 12-5 shows sample HZSPRINT JCL.

//HZSPRINT EXEC PGM=HZSPRNT,TIME=1440,REGION=0M,
//
PARM=('CHECK(*,*)','EXCEPTIONS') 1
Figure 12-5 Sample HZSPRINT JCL
263
1 SYS1.SAMPLIB(HZSPRINT) provides sample JCL and parameters that can be used for the
HZSPRINT utility. Sample output generated from the HZSPRINT utility can be seen in
Figure 12-6.
************************************************************************
*
*
* Start: CHECK(IBMRSM,RSM_MEMLIMIT)
*
*
*
************************************************************************
CHECK(IBMRSM,RSM_MEMLIMIT)
START TIME: 07/08/2007 20:25:51.393858
CHECK DATE: 20041006 CHECK SEVERITY: LOW
* Low Severity Exception *
IARH109E MEMLIMIT is zero
Explanation: Currently, the MEMLIMIT
has not been specified.
Setting MEMLIMIT too low may cause jobs
storage to fail. Setting MEMLIMIT too
of real storage resources and lead to
system loss.
setting in SMFPRMxx is zero, or

that rely on high virtual
high may cause over-commitment
performance degradation or
System Action: n/a

Operator Response: Please report this problem to the system
programmer.
System Programmer Response: An application programmer should consider
coding the MEMLIMIT option on the EXEC JCL card for any job that
requires high virtual storage. This will provide job specific
control over high virtual storage limits. You may also want to
consider using the IEFUSI exit. Finally, consider setting a system
wide default for MEMLIMIT in SMFPRMxx. Consult the listed sources
for more information.
If you are already controlling the allocation limit for high virtual
storage using the IEFUSI exit, you may wish to make this check
inactive to avoid future warnings.
Problem Determination: n/a
Source: Real Storage Manager
Reference Documentation: z/OS MVS Initialization and Tuning Reference
. . .
Figure 12-6 Sample output from HZSPRINT
By using the CK option from the SDSF main menu, you can display the various z/OS Health
Checks available and the status of the checks. Figure 12-7 on page 265 shows a sample of
the checks available on our z/OS #@$3 system.
264
SDSF HEALTH CHECKER DISPLAY #@$3

COMMAND INPUT ===>
NP NAME
ASM_PAGE_ADD
CNZ_CONSOLE_MSCOPE_AND_ROUTCODE
CNZ_SYSCONS_MSCOPE
CNZ_SYSCONS_PD_MODE
CNZ_SYSCONS_ROUTCODE
CNZ_TASK_TABLE
CSTCP_SYSTCPIP_CTRACE_TCPIP
CSTCP_TCPMAXRCVBUFRSIZE_TCPIP
. . .
LINE 1-18 (81)

SCROLL ===> CSR
State
Status
ACTIVE(ENABLED)
EXCEPTION-MEDIUM
ACTIVE(ENABLED)
EXCEPTION-LOW
ACTIVE(ENABLED)
EXCEPTION-MEDIUM
ACTIVE(ENABLED)
EXCEPTION-MEDIUM
ACTIVE(ENABLED) 1 SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED) 2 EXCEPTION-LOW
ACTIVE(ENABLED)
EXCEPTION-LOW
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
ACTIVE(ENABLED)
SUCCESSFUL
Figure 12-7 CK option from SDSF
1 A status of SUCCESSFUL indicates that the check completed cleanly.

2 A status of EXCEPTION-LOW indicates that the check has found an exception to a suggested
value or a potential problem. The exception message might be accompanied by additional
supporting information.
Note: Exceptions that Health Checker issues are categorized into low, medium, and high.
By scrolling further right on the window shown in Figure 12-7, you can display additional
information about the various checks.
To view any checks that are in an exception status in SDSF, tab your cursor to the
appropriate check and type an S to select the check. Using the example in Figure 12-7, we
typed S next to CNZ_CONSOLE_MSCOPE_AND_ROUTCODE check. The resulting detailed
information shown in Figure 12-8 on page 266 was then displayed.
265
CHECK(IBMCNZ,CNZ_CONSOLE_MSCOPE_AND_ROUTCODE)
START TIME: 07/08/2007 20:25:51.078025
CHECK DATE: 20040816 CHECK SEVERITY: LOW
There is a total of 10 consoles (1 active, 9 inactive) that are
configured with a combination of message scope and routing code values
that are not reasonable.
Console Console
Active
Type
Name
System
MCS
#@$3M01
#@$3
MCS
#@$3M02
(Inactive)
SMCS
CON1
(Inactive)
SMCS
CON2
(Inactive)
. . .
* Low Severity Exception *
MSCOPE
*ALL
*ALL
*ALL
*ALL
CNZHF0003I One or more consoles are configured with a combination of

message scope and routing code values that are not reasonable.
Explanation: One or more consoles have been configured to have a
multi-system message scope and either all routing codes or all
routing codes except routing code 11. Note: For active MCS and SMCS
consoles, only the consoles active on this system are checked. For
inactive MCS and SMCS consoles, all consoles are checked. All EMCS
consoles are checked.
System Action: The system continues processing.
Operator Response: Report this problem to the system programmer.
System Programmer Response: To view the attributes of all consoles,
issue the following commands:
DISPLAY CONSOLES,L
DISPLAY EMCS,FULL,STATUS=L
Update the MSCOPE or ROUTCODE parameters of MCS and SMCS consoles on
the CONSOLE statement in the CONSOLxx parmlib member before the next
IPL. For EMCS consoles (or to have the updates to MCS/SMCS consoles
in effect immediately), you may update the message scope and routing
code parameters by issuing the VARY CN system command with either
the MSCOPE, DMSCOPE, ROUT or DROUT parameters. If an EMCS console is
not active, find out which product activated it and contact the
product owner. Effective with z/OS V1R7, you can use the EMCS
console removal service (IEARELEC in SYS1.SAMPLIB) to remove any
EMCS console definition that is no longer needed.
. . .
Problem Determination: n/a
Automation:
n/a
Check Reason: Reduces the number of messages sent to a console in the

sysplex
ND TIME: 07/08/2007 20:25:51.352692
STATUS: EXCEPTION-LOW
Figure 12-8 Detailed information about a specific check
Using the HZSPRINT utility with the LOGSTREAM keyword, check reports can be generated
from the log stream; see Figure 12-9 on page 267. In our Parallel Sysplex environment, the
266
log stream name is called HZS.HEALTH.CHECKER.HISTORY and the CF structure name is

HZS_HEALTHCKLOG. The log stream and CF structure name must have a prefix of HZS.
//HZSPRINT EXEC PGM=HZSPRNT,TIME=1440,REGION=0M,
//
PARM=('LOGSTREAM(HZS.HEALTH.CHECKER.HISTORY)') 1
Figure 12-9 HZSPRINT with LOGSTREAM keyword
1 Specify the name of the log stream.

Figure 12-10 shows sample output from the HZSPRINT utility executed against the log
stream.
************************************************************************
*
*
* Start: CHECK(IBMVSM,VSM_CSA_THRESHOLD)
*
*
Sysplex: #@$#PLEX
System: #@$3
*
*
*
************************************************************************
CHECK(IBMVSM,VSM_CSA_THRESHOLD)
START TIME: 07/09/2007 00:30:52.739128
CHECK DATE: 20040405 CHECK SEVERITY: HIGH
CHECK PARM: CSA(80%),ECSA(80%)
IGVH100I The current allocation of CSA storage is 756K (15% of the total
size of 4760K). The highest allocation during this IPL is 16%. Ensuring
an appropriate amount of storage is available is critical to the long
term operation of the system. An exception will be issued when the
allocated size of CSA is greater than the owner specified threshold of
80%.
* High Severity Exception *
IGVH100E ECSA utilization has exceeded 80% and is now 89%
Explanation: The current allocation of ECSA storage is 89% of 72288K.
7916K (11%) is still available.
The highest allocation during this IPL is 89%.
This allocation exceeds the owner threshold.
System Action: The system continues processing. However, eventual
action may need to be taken to prevent a critical depletion of
virtual storage resources.
Operator Response:
programmer.
...
Please report this problem to the system
Figure 12-10 HZSPRINT from a log stream
12.5 Useful commands

Starting in Figure 12-11, the following commands were all issued as a z/OS modify (F)
command to the HZSPROC z/OS Health Checker started task from a z/OS console.
267
F HZSPROC,DISPLAY
HZS0203I 01.17.08 HZS INFORMATION 358
POLICY(*NONE*)
OUTSTANDING EXCEPTIONS: 25
(SEVERITY NONE: 0
LOW: 9
MEDIUM: 13
HIGH: 3)
ELIGIBLE CHECKS: 80 (CURRENTLY RUNNING: 0)
INELIGIBLE CHECKS: 1
DELETED CHECKS: 0
ASID: 0061
LOG STREAM: HZS.HEALTH.CHECKER.HISTORY - CONNECTED
HZSPDATA DSN: SYS1.#@$3.HZSPDATA
PARMLIB: 00
Figure 12-11 Display the overall status of z/OS Health Checker
Figure 12-12 illustrates how to display each check.

F HZSPROC,DISPLAY,CHECKS
HZS0200I 01.19.08 CHECK SUMMARY
370
CHECK OWNER
CHECK NAME
IBMRRS
RRS_ARCHIVECFSTRUCTURE
IBMRRS
RRS_RSTOFFLOADSIZE
IBMRRS
RRS_DUROFFLOADSIZE
IBMRRS
RRS_MUROFFLOADSIZE
IBMRRS
RRS_RMDOFFLOADSIZE
IBMRRS
RRS_RMDATALOGDUPLEXMODE
IBMCS
CSTCP_TCPMAXRCVBUFRSIZE_TCPIP
IBMCS
CSTCP_SYSTCPIP_CTRACE_TCPIP
IBMUSS
USS_MAXSOCKETS_MAXFILEPROC
IBMUSS
USS_AUTOMOUNT_DELAY
IBMUSS
USS_FILESYS_CONFIG
IBMCS
CSVTAM_CSM_STG_LIMIT
IBMIXGLOGR
IXGLOGR_ENTRYTHRESHOLD
IBMIXGLOGR
IXGLOGR_STAGINGDSFULL
IBMIXGLOGR
IXGLOGR_STRUCTUREFULL
IBMRACF
RACF_UNIXPRIV_ACTIVE
...
STATE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
AE
STATUS
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
EXCEPTION-MED
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
EXCEPTION-LOW
EXCEPTION-LOW
SUCCESSFUL
SUCCESSFUL
EXCEPTION-MED
Figure 12-12 Display of each available check
Figure 12-13 on page 268 illustrates how to display a specific check when the owner is
unknown.
F HZSPROC,DISPLAY,CHECKS,CHECK=(* 1,USS_FILESYS_CONFIG)
391
CHECK OWNER
CHECK NAME
STATE STATUS
IBMUSS
USS_FILESYS_CONFIG
AE
SUCCESSFUL
A - ACTIVE
I - INACTIVE
E - ENABLED
D - DISABLED
G - GLOBAL CHECK
+ - CHECK ERROR MESSAGES ISSUED
Figure 12-13 Display a specific check when the owner of the check is not known
1 When the owner of the check is not known, use an asterisk (*) as a wildcard. This will
display the appropriate information for that particular check.
268
Figure 12-14 illustrates how to display a specific check when the check owner and check
name are known.
F HZSPROC,DISPLAY,CHECKS,CHECK=(IBMRACF,RACF_TSOAUTH_ACTIVE)
986
CHECK OWNER
CHECK NAME
STATE STATUS
IBMRACF
RACF_TSOAUTH_ACTIVE
AE
SUCCESSFUL
A - ACTIVE
I - INACTIVE
E - ENABLED
D - DISABLED
G - GLOBAL CHECK
Figure 12-14 Display a specific check when the check owner and check name are known
Figure 12-15 illustrates how to display all checks relating to a specific check owner.
F HZSPROC,DISPLAY,CHECKS,CHECK=(IBMGRS,* 1)
002
CHECK OWNER
CHECK NAME
STATE
IBMGRS
GRS_RNL_IGNORED_CONV
AEG
IBMGRS
GRS_GRSQ_SETTING
AE
IBMGRS
GRS_EXIT_PERFORMANCE
AE
IBMGRS
GRS_CONVERT_RESERVES
AEG
IBMGRS
GRS_SYNCHRES
AE
IBMGRS
GRS_MODE
AEG
A - ACTIVE
I - INACTIVE
E - ENABLED
D - DISABLED
G - GLOBAL CHECK
STATUS
SUCCESSFUL
SUCCESSFUL
SUCCESSFUL
EXCEPTION-LOW
SUCCESSFUL
SUCCESSFUL
Figure 12-15 Display all checks relating to a specific check owner
1 When the check owner is known but the check name is not known, use an asterisk (*) as a
wildcard. This will display all checks that have that particular check owner.
Figure 12-16 on page 270 illustrates how to display detailed information about a particular
check.
269
F HZSPROC,DISPLAY,CHECKS,CHECK=(IBMGRS,GRS_CONVERT_RESERVES),DETAIL
HZS0201I 02.44.14 CHECK DETAIL
096
CHECK(IBMGRS,GRS_CONVERT_RESERVES)
STATE: ACTIVE(ENABLED)
GLOBAL STATUS: EXCEPTION-LOW
EXITRTN: ISGHCADC
LAST RAN: 07/08/2007 23:13
NEXT SCHEDULED: (NOT SCHEDULED)
INTERVAL: ONETIME
EXCEPTION INTERVAL: SYSTEM
SEVERITY: LOW
WTOTYPE: INFORMATIONAL
SYSTEM DESCCODE: 12
THERE ARE NO PARAMETERS FOR THIS CHECK
REASON FOR CHECK: When in STAR mode, converting RESERVEs can
help improve performance and avoid deadlock.
MODIFIED BY: N/A
DEFAULT DATE: 20050105
ORIGIN: HZSADDCK
LOCALE: HZSPROC
DEBUG MODE: OFF VERBOSE MODE: NO
Figure 12-16 Display detailed information for a particular check
For additional information about z/OS Health Checker, refer to IBM Health Checker for z/OS
Users Guide, SA22-7994.
270
13
Chapter 13.
Managing JES3 in a Parallel

Sysplex
This chapter discusses the following scenarios in a JES3 environment:
Introduction to JES3
JES3 job flow
JES3 in a sysplex
Dynamic System Interchange (DSI)
JES3 networking with TCP/IP
Useful JES3 operator commands
271
13.1 Introduction to JES3

A major goal of operating systems is to process jobs while making the best use of system
resources. Thus, one way of viewing operating systems is as resource managers. Before job
processing, operating systems reserve input and output resources for jobs. During job
processing, operating systems manage resources such as processors and storage. After job
processing, operating systems free all resources used by the completed jobs, thus making
the resources available to other jobs. This process is called resource management.
There is more to the processing of jobs than just the managing of resources needed by the
jobs, however. At any instant, a number of jobs can be in various stages of preparation,
processing, and post-processing activity. To use resources efficiently, operating systems
divide jobs into parts. They distribute the parts of jobs to queues to wait for needed resources.
Keeping track of where things are and routing work from queue to queue is called workflow
management, and it is a major function of any operating system.
With the z/OS JES3 system, resource management and workflow management are shared
between z/OS and JES3. Generally speaking, JES3 performs resource management and
workflow management before and after job execution. z/OS performs resource and workflow
management during job execution.
JES3 considers job priorities, device and processor alternatives, and installation-specified
preferences when preparing jobs for processing job output.
13.2 JES3 job flow

The JES3 job flow, whether in a Parallel Sysplex environment or not, has seven key phases:
1. Input service
JES3 input service accepts and queues all work entering the JES3 system. The global
processor reads the work definitions into the system and creates JES3 job structures for
them. Work is accepted from:
A TSO SUBMIT command
A local card reader
A local tape reader
A disk reader
A remote work station
Another node in a job entry network
The internal reader
2. Converter/interpreter processing
After input service processing, a job passes through the JES3 converter/interpreter
processing (C/I). As a result, JES3 learns about the resources the job requires during
execution. C/I routines provide input to the main device scheduling (MDS) routines by
determining available devices, volumes, and data sets. These service routines process
the jobs JCL to create control blocks for setup and also prevent jobs with JCL errors from
continuing in the system. The C/I section of setup processing is further divided into three
phases:
MVS converter/interpreter (C/I) processing
Prescan processing
272
Postscan processing
The first two phases can occur in either the JES3 address space on the global processor
or in the C/I functional subsystem address space on either the local or the global
processor.
3. Job resource management
The next phase of JES3 job processing is called job resource management. The job
resource management function provides for the effective use of system resources. JES3
main device scheduler (MDS) functionality, alias the setup, ensures the operative use of
non-sharable mountable volumes, eliminates operator intervention during job execution,
and performs data set serialization. It oversees specific types of pre-execution job setup
and generally prepares all necessary resources to process the job. The main device
scheduler routines use resource tables and allocation algorithms to satisfy a jobs
requirements through the allocation of volumes and devices, and, if necessary, the
serialization of data sets.
4. Generalized main scheduling
After a job is set up, it enters JES3 job scheduling. JES3 job scheduling is the group of
services that govern where and when z/OS execution of a JES3 job occurs. Job
scheduling controls the order and execution of jobs running within the JES3 complex.
5. Job execution
Jobs are scheduled to the waiting initiators on the JES3 main processors. For the sysplex
environment, the use of Workload Manager (WLM) allows the optimization of resources to
address spaces using goals defined for the various work using a WLM policy.
6. Output processing
The final part of JES3 job processing is called job output and termination. Output service
routines operate in various phases to process SYSOUT data sets destined for print or
punch devices (local, RJP, or NJE), TSO users, internal readers, external writers, and
writer functional subsystems.
7. Purge processing
Purge processing represents the last JES3 processing step for any job. It releases the
resources used during the job.
13.3 JES3 in a sysplex

JESXCF is an XCF application. It provides, based on XCF coupling services, common
inter-processor and intra-processor communication services for both JES3 and JES2
subsystems, as illustrated in Figure 13-1 on page 274.
273
Figure 13-1 JES3 configuration in a sysplex
13.4 Global-only JES3 configuration

In this configuration there is a single z/OS and JES3 complex, as illustrated in Figure 13-2.
The sysplex configuration is defined as either PLEXCFG=XCFLOCAL or
PLEXCFG=MONOPLEX in the IEASYSxx PARMLIB member.
Figure 13-2 Global-Only configuration
274
13.5 Global-local JES3 single CEC

With multiple z/OS images on a single physical CEC, the sysplex configuration must be
specified as PLEXCFG=MILTISYSTEM in the IEASYSxx parmlib member. The JES3
systems communicate with each other using XCF services using either CTCs or the Coupling
Facility, as illustrated in Figure 13-3.
Figure 13-3 Global-Local configuration on a single CEC
13.6 Global-Local JES3 multiple CEC

With multiple z/OS images in multiple CECs, the PLEXCFG=MULTISYSTEM must be
specified in the IEASYSxx parmlib member and the use of a Sysplex Timer is required see
Figure 13-4.
Figure 13-4 Global-Local configuration on multiple CECs
275
13.7 z/OS system failure actions for JES3

The following are actions related to JES3 if a z/OS system fails:
If the z/OS system that failed was the JES3 global processor, we recommend that you
bring that system back up as quickly as possible. You could change one of the JES3 local
processors to be the JES3 global using dynamic system interchange (DSI). However, in
some environments, this process is non-trivial, and might not even be possible. Whether
or not you switch one of your JES3 local processors to be the global, you should not
attempt to use Automatic Restart Manager to do cross-system restarts of any subsystems
running on a JES3 global system.
If the system that failed was a JES3 local processor, then after you are certain that the
system has been partitioned out of the sysplex (as indicated by the IXC105I message
SYSPLEX PARTITIONING HAS COMPLETED...), issue the command *START,main,FLUSH to
ensure that those elements registered with Automatic Restart Manager can successfully
restart on another JES3 local system.
13.8 Dynamic system interchange

Dynamic system interchange (DSI) is the backup facility to be used if a permanent machine
or program failure occurs on the global, or if system reconfiguration is necessary for
preventive maintenance.
DSI allows JES3 to continue operation by switching the global function to a local main in the
same JES3 complex. If a failure occurs during DSI, try a hot start. A failure during connect
processing could be the cause of the failure. If the failure recurs, a warm start is required.
The DSI procedure consists of a sequence of commands entered on either the old or the new
global. Your systems programmer should have a DSI procedure tailored for your installation
and have it updated to reflect any changes to your installations configuration.
Important: Ensure that the old global's JES3 address space is no longer functioning when
the new global is being initialized during the DSI process. This includes the JES3DLOG
address space that might have been executing on the global z/OS system as well.
13.9 Starting JES3 on the global processor

The types of starts and restarts for the global processor are:
Cold start
Warm start
Warm start with analysis
Warm start to replace a spool data set
Warm start with analysis to replace a spool data set
Hot start with refresh
Hot start with refresh and analysis
Hot start
Hot start with analysis
You must use a cold start when starting JES3 for the first time. JES3 initialization statements
are read as part of cold start, warm start, and hot start with refresh processing.
276
If JES3 detects any error in the initialization statements, it prints an appropriate diagnostic
message on the console or in the JES3OUT data set. JES3 ends processing if it cannot
recover from the error.
13.10 Starting JES3 on a local processor

Use a local start to restart JES3 on a local main after a normal shutdown on the local, after
JES3 ends because of a failure in either the local JES3 address space or z/OS. You must also
start each local main after you perform a cold start or any type of warm start on the global or
after you use a hot start to remove or reinstate a spool data set on the global processor. You
can perform a local start any time the global is active.
You do not have to IPL the local main before you perform the local start unless one of the
following is true:
z/OS was shut down on the local main.
z/OS was not previously started.
z/OS failed on the local main.
A cold start or any type of warm start was performed on the global processor.
13.11 JES3 networking with TCP/IP

Prior to z/OS V1R8, the network job entry (NJE) function in JES3 was limited to the binary
synchronous (BSC) and system networking architecture (SNA) protocols. Both of these
communication protocols require dependent hardware which is, or soon will be, out of service.
There are several solutions that provide SNA connectivity over TCP/IP (such as Enterprise
Extender). However, compared to a pure TCP/IP network, these solutions suffer from
performance and interoperability problems, and are usually more difficult to maintain.
Starting with z/OS V1R8, JES3 provides support for NJE over TCP/IP. JES2 has been
providing NJE over TCP/IP support since z/OS V1R7. VM (RSCS), VSE, and AS/400 have
been providing NJE over TCP/IP support for several releases. JES3 on z/OS V1R8 is now
able to communicate with JES2, VM, VSE, AS/400, or any other applicable component, using
NJE over TCP/IP.
As TCP/IP becomes the standard for networking today, JES now implements NJE using the
TCP/IP protocol.
To send data from one node to another using TCP/IP, a virtual circuit is established between
the two nodes. The virtual circuit allows TCP/IP packets to be sent between the nodes. Each
node is assigned an IP address.
The commands to start/stop network devices are similar to managing the SNA devices for
NJE.
Figure 13-5 on page 278 displays a TCP/IP NJE configuration.
277
Figure 13-5 TCP/IP NJE configuration
Note: The NETSRV address space is a JES component used by both JES2 and JES3.
The NETSRV address space is a started task that communicates with JES to spool and
de-spool jobs and data sets. In Figure 13-6, a sample JES3 node definition for nodes BOSTON
and NEW YORK are defined.
Figure 13-6 JES3 node definitions
The networking flow between the nodes Boston and New York is illustrated in Figure 13-7 on
page 279.
278
Figure 13-7 Networking flow between nodes
13.11.1 JES3 TCP/IP NJE commands

This section discusses JES3 TCP/IP NJE commands and their function. Figure 13-8 shows
the use of I NETSERV.
Figure 13-8 Inquire on NETSRV
Figure 13-9 shows how to alter the port used.
Figure 13-9 Altering the port used
Table 13-1 on page 280 lists useful commands for TCP/IP NJE and provides a brief
description of their purpose.
279
Table 13-1 Commands for TCP/IP JNE and their purpose

Command
Purpose
*I,SOCKET=ALL
Tell me what sockets I have and what the current connections are.
*I,NETSERV=ALL
Tell me what Netservs I have.
*I,SOCKET=name
Tell me about a particular socket.
*I,NETSERV=name
Tell me about a particular Netserv.
*I,NJE,NAME=node
Produces specific TCP/IP NJE information if the node is TYPE=TCP.
The following figures display typical output from these as issued to a z/OS 1.9 system.
Figure 13-10 illustrates how to inquire about sockets and connections.
*I,SOCKET=ALL
IAT8709 SOCKET INQUIRY RESPONSE 836
INFORMATION FOR SOCKET WTSCNET
NETSERV=JES3NS1, HOST=WTSCNET.ITSO.IBM.COM, PORT=
0,
NODE=WTSCNET, JTRACE=NO, VTRACE=NO, ITRACE=NO, ACTIVE=NO,
SERVER=NO
INFORMATION FOR SOCKET @0000001
NETSERV=JES3NS1, HOST=, PORT=
0, NODE=WTSCNET, JTRACE=NO,
VTRACE=NO, ITRACE=NO, ACTIVE=YES, SERVER=YES
END OF SOCKET INQUIRY RESPONSE
Figure 13-10 Sockets and current connections
Figure 13-11 illustrates how to inquire about available NETSERVs.

*I,NETSERV=ALL
IAT8707 NETSERV INQUIRY RESPONSE 832
INFORMATION FOR NETSERV JES3NS1
SYSTEM=SC65, HOST=WTSC65.ITSO.IBM.COM, PORT=
JTRACE=NO, VTRACE=NO, ITRACE=NO, ACTIVE=YES
SOCKETS DEFINED IN THIS NETSERV
SOCKET
ACTIVE
NODE
SERVER
WTSCNET NO
WTSCNET NO
@0000001 YES
WTSCNET YES
END OF NETSERV INQUIRY RESPONSE
0, STACK=TCPIP,
Figure 13-11 NETSERVs available
Figure 13-12 illustrates how to display a specific socket.

*I,SOCKET=WTSCNET
IAT8709 SOCKET INQUIRY RESPONSE 840
INFORMATION FOR SOCKET WTSCNET
NETSERV=JES3NS1, HOST=WTSCNET.ITSO.IBM.COM, PORT=
0,
NODE=WTSCNET, JTRACE=NO, VTRACE=NO, ITRACE=NO, ACTIVE=NO,
SERVER=NO
END OF SOCKET INQUIRY RESPONSE
Figure 13-12 Display a specific socket
280
Figure 13-13 illustrates how to display a specific NETSERV.

*I,NETSERV=JES3NS1
IAT8707 NETSERV INQUIRY RESPONSE 842
INFORMATION FOR NETSERV JES3NS1
SYSTEM=SC65, HOST=WTSC65.ITSO.IBM.COM, PORT=
JTRACE=NO, VTRACE=NO, ITRACE=NO, ACTIVE=YES
SOCKETS DEFINED IN THIS NETSERV
SOCKET
ACTIVE
NODE
SERVER
WTSCNET NO
WTSCNET NO
@0000001 YES
WTSCNET YES
END OF NETSERV INQUIRY RESPONSE
0, STACK=TCPIP,
Figure 13-13 Display a specific NETSERV
Figure 13-14 illustrates how to display specific TCP/IP NJE information.

*I,NJE,NAME=WTSCNET
IAT8711 NODE INQUIRY RESPONSE 845
INFORMATION FOR NODE WTSCNET
TYPE=TCPIP, JT=1, JR=1, OT=1, OR=1, SS=NO, TLS=NO, ACTIVE=YES,
PWCNTL=SENDCLR
SOCKETS DEFINED FOR THIS NODE
SOCKET
ACTIVE SERVER NETSERV SYSTEM
WTSCNET NO
NO
JES3NS1 SC65
@0000001 YES
YES
JES3NS1 SC65
END OF NODE INQUIRY RESPONSE
Figure 13-14 Specific TCP/IP NJE information
13.12 Useful JES3 operator commands

Table 13-2 lists commonly used JES3 commands. It also lists the equivalent JES2 or z/OS
command. Additional information about JES3 commands can be found in z/OS V1R10.0
JES3 Commands, SA22-7540.
Table 13-2 Comparison of common JES3, JES2, and z/OS commands and actions
Action
JES3 command
JES2 command
Start a Process or Device
*Start or *S
$S
Stop a Process or Device
*Cancel or *C
$P
Restart a Process or
Device
*Restart or *R
$R
Halt a Process or Device
(see Cancel)
$Z
Cancel a Process or
Device
*Cancel or *C
$C
Modify or Set or Reset
*Modify or *F
$T
Hold a Job or Device
*Modify or *F
$H
Release a Job or Device
*Modify or *F
$O
z/OS command
281
Action
JES3 command
JES2 command
Repeat
*Restart or *R
$N
Inquire or Display
*Inquire or *I
$D
DISPLAY
DeviceOnline/Offline
*Vary or *V
$S, $P
VARY
Stop Local or Global

Processor
*Return
$PJES2
Send Command to
Remote Node
*Send or *T
$M or $N
ROUTE
Send Message to
Console(s)
*Message or *Z
$D M
SEND
Inquire status of a printer
*I D D/uuu
$D U,PRTS
$D U,PRTnnn
Inquire number of pages

left in current job
*S dev,P
$L Jnnnn,ALL or
$L STCnnn,ALL
Start printer
*S devname
$S PRTnnn
Modify or set printer

characteristics
*F,D=devname or
*S uuu,WC=class(es)
$T PRTnnn or
$T
PRTnnn,Q=class(es)
Interrupt printer and return

output to output queue
$I PRTnnn
Halt (temporarily) a printer
$Z PRTnnn
z/OS command
Restart a printer
*R,devname
$E PRTnnn
Restart and repeat printer
*R devname,G
$N PRTnnn
Restart and requeue

printer
*R devname,J
$E PRTnnn
Restart printer and

schedule new pass
*R devname,RSCD
$E PRTnnn
Backspace printer nnnn

pages
*R devname,C or
N,R=-ppppP
$B PRTnnn,pppp or
$B PRTnnn,D to the
start of the data set
Forward printer pppp

pages
*R devname,R=+ppppP
$F PRTnnn,pppp or
$F PRTnnn,D to the
end of the data set
Cancel printer
*C devname
$C PRTnnn
Vary printer or device

offline/online
*V devname,OFF or ON
Start a FSS
*MODIFY,F,FSS=fssna
me,ST=Y
Stop a FSS
*C devname to the last

device for the FSS
V ddd,OFFLINE
or ONLINE
1 FSS is started by the z/OS command S fssprocname and then a JES2 command $S PRTnnn.
2 FSS stops if specified in the parameters.
282
14
Chapter 14.
Managing consoles in a Parallel

Sysplex
This chapter provides information about console operations in a Parallel Sysplex
environment, including;
Console management
Operating from the HMC
Console buffer shortages
Console command routing
Message Flood Automation (MFA)
z/OS Management Console
283
14.1 Introduction to managing consoles in a Parallel Sysplex

Historically, operators of a z/OS image have received messages and entered commands from
multiple console support (MCS) consoles. Today, with console support available across
multiple systems, a sysplex comprised of many z/OS images can be operated from a single
console. Because the operator has a single point of control for all images, MCS often refers to
multisystem console support.
In a sysplex, MCS consoles can:
Be attached to any system
Receive messages from any system in the sysplex
Route commands to any system in the sysplex
Therefore, the following considerations apply when defining MCS consoles in this
environment:
There is no requirement that each system have consoles attached.
A sysplex, which can be up to 32 systems, can be operated from a single console.
Multiple consoles can have master command authority.
There is no need to define a master console, or define alternate console groups for
console switch operation.
In z/OS V1.8, the single master console was eliminated, which removed a single point of
failure. The console switch function was also removed, which removed a potential point of
failure because you can now define more than two consoles with master console authority.
This is discussed in more detail in 14.2.1, Sysplex master console on page 286.
There are four types of operator consoles available for use in a sysplex:
284
MCS
MCS consoles are display devices that are attached to a z/OS system to
provide communication between operators and z/OS. MCS consoles are
defined to a local non-SNA control unit (for example an OSA Integrated
Console Controller, or 2074). Currently you can define a maximum of 99
MCS consoles for the entire Parallel Sysplex. In a future release of z/OS,
IBM plans to increase the maximum number of MCS and SNA MCS
(SMCS) consoles that can be defined and active in a configuration from
99 per sysplex to 99 per system in the sysplex.
SMCS
SNA MCS consoles use z/OS Communications Server to communicate

with the system and may be remotely attached to the system. SMCS
consoles are only available for use when the z/OS Communications Server
is active. See 14.2.3, SNA MCS consoles on page 287 for more details.
EMCS
Extended MCS consoles are defined and activated by authorized programs

acting as operators. An extended MCS console is actually a program that
acts as a console. See 14.2.2, Extended MCS consoles on page 286 for
more details.
Hardware
In this context, the term hardware (or system) consoles refers to the
interface provided by the Hardware Management Console (HMC) on an
IBM System z processor. It is referred to as SYSCONS. See 14.4,
Operating z/OS from the HMC on page 291 for more details.
14.2 Console configuration

You can use the DISPLAY CONSOLES command to display the status of all consoles, or the
status of a specific console in the sysplex, including MCS and SMCS. For information related
to extended MCS (EMCS) consoles, use the DISPLAY EMCS command. This command is
explained in more detail in 14.2.2, Extended MCS consoles on page 286.
There are a number of parameters you can use when displaying the console configuration to
obtain specific information. Some of the command parameter options are discussed in this
chapter. For more detailed information about this topic, refer to MVS System Commands,
SA22-7627.
An example of a response from the DISPLAY CONSOLE command is shown in Figure 14-1. This
figure is referenced throughout this chapter and shows an example of two MCS consoles.
D C
IEE889I 00.32.12 CONSOLE DISPLAY 508
MSG: CURR=0
LIM=3000 RPLY:CURR=1
LIM=999 SYS=#@$1
PFK=00
CONSOLE
ID --------------- SPECIFICATIONS --------------SYSLOG
COND=H
AUTH=CMDS
NBUF=N/A
ROUTCDE=ALL
OPERLOG 1
COND=H
AUTH=CMDS
NBUF=N/A
ROUTCDE=ALL
#@$3M01 2
01 COND=A 5
AUTH=MASTER 6
NBUF=N/A
08E0
3
AREA=Z
MFORM=T,S,J,X
#@$3
4
DEL=R
RTME=1/4
RNUM=20
SEG=38
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=OPTIONAL
CMDSYS=#@$3 7
MSCOPE=*ALL 8
#@$2M01
11 COND=A
AUTH=MASTER
NBUF=N/A
08E0
AREA=Z
MFORM=T,S,J,X
#@$2
DEL=R
RTME=1/4
RNUM=20
SEG=38
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=OPTIONAL
. . .
Figure 14-1 Display console configuration
1 OPERLOG is active.
2 Name of the console as defined in the CONSOLxx member.
3 Device address of the console.
4 The system where the console is defined.
5 The status of the console. In this case, it is A for Active.
6 The console has master command authority.
7 The system that the command is directed to, if no command prefix is entered.
8 Messages are received at this console from these systems. *ALL indicates messages from
all active systems in the sysplex will be received on this console.
Chapter 14. Managing consoles in a Parallel Sysplex
285
14.2.1 Sysplex master console

Prior to z/OS 1.8, in a stand-alone z/OS environment there was one master console for each
system, and in a sysplex there was one master console for all systems within the sysplex. The
master console was identified as an M on the COND= field 1 of the DISPLAY CONSOLE
. . .
#@$2M01
08E0
#@$2
. . .
11
COND=M 1
AUTH=MASTER
NBUF=N/A
AREA=Z
MFORM=T,S,J,X
DEL=R
RTME=1/4
RNUM=20
SEG=38
CON=N
Figure 14-2 Master console prior to z/OS 1.8
In z/OS V1.8, the single master console was eliminated, which removed a single point of
failure. The functions associated with the master console, including master command
authority and the ability to receive messages delivered via the INTERNAL or INSTREAM
message attribute, can be assigned to any console in the configuration, including EMCS
consoles. The console switch function has also been removed, which removed another
potential point of failure because you are now able to define more than two consoles with
master console authority.
The display master console command (D C,M) no longer identifies a console as an M on the
COND= field. In Figure 14-3, the three systems in the sysplex have the COND field that is not
M 1 (in this case A for Active, but there are other possible conditions) and the AUTH field as
MASTER 2, which means the console is authorized to enter any operator command. All three
consoles have master console authority in the sysplex, so there is no longer a requirement to
switch a console if one is deactivated or fails.
D C,M
. . .
#@$3M01
08E0
#@$3
. . .
#@$2M01
08E0
#@$2
. . .
#@$1M01
08E0
#@$1
. . .
01
11
13
COND=A 1
AUTH=MASTER 2
NBUF=0
AREA=Z
MFORM=T,S,J,X
DEL=R
RTME=1/4
RNUM=20
SEG=38
CON=N
COND=A 1
AUTH=MASTER 2
NBUF=N/A
AREA=Z
MFORM=T,S,J,X
DEL=R
RTME=1/4
RNUM=20
SEG=14
CON=N
COND=A 1
AUTH=MASTER 2
NBUF=N/A
AREA=Z
MFORM=T,S,J,X
DEL=R
RTME=1/4
RNUM=20
SEG=38
CON=N
Figure 14-3 Consoles with master authority
14.2.2 Extended MCS consoles

Extended MCS consoles are defined and activated by authorized programs acting as
operators. An extended MCS console is actually a program that acts as a console. It is used
to issue z/OS commands and to receive command responses, unsolicited message traffic,
and the hardcopy message set. For example, IBM products such as TSO/E, SDSF, and
286
NetView utilize EMCS functions. To display information related to extended MCS (EMCS)
consoles, use the DISPLAY EMCS command.
There are a number of parameters you can use when displaying the EMCS console
configuration to obtain specific information. Review MVS System Commands, SA22-7627, for
more detailed information about this topic. Some of the command parameter options are
discussed in this chapter. An example of a response from the DISPLAY EMCS command is
shown in Figure 14-4. The command entered used the S parameter to display a summary of
EMCS consoles, which includes the number and names for the consoles that meet the
criteria.
D EMCS,S
IEE129I 21.50.06 DISPLAY EMCS 878
DISPLAY EMCS,S
NUMBER OF CONSOLES MATCHING CRITERIA: 18
*DICNS$1 *DICNS$2 *DICNS$3 COTTRELC HAIN
FOSTER
#@$1
*ROUTE$1 #@$2
*ROUTE$2 #@$3
*ROUTE$3 *SYSLG$1 *OPLOG01
*SYSLG$2 *OPLOG02 *SYSLG$3 *OPLOG03
Figure 14-4 EMCS console summary
To obtain more information about the EMCS consoles or a specific console, you can use the
I (info) or F (full) parameter. Figure 14-5 shows output from the D EMCS,F command for a
specific EMCS console, CN=*SYSLG$3. The output is similar to the D C command.
D EMCS,F,CN=*SYSLG$3
CNZ4101I 21.53.05 DISPLAY EMCS 887
DISPLAY EMCS,F,CN=*SYSLG$3
CN=*SYSLG$3 1 STATUS=A 2 CNID=03000005 KEY=SYSLOG
SYS=#@$3
ASID=000B JOBNAME=-------- JOBID=-------HC=N AUTO=N DOM=NONE
TERMNAME=*SYSLG$3
MONITOR=-------CMDSYS=#@$3 3
LEVEL=ALL
AUTH= MASTER 4
MSCOPE=#@$3 5
ROUTCDE=NONE
INTIDS=N UNKNIDS=N
ALERTPCT=100
QUEUED=0
QLIMIT=50000
SIZEUSED=5184K
MAXSIZE=2097152K
Figure 14-5 EMCS console detail
1 Name of the console as defined by the program activating the console.

2 The status of the console. In this case, it is A for Active.
3 The system that the command is directed to, if no command prefix is entered.
4 The console has master command authority.
5 Messages are received at this console from these systems.
14.2.3 SNA MCS consoles

SMCS consoles use z/OS Communications Server to communicate with the system, and may
be remotely attached to the system. SMCS consoles are only available for use when the
287
z/OS Communications Server is active and have the appropriate VTAM and console
definitions set up.
The SMCS consoles are defined in the CONSOLxx member, and can be defined with the
same configuration as MCS consoles, including master command authority. Using the D C,L
command, as seen in Figure 14-6, the SMCS console is identified by the COND field A,SM 1
which means the console is an active SMCS console. In this example, the console also has
the AUTH field as MASTER 2, which means the console has master command authority and is
authorized to enter any operator command.
D C,L
. . .
CON1
03
COND=A,SM 1 AUTH=MASTER 2
NBUF=N/A
AREA=Z
MFORM=T,S,J,X
DEL=R
RTME=1/4
RNUM=28
SEG=28
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=REQUIRED
CMDSYS=*
MSCOPE=*ALL
INTIDS=N UNKNIDS=N
. . .
Figure 14-6 Display SMCS consoles
A D C,SMCS command can also be used to display the status and APPLID of the SMCS VTAM
configuration on each system in the sysplex, as shown in Figure 14-7.
D C,SMCS
GENERIC=SCSMCS$$
SYSTEM
APPLID
SMCS STATUS
#@$2
SCSMCS$2 ACTIVE
#@$1
SCSMCS$1 ACTIVE
#@$3
SCSMCS$3 ACTIVE
Figure 14-7 Display SMCS VTAM information
14.2.4 Console naming

The MCS and SMCS consoles have an ID number assigned to them, but they must also have
a name assigned in the CONSOLxx member. All consoles in the sysplex must be assigned a
console name. An IPL will stall during console initialization if consoles have not been named
in the CONSOLxx member.
Naming the consoles makes it possible to create console groups. It also reduces the
possibility of exceeding the current 99 console limit for the sysplex. IBM plans to increase the
maximum number of MCS and SMCS consoles that can be defined and active in a
configuration from 99 per sysplex to 99 per system in the sysplex in a future release. If a
system is re-IPLed, then all consoles attached to the system will use the same console IDs
when they reuse a name.
288
14.2.5 MSCOPE implications

In a sysplex, the console MSCOPE (message scope) parameter is required to specify which
systems can send messages to a console. This parameter is initially set at IPL using
information defined to the system in the CONSOLxx member. You can display the current
MSCOPE settings of a console by issuing the D C command.
A new MSCOPE value can be set by issuing a VARY CN command, for example:
V CN(<cnsl>),MSCOPE=<sys>
It is possible to add or delete system names from the current MSCOPE value of a console.
To add system names, enter:
V CN(<cnsl>),AMSCOPE=<sys>
To delete system names, enter:
V CN(<cnsl>),DMSCOPE=<sys>
The <cnsl> can be:
*
The console that you are currently issuing commands from
#@$1M01
A specific console name
(#@$1M01,#@$2M01,.)
A list of specific console names
The <sys> can be:

*
The system the console is connected to
*ALL
All active systems in the sysplex
#@$1
A specific system
(#@$1,#@$2,...)
A list of systems
For example, Figure 14-8 shows a display of an EMCS console named TEST01 which shows
an MSCOPE setting of *ALL 1.
D EMCS,I,CN=TEST01
DISPLAY EMCS,I,CN=TEST01
CN=TEST01
STATUS=A
CNID=01000004 KEY=SDSF
. . .
MSCOPE=*ALL 1
ROUTCDE=NONE
INTIDS=N UNKNIDS=N
Figure 14-8 Display MSCOPE information before change
In Figure 14-9, a V CN command is issued from console TEST01 to change its MSCOPE from
*ALL to #@$3.
V CN(*),MSCOPE=#@$3
IEE712I VARY CN PROCESSING COMPLETE
Figure 14-9 Changing MSCOPE of a console
289
In Figure 14-10, a subsequent display of the EMCS console named TEST01 shows the
MSCOPE setting has changed to #@$3 1.
D EMCS,I,CN=TEST01
DISPLAY EMCS,I,CN=TEST01
CN=TEST01
STATUS=A
CNID=01000004 KEY=SDSF
. . .
MSCOPE=#@$3 1
ROUTCDE=NONE
INTIDS=N UNKNIDS=N
Figure 14-10 Display MSCOPE information after change
You can monitor your sysplex from a single console in the sysplex if its MSCOPE value is set
to *ALL on that console. Keep in mind that all consoles with an MSCOPE value of *ALL will
receive many more messages than consoles defined with an MSCOPE of a single system.
Thus, there is more chance of running into a console buffer shortage. This is discussed in
14.5, Console buffer shortages on page 295.
14.2.6 Console groups

Console groups can be defined using the CNGRPxx PARMLIB member. You can specify
MCS, SMCS, and extended MCS consoles as members of these groups. You can use
console groups to specify the order in which consoles are to receive messages, or to identify
the consoles that must be inactive for the system to place the system console into problem
determination state. See 14.4, Operating z/OS from the HMC on page 291 for more
information about this topic.
When a system joins a sysplex, the system inherits any console group definitions that are
currently defined in the sysplex. Its own console group definitions in the INIT statement in
CONSOLxx are ignored. If there are no console groups defined when a system joins the
sysplex, then the joining systems parmlib definitions will be in effect for the entire sysplex.
After the system is up, any system in the sysplex can issue the SET CNGRP command to add or
change the console group definitions; see Figure 14-11.
D CNGRP
IEE679I 02.34.19 CNGRP DISPLAY 945
NO CONSOLE GROUPS DEFINED
SET CNGRP=01
IEE712I SET CNGRP PROCESSING COMPLETE
D CNGRP
CONSOLE GROUPS ACTIVATED FROM SYSTEM #@$3
---GROUP--- ---------------------MEMBERS-------MASTER
01 #@$1M01 #@$2M01 #@$3M01
HCGRP
01 *SYSLOG*
SET CNGRP=NO
IEE712I SET CNGRP PROCESSING COMPLETE
D CNGRP
NO CONSOLE GROUPS DEFINED
Figure 14-11 Activating console groups
290
14.3 Removing a console

The VARY CN command is used to set attributes for MCS, SMCS, and extended MCS
consoles. The consoles specified in the VARY CN commands must be defined as consoles in
the CONSOLxx parmlib member. Extended MCS consoles can also be accepted.
As seen in Figure 14-12, you can use the VARY CN command to vary a console offline 1 and
vary the console back online 2. The attributes for the console definition are taken from the
CONSOLxx member when the console is brought back online. There are a number of
parameters you can use when using the VARY CN command to change console attributes;
refer to MVS System Commands, SA22-7627, for more details.
D C
. . .
#@$2M01
01 COND=A
AUTH=MASTER
NBUF=N/A
08E0
AREA=Z
MFORM=T,S,J,X
#@$2
DEL=R
RTME=1/4
RNUM=20
SEG=20
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=OPTIONAL
CMDSYS=#@$2
MSCOPE=*ALL
INTIDS=N UNKNIDS=N
. . .
V CN(#@$2M01),OFFLINE 1
IEE303I #@$2M01 OFFLINE
V CN(#@$2M01),ONLINE
2
MSG: CURR=0
LIM=999 SYS=#@$2
PFK=00
CONSOLE
ID --------------- SPECIFICATIONS --------------#@$2M01
01 COND=A
AUTH=MASTER
NBUF=0
08E0
AREA=Z
MFORM=T,S,J,X
#@$2
DEL=R
RTME=1/4
RNUM=20
SEG=20
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=OPTIONAL
CMDSYS=#@$2
MSCOPE=*ALL
INTIDS=N UNKNIDS=N
Figure 14-12 Removing and reinstating a console
14.4 Operating z/OS from the HMC

You can operate a z/OS system or an entire sysplex using the operating system message
facility of the Hardware Management Console (HMC). This facility is also known as the
SYSCONS console and is considered an EMCS type of console. You would generally only
use this facility if there were problems with the consoles defined with master console authority
in the CONSOLxx parmlib member. The procedures for using the HMC to operate z/OS are
not unique to a sysplex, and detailed information can be found in Hardware Management
Console Operations Guide, SC28-6837.
291
There are various scenarios where the use of the HMC as the SYSCONS may be required.
One of these scenarios is at IPL time, if no other consoles are available.
Normally a locally attached console is used when a system is IPLed. The console is defined
as a Nucleus Initialization Program (NIP) console in the operating system configuration
(IODF). If none of the consoles specified as a NIP console are available, or if there are none
specified for the system, then the system will IPL using the SYSCONS console as the NIP
console. If there is no working HMC available, then the Support Element on the processor will
be used instead. When the SYSCONS console is used during IPL, or for receiving other
messages that may be sent to it from the operating system, it is important that the operator
knows how to use the console for this purpose.
To use the SYSCONS console on the HMC, you must select the Operating System
Messages (OSM) task and the appropriate system on the HMC. The HMC will open a window
which will be the SYSCONS console for the system. During an IPL process, the messages
are automatically displayed on the SYSCONS console. If there are any replies required
during the NIP portion of the IPL, the operator can reply using the Respond button on the
window, as shown in Figure 14-13. If you need to use the SYSCONS console for command
processing, you can use the Send button to send a command to z/OS. You must first enter the
VARY CN(*),ACTIVATE command, as shown in the command line of Figure 14-13, to allow the
SYSCONS console to send commands and receive messages.
Figure 14-13 Activate the SYSCONS
This command can only be entered at the SYSCONS console. If you try to enter any other
z/OS command prior to this command, you receive a reply stating that you must enter the
VARY CONSOLE command to enable system console communications, as shown in
292
Figure 14-14 Command rejected
As a result of entering this command, and if z/OS is able to establish communication with the
SYSCONS console, there is a response to indicate that the vary processing is complete, as
Figure 14-15 Console activation complete
Messages are now displayed from systems as specified in the MSCOPE parameter for the
SYSCONS. Also, almost any z/OS command can now be entered, with a few restrictions. If
there is no response to the command, it may indicate that the system is not active, or the
interface between z/OS and the Support Element (SE) is not working. There is also the
possibility that the ROUTCDE setting on the SYSCONS is set to NONE. You can check this
293
by using the D C,M command; if the ROUTCDE parameter is not set to ALL and you want to
see all the messages for the system or sysplex, then enter;
V CN(*),ROUT=ALL
The SYSCONS console would normally only be used when IPLing systems with no local
attached consoles to complete the IPL, or when messages are sent to it from z/OS in
recovery situations.
Although the SYSCONS console for a system may be accessed on multiple HMCs, you do
not have to issue the VARY CONSOLE command on each HMC. It only needs to be entered once
for the system. It remains active for the duration of the IPL, or until the VARY CN,DEACT
command (to deactivate the system console) is entered.
To display the SYSCONS console status for all systems in the sysplex, use the DISPLAY
CONSOLE command as shown Figure 14-16.
D C,M
MSG: CURR=0
LIM=20
SYS=AAIL
PFK=00
CONSOLE
ID --------------- SPECIFICATIONS --------------. . .
AAILSYSC
COND=A,PD
AUTH=MASTER
SYSCONS
MFORM=M
LEVEL=ALL,NB
AAIL
ROUTCDE=NONE
CMDSYS=AAIL
MSCOPE=*ALL
AUTOACT=-------INTIDS=N UNKNIDS=N
. . .
Figure 14-16 Display the SYSCONS console
For each system that is active or has been active in the sysplex since the sysplex was
initialized, there is a SYSCONS console status displayed, along with all the other consoles in
the sysplex. The COND status for the SYSCONS has three possible values. They are:
A
The system is active in the sysplex, but the SYSCONS is not available.
A,PD
The SYSCONS is available, and in Problem Determination mode.
The system is not active in the sysplex, therefore no SYSCONS is available.
The status of A indicates that the system has been IPLed, but there has not been a VARY
CONSOLE command issued from the SYSCONS for the system. The status of A,PD indicates
that the system is IPLed, and there has been a VARY CONSOLE command issued from the
SYSCONS for the system. The status of N indicates that the system associated with the
SYSCONS is not active in the sysplex. The console appears in the list because the system
had been active in the sysplex. It could also indicate that the interface between z/OS and the
SE is not working, although this would be rare.
There is an MSCOPE parameter for this console, and it should be set appropriately. See
14.2.5, MSCOPE implications on page 289 for further details.
294
14.5 Console buffer shortages

A console buffer shortage condition may occur more frequently in a sysplex environment
because of the increased number of messages being sent to certain consoles, especially
consoles with an MSCOPE=*ALL.
You can use the DISPLAY CONSOLE,BACKLOG command to determine the console buffer
conditions, as shown in Figure 14-17. In this example there is no buffer shortage and the
number of write to operator (WTO) message buffers in use is zero (0) 1. This is what you
would expect to see under normal conditions.
D C,B
MSG: CURR=0 1 LIM=3000 RPLY:CURR=3
LIM=999 SYS=#@$3
PFK=00
CONSOLE
ID --------------- SPECIFICATIONS --------------NO CONSOLES MEET SPECIFIED CRITERIA
WTO BUFFERS IN CONSOLE BACKUP STORAGE =
0
ADDRESS SPACE WTO BUFFER USAGE
NO ADDRESS SPACES ARE USING MORE THAN 1000 WTO BUFFERS
MESSAGES COMING FROM OTHER SYSTEMS - WTO BUFFER USAGE
NO WTO BUFFERS ARE IN USE FOR MESSAGES FROM OTHER SYSTEMS
Figure 14-17 Console buffer display - normal
In the event of a WTO buffer shortage, you can use the DISPLAY CONSOLE,BACKLOG command
to determine the console buffer conditions, as shown in Figure 14-18. The command will
display details about the affected console and will also display any jobs using more than 1000
console buffers. This can help to determine the most appropriate corrective action.
D C,B
MSG: CURR=****1 LIM=3000 RPLY:CURR=3
LIM=999 SYS=#@$3
PFK=00
CONSOLE
ID --------------- SPECIFICATIONS --------------#@$3M01 2
13 COND=A
AUTH=MASTER
NBUF=3037 3
08E0
AREA=Z
MFORM=T,S,J,X
#@$3
DEL=N
RTME=1/4
RNUM=20
SEG=20
CON=N
USE=FC
LEVEL=ALL
PFKTAB=PFKTAB1
ROUTCDE=ALL
LOGON=OPTIONAL
CMDSYS=#@$3
MSCOPE=*ALL
INTIDS=N UNKNIDS=N
WTO BUFFERS IN CONSOLE BACKUP STORAGE =
18821
4
ADDRESS SPACE WTO BUFFER USAGE
ASID = 002F
JOBNAME = TSTWTO2 NBUF =
7432
5
ASID = 002E
7231
ASID = 0030
7145
MESSAGES COMING FROM OTHER SYSTEMS - WTO BUFFER USAGE
SYSTEM = #@$2
NBUF =
20
6
SYSTEM = #@$1
NBUF =
20
Figure 14-18 Console buffer display - buffer shortage
295
1 The number of write-to-operator (WTO) message buffers in use by the system at this time. If
the number is greater than 9999, asterisks (*) will appear.
2 The name of the console experiencing a buffer shortage.
3 The number of WTO message buffers currently queued to this console. If the number is
greater than 9999, asterisks (*) will appear.
4 All WTO buffers are in use, and the communications task (COMMTASK) is holding WTO
requests until the WTO buffer shortage is relieved. The number shown is the number of WTO
requests that are being held.
5 This shows the address space that is using more than 33% of the available WTO buffers.
The NBUF shows the number of WTO buffers in use by the specified ASID and job. In this
case, it may be appropriate to cancel the jobs sending the large number of messages to the
console.
6 Messages coming from other systems in the sysplex are using WTO message buffers. This
shows each system that has incoming messages in WTO buffers. The system name and the
number of buffers being used for messages from that system is shown.
There are a number of actions that can be attempted to try to relieve a WTO buffer shortage
condition. The console with the buffer shortage may not be local to an operations area, so
physically resolving some issues may not be an option.
Here are suggested actions to help relieve a buffer shortage:
Respond to any WTOR requesting an operator action.
Re-route the messages to another console by entering:
K Q,R=consname1,L=consname2
Here consname1 is the name of the console to receive the re-routed messages, and
consname2 is the name of the console whose messages are being rerouted. This only
reroutes messages already in the queue for the console.
It may be appropriate to cancel any jobs, identified using the D C,B command, that are
using a large number of buffers. The job or jobs may be flooding the console with
messages, and cancelling the job may help relieve the shortage.
Determine if there are outstanding action messages by using the DISPLAY REPLIES
D R,L,CN=(ALL)
IEE112I 11.11.45 PENDING REQUESTS 894
RM=3
IM=36
CEM=18
EM=0
RU=0
IR=0
AMRF
ID: R/K
T TIME
SYSNAME JOB ID
MESSAGE TEXT
. . .
27065 C 09.19.14 SC55
*ATB052E LOGICAL UNIT SC55HMT
FOR TRANSACTION SCHEDULER ASCH
NOT ACTIVATED IN THE APPC
CONFIGURATION. REASON CODE =
5A.
27063 C 09.18.08 SC54
*ATB052E LOGICAL UNIT SC54HMT
FOR TRANSACTION SCHEDULER ASCH
NOT ACTIVATED IN THE APPC
CONFIGURATION. REASON CODE =
5A.
. . .
Figure 14-19 Display outstanding messages
296
You can then use the K C (control console) command to delete the outstanding action
messages that the action message retention facility (AMRF) has retained. An example of
this is shown in Figure 14-20.
K C,A,27063-27065
IEE146I K COMMAND ENDED-2 MESSAGE(S) DELETED
Figure 14-20 Deleting outstanding action messages
Vary the particular console offline by entering:

VARY CN(consname),OFFLINE
This would release the console buffers for that particular console. However, it may only
temporarily relieve the problem and may not resolve the underlying cause of the buffer
shortage, because any heavy message traffic may be rerouted to the next console in the
console group.
Change the value of message buffers if necessary. There are three types of message
buffers. They are:
Write-to-operator (WTO) buffers
When the number of WTO buffers reaches this number, the system places any
program that issues a WTO into a wait until the number of WTO buffers decreases to a
value less than the limit. This is represented as MLIM as seen in message IEE144I in
Figure 14-21.
Write-to-operator-with-reply (WTOR) buffer
The current limit of outstanding WTOR messages that the system or sysplex can hold
in buffers. When the number of WTOR buffers reaches this value, the system places
any program that issues a WTOR into a wait until the number of WTOR buffers
decreases to a value less than the limit. This is represented as RLIM as seen in
message IEE144I in Figure 14-21.
Write-to-log (WTL) as to the SYSLOG buffer
The current limit of messages that can be buffered to the SYSLOG processor. When
the number of messages buffered up for the SYSLOG processor reaches this value,
subsequent messages to be buffered to the SYSLOG processor will be lost until the
number of buffered messages decreases to a value less than the limit. This is
represented as LOGLIM as seen on message IEE144I in Figure 14-21.
To determine the current limit for each buffer, use the K M (control message) command.
K M,REF
IEE144I K M,AMRF=N,MLIM=3000,RLIM=0999,UEXIT=N,LOGLIM=006000,
ROUTTIME=005,RMAX=0999
Figure 14-21 Control message command
You can also use this command to change the limits of each buffer. For example,
Figure 14-22 shows the K M command to change the WTO message limit.
K M,MLIM=8000
IEE712I CONTROL
PROCESSING COMPLETE
Figure 14-22 Change WTO buffers - MLIM
297
Increasing the limits specified may require the use of more private storage in the console
address space (for MLIM) and ECSA for RLIM and LOGLIM, which may create other
system performance concerns. The maximum values of each type of buffer are listed in
Table 14-1.
Table 14-1 Maximum console buffer values
Type
Parameter
Maximum
WTO
MLIM
9999
WTOR
RLIM
9999
WTL
LOGLIM
999999
Deleting the message queue by using a K Q command is also an option. Use this
command to delete messages that are queued to an MCS or SMCS console (not EMCS).
This action affects only messages currently on the console's queue. Subsequent
messages are queued as usual, so the command may need to be issued a number of
times. Remember that the messages deleted from the queue will be lost. Use the following
command:
K Q[,L=consname]
Here L=consname is the name of the console whose message queue is to be deleted, or
blank defaults to the console where the command is issued.
14.6 Entering z/OS commands

When operating in a sysplex environment, it is possible to enter commands for execution on
any system or systems in the sysplex. This section examines the various methods that you
can use to accomplish this.
14.6.1 CMDSYS parameter

Each console has a CMDSYS parameter specified when it is connected to a system in a
sysplex. Although this parameter is usually pointing to the system where the console is
assigned, or logged on to, it can point to any system in the sysplex. The CMDSYS parameter
specifies the name of the system for which this console has command association, which
means any command issued from the console will be issued to the system assigned on the
CMDSYS parameter. This system might be different from the system where this console is
physically attached.
Some systems do not have a console physically connected to them. In this case, it may make
sense to set up a console on another system with a CMDSYS parameter associated to the
system with no consoles attached. All commands entered on a console will be executed on
the system specified in the CMDSYS parameter, as long as no other routing options are
used.
To determine the CMDSYS parameter of an MCS console, view the IEE612I message line on
the bottom of the console display (MCS and SMCS consoles); see Figure 14-23.
IEE612I CN=#@$3M01
DEVNUM=08E0 SYS=#@$3
Figure 14-23 CMDSYS parameter on the console
298
CMDSYS=#@$3
If you are using an EMCS console, you must use the DISPLAY CONSOLE command to
determine the CMDSYS setting. To alter the CMDSYS value for your console, use the
CONTROL VARY command. For example, to change the CMDSYS value for console #@$3M01
to #@$1, enter:
K V,CMDSYS=#@$1,L=#@$3M01
There is no response to this command on the console. The IEE612I message is updated with
the new CMDSYS value. The L= parameter is only required if the change is to be made on a
console where the command is not entered.
IEE612I CN=#@$3M01
DEVNUM=08E0 SYS=#@$3
CMDSYS=#@$1
Figure 14-24 CMDSYS parameter on the console after change
This change will remain in effect until the next IPL, or until another CONTROL command is
entered. If you want this change to be permanent, the system programmer would need to
change the CONSOLxx parmlib member.
14.6.2 Using the ROUTE command

The z/OS ROUTE command routes an operator command for execution on one or more
systems in the sysplex. The response to the command is returned to the issuing console
unless redirected by an L= parameter. The ROUTE command allows the operator to specify
which system or systems will execute the command using various operands. The response to
the command is displayed on the console where the command is issued.
The following examples use the DISPLAY TIME command as the command to be executed on
a system or systems. In the first example, the console used has a CMDSYS value of #@$3
and the command is routed to execute on #@$1.
#@$3
RO #@$1,D T
IEE136I LOCAL: TIME=19.15.22 DATE=2007.197
TIME=23.15.22 DATE=2007.197
UTC:
Figure 14-25 ROUTE command to one system
You are not limited to just one system name when using the ROUTE command. An example of
executing a command on more than one system in the sysplex is shown in Figure 14-26.
#@$3
RO (#@$1,#@$2),D T
#@$3
IEE421I RO (LIST),D T 471
SYSNAME RESPONSES --------------------------------------------------#@$1
IEE136I LOCAL: TIME=19.57.18 DATE=2007.197 UTC:
TIME=23.57.18 DATE=2007.197
#@$2
TIME=23.57.18 DATE=2007.197
Figure 14-26 ROUTE command to multiple systems
To execute a command on all systems in the sysplex without listing all the system names in
the ROUTE command, you can enter:
RO *ALL,D T
299
This will result in a response from each system active in the sysplex.
To execute a command on all other systems in the sysplex except the one specified in the
SYS field in the IEE612I message, you can enter:
RO *OTHER,D T
System grouping
Another way of using the ROUTE command is to use a system group name. The system group
names can be set up by the system programmer using the IEEGSYS sample program and
procedure in SYS1.SAMPLIB.
In our examples, we have defined system groups as shown in Figure 14-27.
GROUP(TEST)
GROUP(DEVL)
GROUP(PROD)
NAMES(#@$1)
NAMES(#@$1,#@$2)
NAMES(#@$2,#@$3)
Figure 14-27 System grouping parameter
All active systems included in the system group will execute the command; see Figure 14-28.
These group names can be used to route commands to the active systems in the group.
#@$3
RO PROD,D T
#@$3
IEE421I RO PROD,D T 161
SYSNAME RESPONSES --------------------------------------------------#@$2
TIME=03.17.23 DATE=2007.198
#@$3
Figure 14-28 System group command
You can include more than one system group name, or include both system group names
and system names when using the ROUTE command. Be sure to put multiple names in
brackets with a comma (,) separating each name, for example:
RO (DEVL,#@$3),D T
Here DEVL is a system group and #@$3 is a system name that is outside the defined system
group, but is still in the sysplex.
14.6.3 Command prefixes

Another method of controlling where commands will be executed is to use command
prefixing. For command prefixes to be used, they need to be set up by the system
programmer using the IEECMDPF sample program and procedure in SYS1.SAMPLIB. This
program sample program defines a command prefix equal to the system name.
You may already be familiar with the JES2 or JES3 prefixes used on your systems. Additional
prefixes can be registered on a system that will define a system with a prefix using the
IEECMDPF program. You can determine the command prefixes that are registered on your
system by using the DISPLAY OPDATA command. Figure 14-29 on page 301 shows that the
command prefix entries are defined with an owner of IEECMDPF.
300
D OPDATA
PREFIX
OWNER
SYSTEM
SCOPE
$
JES2
#@$2
SYSTEM
$
JES2
#@$1
SYSTEM
$
JES2
#@$3
SYSTEM
%
RACF
#@$2
SYSTEM
%
RACF
#@$1
SYSTEM
%
RACF
#@$3
SYSTEM
#@$1
IEECMDPF
#@$1
SYSPLEX
#@$2
IEECMDPF
#@$2
SYSPLEX
#@$3
IEECMDPF
#@$3
SYSPLEX
REMOVE
NO
NO
NO
NO
NO
NO
YES
YES
YES
FAILDSP
SYSPURGE
SYSPURGE
SYSPURGE
PURGE
PURGE
PURGE
SYSPURGE
SYSPURGE
SYSPURGE
Figure 14-29 Display command prefix
When a command prefix has been defined, a command can be prefixed with the appropriate
system prefix to have a command executed on that system, without having to prefix the
command with the ROUTE command. For example, in Figure 14-30 we used the prefix #@$1 to
send a command to system #@$1 from a console on #@$3.
#@$3
#@$1 D T
#@$1
IEE136I LOCAL: TIME=23.40.36 DATE=2007.197
TIME=03.40.36 DATE=2007.198
UTC:
Figure 14-30 Command prefix output
There is no requirement to put a space between the prefix and the beginning of the
command.
14.7 Message Flood Automation

Some z/OS systems experience cases where a user program or a z/OS process itself issues
a large number of messages to the z/OS consoles in a short time. For example, a user
program may enter an unintentional loop that includes a WTO call. Hundreds (or even
thousands) of messages a second are not uncommon.
These messages are often very similar or identical, but are not necessarily so. Techniques to
identify similar messages can be very difficult and time-consuming. Message Flood
Automation (MFA) has been developed to help address this problem. Its intention is, where
possible, to identify runaway WTO conditions that can cause severe disruptions to z/OS
operation and to take installation-specified actions in these cases.
From z/OS V1.9 onward, the Consoles component has included the Message Flood
Automation function. This function was also made available via APAR OA17514 for z/OS
V1.6 and higher.
Message Flood Automation provides specialized, policy-driven automation for dealing with
high volumes of messages occurring at very high message rates. The policy can be set in a
parmlib member, and then examined and modified through operator commands. The policy
specifies the types of messages that are to be monitored, the criteria for establishing the
onset and ending of a message flood, and the actions that may be taken if a flood occurs.
301
Multiple levels of policy specification allow criteria and actions to be applied to message
types, jobs, or even individual message IDs. The actions that may be taken during a message
flood include:
Preventing the flood messages from being displayed on a console.
Preventing the flood messages from being logged in the SYSLOG or OPERLOG.
Preventing the flood messages from being queued for automation.
Preventing the flood messages from propagating to other systems in a sysplex (if the
message is not displayed, logged or queued for automation).
Preventing the flood messages from being queued to the Action Message Retention
Facility (AMRF) if the message is an action message.
Taking action against the address space issuing the flood messages, by issuing a
command (typically a CANCEL command).
Message Flood Automation display commands

Some examples of the DISPLAY MSGFLD (or D MF) Message Flood Automation display
commands are shown in this section.
The STATUS keyword can be specified to display the current enablement status of Message
Flood Automation as well as the name of the currently active MSGFLDxx parmlib member, as
D MSGFLD,STATUS
MSGF042I Message Flood Automation V1R2M05 10/15/04 ENABLED.
Policy INITIALIZED.
Using PARMLIB member: MSGFLD00
Intensive modes: REGULAR-OFF ACTION-OFF SPECIFIC-OFF
Message rate monitoring DISABLED.
0 msgs
0 secs
Figure 14-31 Display Message Flood Automation status
The PARAMETERS keyword can be specified to display the current values of all of the
parameters for all of the msgtypes. The msgtypes are either regular, action, or specific, as
D MSGFLD,PARAMETERS
MSGF901I Message Flood Automation parameters
Message type REGULAR
ACTION SPECIFIC
MSGCOUNT =
5
22
8
MSGTHRESH =
30
30
10
JOBTHRESH =
30
30
INTVLTIME =
1
1
1
JOBIMTIME =
2
2
SYSIMTIME =
2
2
5
NUMJOBS
=
10
10
Figure 14-32 Display Message Flood Automation parameters
The DEFAULTS keyword can be specified to display the current default actions to be taken for
all of the msgtypes, as shown in Figure 14-33 on page 303.
302
D MSGFLD,DEFAULTS
MSGF904I Message Flood Automation DEFAULTS
Message type REGULAR ACTION SPECIFIC
LOG
=
Y
Y
N
AUTO
=
Y
Y
N
DISPLAY
=
N
N
N
CMD
=
N
N
RETAIN
=
N
N
Figure 14-33 Display Message Flood Automation defaults
The JOBS keyword can be specified to display the current default actions to be taken for all of
the jobs that have been defined in the active MSGFLDxx parmlib member, as shown in
Figure 14-34.
D MSGFLD,JOBS
MSGF905I Message Flood Automation JOB actions
REGULAR messages LOG AUTO DISPLAY CMD RETAIN
JOB D1%%MSTR Y
N
N
N
ACTION messages LOG AUTO DISPLAY CMD RETAIN
JOB D2%%MSTR Y
N
N
N
N
Figure 14-34 Display Message Flood Automation jobs
The MSGS keyword can be specified to display the current default actions to be taken for all of
the messages that have been defined in the active MSGFLDxx parmlib member, as shown in
Figure 14-35.
D MSGFLD,MSGS
MSGF906I Message Flood Automation MSG actions
SPECIFIC messages LOG AUTO DISPLAY CMD RETAIN
MSG IOS000I
N
N
N
N
MSG IOS002A
N
N
N
N
MSG IOS291I
N
N
N
N
MSG IEA476E
N
N
N
N
MSG IEA491E
N
N
N
N
MSG IEA494I
N
N
N
N
MSG IEA497I
N
N
N
N
MSG IOS251I
N
N
N
N
MSG IOS444I
N
N
N
N
MSG IOS450E
N
N
N
N
. . .
Figure 14-35 Display Message Flood Automation messages
The MODE keyword can be specified to display the current intensive mode states for the three
message types, as shown in Figure 14-36.
D MSGFLD,MODE
MSGF040I Intensive modes: REGULAR-OFF
ACTION-OFF
SPECIFIC-OFF
Figure 14-36 Display Message Flood Automation mode
303
14.8 Removing consoles using IEARELCN or IEARELEC

The current limit is 99 consoles in a sysplex, so it is possible to eventually run out of consoles.
Some customers have performed an unwanted IPL to recapture orphan consoles.
To remove a console definition for MCS and SMCS consoles, you can use the sample JCL
for program IEARELCN in SYS1.SAMPLIB. The use of the IEARELCN program is described
in MVS Planning: Operations, SA22-7601. Similarly, to remove a console definition for EMCS
consoles, you can use the sample JCL for program IEARELEC in SYS1.SAMPLIB.
In a sysplex, deleting a console definition releases the console ID associated with the console
and makes it available for other console definitions. Thus, you have flexibility controlling the
number of console IDs you need in an active console configuration. For example, if you
define 10 consoles in CONSOLxx and you have used the VARY CONSOLE OFFLINE command
for one of the consoles (so it is inactive), the system still associates the console ID with the
inactive console. Using the console service, you can delete the console definition, thereby
making the console ID available for reuse. When you add a new console, the system
reassigns the console ID. This action would need to be done by a system programmer.
14.9 z/OS Management Console

There is also a graphical user interface (GUI) for monitoring your systems, and it is known as
the IBM Tivoli OMEGAMON z/OS Management Console. It displays and collects health
check and availability information about z/OS systems and sysplex resources, and reports
the information in the Tivoli Enterprise Portal GUI. The product workspaces provide health
check information for z/OS systems, and configuration status information for z/OS systems
and sysplex resources. The user interface contains expert advice on alerts.
The OMEGAMON z/OS Management Console contains a subset of the availability and
sysplex monitoring functions of the IBM Tivoli OMEGAMON XE on z/OS product. In addition,
it uses IBM Health Checker for z/OS (this must also be active on your systems) to monitor
systems for potential problems, and it monitors Health Checker for z/OS for problems.
The OMEGAMON z/OS Management Console displays the following types of z/OS data:
Availability
304
Operations status, including outstanding WTOR and WTO buffers remaining

Address space data
Coupling facility policy information
Coupling facility systems data
Coupling facility structures data
Coupling facility structure connections data
Coupling facility paths data
LPAR clusters data
UNIX System Services address spaces and processes data
Paging data set data
Cross-system coupling facility (XCF) systems data
XCF paths data
IBM Health Checker for z/OS data

Including status of the product and checks. The checks performed by the IBM Health
Checker for z/OS identify potential problems before they affect your availability or, in
worst cases, cause outages. IBM Health Checker for z/OS periodically runs the
element, product, vendor, or installation checks to look at the current active z/OS and
sysplex settings and definitions for a system, and compares the values to those
suggested by IBM or defined by you. See Chapter 12, IBM z/OS Health Checker on
page 257 for more detailed information about this topic.
The Tivoli OMEGAMON z/OS Management Console has a Java-based interface known as
the Tivoli Enterprise Portal, which you can access via a Web browser. You can set threshold
levels and flags as desired to alert you when the systems reach critical points. Refer to IBM
OMEGAMON z/OS Management Console Users Guide, GC32-1955, for more details.
Workspace
A workspace is the work area of the Tivoli Enterprise Portal application window. It is
comprised of one or more views. A view is a pane in the workspace (typically a chart, graph,
or table) showing data collected by a monitoring agent. As you select items in the Navigator,
each workspace presents views relevant to your selection. Every workspace has at least one
view, and every view has a set of properties associated with it. You can customize the
workspace by working in the Properties Editor to change the style and content of each view.
You can also change, add, and delete views on a workspace.
An example of the sysplex information available from the z/OS Management Console is the
Coupling Facility Systems Data for Sysplex workspace report shown in Figure 14-37. This
report displays status and storage information about the Coupling Facilities defined to the
sysplex. This workspace contains views such as;
The Dump Table Storage bar chart, which shows each Coupling Facility; the number of
4 K pages of storage reserved for dump tables; the number of pages currently holding
dumps; and the percentage of allocated storage currently being used.
The Coupling Facility Systems Information table displays basic status and storage
statistics for each Coupling Facility. From this table, you can link to the other workspaces
for the selected Coupling Facility
Figure 14-37 z/OS Management Console - CF view
305
Another example of the information available in the z/OS Management Console is the z/OS
Health Checker information. An example of this is the Health Monitor Checks workspace,
which provides a summary of information about each health check. This workspace displays
data provided by the Health Checker Checks attribute group, as seen in Figure 14-38.
Figure 14-38 z/OS Management Console - Health Checker view
The Health Monitor Checks workspace contains views such as:

The Exception Check Counts bar chart, which shows the number of exceptions detected
in the most recent iteration of each check.
The Run Counts bar chart, which shows the number of times each check has been
invoked since the last time it was initialized or refreshed.
The Health Checker Checks table, which displays identifying information, status, and
other information about each check. From this table, you can link to the Check Messages
workspace for the selected check.
306
15
Chapter 15.
z/OS system logger

considerations
This chapter describes the operational aspects of the system logger, including:
Starting and stopping the system logger address space
Displaying system logger information; structures and log streams
System logger offload monitoring
Handling a shortage of system logger directory extents
System logger structure rebuilds
Logrec use of log streams as an exploiter of system logger
For more detailed information about the z/OS system logger, see Systems Programmers
Guide to: z/OS System Logger, SG24-6898, and MVS Setting Up a Sysplex, SA22-7625.
307
15.1 Introduction to z/OS system logger

The z/OS system logger is a z/OS component designed to support system and subsystem
components in a Parallel Sysplex. It implements a set of services that enables applications to
write, read, and delete log data into what is known as a log stream. A log stream is a
sequential series of log records written by the system logger at the request of a log writer (an
exploiter like CICS). The records are written in the order of their arrival and may be retrieved
sequentially, forward or backward, or uniquely, by a log token (key).
System, subsystem, or application components can exploit the system logger functions. The
system logger takes the responsibility for tasks such as saving log data, retrieving the data
(potentially from any system in the sysplex), archiving the data, and expiring the data. In
addition, system logger provides the ability to have a single, merged log, containing log data
from multiple instances of an application within the sysplex.
Log data managed by the system logger may reside in processor storage, in a Coupling
Facility structure, on DASD, or potentially, on tape. However, regardless of where system
logger is currently storing a given log record, from the point of view of the exploiter, all the log
records are kept in a single file that is a limited size.
The task of tracking where a specific piece of log data is at any given time is handled by
system logger. Additionally, system logger will manage the utilization of its storage; as the
space in one medium starts filling up (a Coupling Facility structure, for example), logger will
move old data to the next level in the hierarchy (an offload dataset on DASD, for example).
The location of the data, and the migration of that data from one level to another, is
transparent to the application and is managed completely by system logger, as illustrated in
Figure 15-1.
Figure 15-1 Logical and physical views of system logger-maintained log data
308
There are basically two types of users of system logger. One type of exploiter uses the
system logger as an archival facility for log data; for example, OPERLOG or LOGREC. The
second type of exploiter typically uses the data more actively, and explicitly deletes it when it
is no longer required; for example, the CICS DFHLOG. CICS stores information in DFHLOG
about running transactions, and deletes the records as the transactions complete. These are
called active exploiters.
15.1.1 Where system logger stores its data

When an application passes log data to system logger, the data can initially be stored on
DASD, in what is known as a DASD-only log stream, or it can be stored in a Coupling Facility
(CF) in what is known as a CF-Structure log stream. The major differences between these
two types of log stream configurations are the storage medium system logger uses to hold
interim log data. Interim storage is the primary storage used to hold log data that has not yet
been offloaded. Another major difference is how many systems can use the log stream
concurrently;
In a CF log stream, interim storage for log data is in CF list structures. This type of log
stream supports the ability for exploiters on more than one system to write log data to the
same log stream concurrently. Log data that is in interim storage is duplexed to protect
against data loss conditions. This data is usually duplexed to a data space, although log
streams residing in a CF structure may optionally be duplexed to a staging data set, or
utilize system managed duplexing.
In a DASD-only log stream, interim storage for log data is contained in a data space in the
z/OS system. The dataspaces are associated with the system logger address space,
IXGLOGR. DASD-only log streams can only be used by exploiters on one system at a
time. Log data that is in interim storage is duplexed to protect against data loss conditions,
which for DASD-only log streams is usually to a staging data set.
15.2 Starting and stopping the system logger address space

The system logger address space (IXGLOGR) starts automatically, as an MVS system
component address space, during an IPL on each system image.
If the IXGLOGR address space fails for some reason the system will automatically restart the
address space, unless it is terminated using the FORCE IXGLOGR,ARM command.
If for any reason, you need to terminate the system logger address space, you will need to
issue the FORCE IXGLOGR,ARM command. The only way to restart the system logger address
space is through the S IXGLOGRS procedure. After a FORCE command has been issued
against the system logger address space, the system issues IXG056I and IXG067E
messages to prompt the operator to manually restart the system logger address space, as
IXG056I SYSTEM LOGGER ADDRESS SPACE HAS ENDED. 171
OPERATOR ISSUED FORCE COMMAND. MANUAL RESTART REQUIRED.
IXG067E ISSUE S IXGLOGRS TO RESTART SYSTEM LOGGER.
Figure 15-2 System logger address space restart
IXGLOGRS is the command processor to start the system logger address space. IXGLOGRS
only starts the system logger address space (IXGLOGR) and then it immediately ends.
Chapter 15. z/OS system logger considerations
309
While an application is connected to a log stream, the supporting instance of the z/OS system
logger might fail independently of the exploiting application. When the z/OS system logger
address space fails, connections to log streams are automatically disconnected by the
system logger. All requests to connect are rejected. When the recovery processing
completes, the system logger is restarted and an Event Notification Facility (ENF) is
broadcast. On receipt of the ENF, applications may connect to log streams and resume
processing. During startup, system logger runs through a series of operations for all CF
structure-based log streams to attempt to recover and clean up any failed connections, and to
ensure that all data is valid.
15.3 Displaying system logger status

The display logger command can be used to determine the operational status of system
logger, the status of individual log streams from a local and sysplex view, and the utilization of
CF list structures. The command output is delivered via system logger message IXG601I.
The display logger command syntax is shown in Figure 15-3. The D LOGGER command has
sysplex scope when you use either L or C,SYSPLEX options.
D LOGGER[,{STATUS|ST}
]
[,{CONNECTION|CONN|C}[,LSNAME|LSN=logstreamname] ]
[,Jobname|JOB|J=mvsjobname]
[,{SUMM|S }
]
{Detail|D}
[,DASDONLY
]
[,SYSPLEX
]
[,{LOGSTREAM|L}[,LSName=logstreamname
]
[,STRNAME|STRN=structurename]
[,DASDONLY
]
[,{STRUCTURE|STR}[,STRNAME|STRN=structurename]
[,L={a|name|name-a}
Figure 15-3 Display Logger command syntax
Note: An asterisk (*) can be used as a wildcard character with the DISPLAY LOGGER
command. Specify an asterisk as the search argument, or specify an asterisk as the last
character of a search argument. If used, the wildcard must be the last character in the
search argument, or the only character.
To display the current operational status of the system logger, use the D LOGGER,ST
D LOGGER,ST
IXG601I
18.54.39 LOGGER DISPLAY 088
SYSTEM LOGGER STATUS
SYSTEM
SYSTEM LOGGER STATUS
------------------------#@$3
ACTIVE
Figure 15-4 Display Logger status
310
To check the state of a log stream and the number of systems connected to the log stream,
use the D LOGGER,LOGSTREAM command. The amount of output displayed will depend on the
number of log streams defined in the LOGR policy. See Figure 15-5 for an example of the
output.
D LOGGER,LOGSTREAM
IXG601I
INVENTORY INFORMATION BY LOGSTREAM
LOGSTREAM 1
STRUCTURE 2
----------------#@$C.#@$CCM$1.DFHLOG2
CIC_DFHLOG_001
#@$C.#@$CCM$1.DFHSHUN2
CIC_DFHSHUNT_001
#@$C.#@$CCM$2.DFHLOG2
CIC_DFHLOG_001
#@$C.#@$CCM$2.DFHSHUN2
CIC_DFHSHUNT_001
. . .
#@$3.DFHLOG2.MODEL
CIC_DFHLOG_001
#@$3.DFHSHUN2.MODEL
CIC_DFHSHUNT_001
ATR.#@$#PLEX.DELAYED.UR
RRS_DELAYEDUR_1
SYSNAME: #@$1
DUPLEXING: LOCAL BUFFERS
SYSNAME: #@$2
SYSNAME: #@$3
GROUP: PRODUCTION
. . .
IGWTV003.IGWSHUNT.SHUNTLOG LOG_IGWSHUNT_001
IGWTV999.IGWLOG.SYSLOG
*DASDONLY*
ING.HEALTH.CHECKER.HISTORY LOG_SA390_MISC
SYSPLEX.LOGREC.ALLRECS
SYSTEM_LOGREC
SYSPLEX.OPERLOG
SYSTEM_OPERLOG
. . .
#CONN
-----000000
000000
000000
000000
STATUS
-----AVAILABLE
AVAILABLE
AVAILABLE
AVAILABLE
000000 *MODEL*
000000 *MODEL*
000003 IN USE
3
4
5
6
7
000000
000000
000000
000000
000003
AVAILABLE
AVAILABLE
AVAILABLE
LOSS OF DATA
IN USE
Figure 15-5 Display Logger Logstream
1 Logstream name.
2 Structure name defined in the CFRM policy or *DASDONLY* when a DASD only configured
log stream is displayed.
3 The number of active connections from this system to the log stream, and the log stream
status.
Some examples of the status are:
4 Available: The log stream is available for connects.
5 Model: The log stream is a model and is exclusively for use with the LIKE parameter to set
up general characteristics for other log stream definitions.
6 In use: The log stream is available and has a current connection.
7 DUPLEXING: LOCAL BUFFERS indicates that the duplex copy of log stream resides in
system loggers data space.
8 Loss of data: There is a loss of data condition present in the log stream.
311
To display all defined log streams that have a DASD-only configuration, use the D
LOGGER,L,DASDONLY command. See Figure 15-6 for an example of the output.
D LOGGER,L,DASDONLY
IXG601I
LOGSTREAM
STRUCTURE
----------------BDG.LOG.STREAM
*DASDONLY*
IGWTV999.IGWLOG.SYSLOG
*DASDONLY*
#CONN
-----000000
000000
STATUS
-----AVAILABLE
AVAILABLE
Figure 15-6 Display Logger DASD only log stream
To check the number of connections to the log streams, use the D LOGGER,CONN command.
This command displays only log streams that have connectors on the system where the
command has been issued. See Figure 15-7 for an example of the output.
D LOGGER,CONN
IXG601I
CONNECTION INFORMATION BY LOGSTREAM FOR SYSTEM #@$3
LOGSTREAM
STRUCTURE
#CONN STATUS
---------------------- -----ATR.#@$#PLEX.RM.DATA
RRS_RMDATA_1
000001 IN USE
ATR.#@$#PLEX.MAIN.UR
RRS_MAINUR_1
000001 IN USE
SYSPLEX.OPERLOG
SYSTEM_OPERLOG
000002 IN USE
RRS_DELAYEDUR_1 000001 IN USE
ATR.#@$#PLEX.RESTART
RRS_RESTART_1
000001 IN USE
#@$#.SQ.MSGQ.LOG
I#$#LOGMSGQ
000001 IN USE
#@$#.SQ.EMHQ.LOG
I#$#LOGEMHQ
000001 IN USE
Figure 15-7 Display Logger connections
To display which jobnames are connected to the log stream, you can use the D
LOGGER,CONN,LSN=<logstream>,DETAIL command. This command displays only those log
streams that have connectors on the system where the command has been issued. See
Figure 15-8 for an example of the output using the sysplex OPERLOG as the log stream
example.
D LOGGER,C,LSN=SYSPLEX.OPERLOG,DETAIL
IXG601I
CONNECTION INFORMATION BY LOGSTREAM FOR SYSTEM #@$3
LOGSTREAM
STRUCTURE
#CONN STATUS
---------------------- -----SYSPLEX.OPERLOG
SYSTEM_OPERLOG
000002 IN USE
DUPLEXING: STAGING DATA SET
STGDSN: IXGLOGR.SYSPLEX.OPERLOG.#@$3
VOLUME=#@$#W1 SIZE=004140 (IN 4K)
% IN-USE=001
GROUP: PRODUCTION
JOBNAME: CONSOLE
ASID: 000B
R/W CONN: 000000 / 000001
RES MGR./CONNECTED: *NONE*
/ NO
IMPORT CONNECT: NO
Figure 15-8 Display Logger R/W connections
312
To display which log streams are allocated to a particular structure, use the D LOGGER,STR
command. The display shows whether a log stream is defined to the structure and whether it
is connected. See Figure 15-9 for an example of the output.
D LOGGER,STR
IXG601I
INVENTORY INFORMATION BY STRUCTURE
STRUCTURE
GROUP
CONNECTED
--------------------CIC_DFHLOG_001
PRODUCTION
YES
NO
NO
#@$C.#@$CWC2A.DFHLOG2
NO
. . .
LOG_TEST_001
*NO LOGSTREAMS DEFINED*
N/A
RRS_ARCHIVE_2
*NO LOGSTREAMS DEFINED*
N/A
RRS_DELAYEDUR_1
PRODUCTION
YES
RRS_MAINUR_1
PRODUCTION
YES
RRS_RESTART_1
PRODUCTION
YES
RRS_RMDATA_1
PRODUCTION
ATR.#@$#PLEX.RM.DATA
YES
SYSTEM_LOGREC
PRODUCTION
SYSPLEX.LOGREC.ALLRECS
NO
SYSTEM_OPERLOG
PRODUCTION
SYSPLEX.OPERLOG
YES
Figure 15-9 Display Logger structures
15.4 Listing logstream information using IXCMIAPU

You can use the IXCMIAPU program to list additional information when reviewing possible
loss of data in a log stream, as reported in Figure 15-5 on page 311. Notify your system
programmer if any loss of data is identified.
The IXCMIAPU program can be used to list additional information about log streams. One of
the features of the IXCMIAPU program for system logger (DATA TYPE(LOGR)) is its reporting
ability. You can specify either LIST LOGSTREAM(lsname) or LIST STRUCTURE(strname),
depending on the type of specific results you are looking for.
Specifying LIST STRUCTURE(strname) DETAIL(YES), where strname is the CF list structure
name (wildcards are supported), generates a report listing the structure definition values, the
effective average buffer size, and the log streams defined to structures listed.
Specifying LIST LOGSTREAM(lsname) DETAIL(YES), where lsname is the log stream name
(wildcards are supported), generates a report listing all of the log streams matching the
portion of the name specified. The output includes the log stream definition, names of any
associated or possible orphan data sets, connection information, and structure definitions for
313
the CF Structure-based log streams. Without the DETAIL(YES) keyword, only the log stream
definition are reported in the sysout.
//LOGRLIST JOB (0,0),'LIST LOGR POL',CLASS=A,REGION=4M,
//
MSGCLASS=X,NOTIFY=&SYSUID
//STEP1
EXEC PGM=IXCMIAPU
//SYSPRINT DD
SYSOUT=*
//SYSABEND DD
SYSOUT=*
//SYSIN
DD
*
DATA TYPE(LOGR) REPORT(NO)
LIST LOGSTREAM NAME(SYSPLEX.LOGREC.ALLRECS) DETAIL(YES)
/*
Figure 15-10 JCL to list log steam data using IXCMIAPU
The output of the LOGR list report is shown in Figure 15-11 on page 315. The report shown is
of the LOGREC log stream which has possible loss of data.
The listing shows the following:
1 Log stream information about how the log stream was defined.
2 Timing of the possible loss of data.
3 Log stream connection information, which shows the systems connected to the log stream
and their connection status.
4 Offload data set name prefix. The data set is a linear VSAM data set.
5 Information about the offload data sets in the log stream, the sequence numbers, and the
date and time of the oldest record in the data set.
314
LOGSTREAM NAME(SYSPLEX.LOGREC.ALLRECS) STRUCTNAME(SYSTEM_LOGREC) LS_DATACLAS(LOGR24K) 1

LS_MGMTCLAS() LS_STORCLAS() HLQ(IXGLOGR) MODEL(NO) LS_SIZE(1024)
STG_MGMTCLAS() STG_STORCLAS() STG_DATACLAS() STG_SIZE(0)
LOWOFFLOAD(0) HIGHOFFLOAD(80) STG_DUPLEX(NO) DUPLEXMODE()
RMNAME() DESCRIPTION() RETPD(60) AUTODELETE(YES) OFFLOADRECALL(YES)
DASDONLY(NO) DIAG(NO) LOGGERDUPLEX(UNCOND) EHLQ(NO_EHLQ) GROUP(PRODUCTION)
LOG STREAM ATTRIBUTES:
POSSIBLE LOSS OF DATA, LOW BLKID: 0000000000A5D115, HIGH BLKID: 0000000200A5D115
LOW GMT: 06/12/07 06:06:21, WHEN GMT: 06/13/07 11:09:56
User Data:
0000000000000000000000000000000000000000000000000000000000000000
0000000000000000000000000000000000000000000000000000000000000000
3
LOG STREAM CONNECTION INFO:

SYSTEMS CONNECTED: 3
SYSTEM
NAME
-------#@$3
#@$1
#@$2
STRUCTURE
VERSION
---------------C0D636D2B8F5CDCC
C0D636D2B8F5CDCC
C0D636D2B8F5CDCC
CON
ID
-01
03
02
CONNECTION
VERSION
---------00010027
0003001B
00020021
CONNECTION
STATE
---------Active
Active
Active
LOG STREAM DATA SET INFO:

STAGING DATA SET NAMES: IXGLOGR.SYSPLEX.LOGREC.ALLRECS.<SUFFIX>
NUMBER OF STAGING DATA SETS: 0
DATA SET NAMES IN USE: IXGLOGR.SYSPLEX.LOGREC.ALLRECS.<SEQ#>
Ext.
----*00001
<SEQ#>
-------A0000001
A0000002
A0000003
Lowest Blockid
---------------000000000043777C
000000000086F62E
0000000200A5D115
Highest GMT
----------------05/25/07 13:46:36
06/12/07 06:06:21
07/03/07 04:56:38
Highest Local
Status
----------------- ------05/25/07 09:46:36
06/12/07 02:06:21
07/03/07 00:56:38 CURRENT
4
5
NUMBER OF DATA SETS IN LOG STREAM: 3

POSSIBLE ORPHANED LOG STREAM DATA SETS:
Figure 15-11 Output from list log steam data using IXCMIAPU
For both CF structure-based and DASD only log streams, system logger marks a log stream
as permanently damaged when it cannot recover log data from either DASD staging data sets
or the local buffers after a system, sysplex, or Coupling Facility failure. Applications are
notified of the damage via system logger services and reason codes. Recovery actions are
necessary only if warranted for the application. Notify your system programmer if any loss of
data is identified.
Important: Never delete offload data sets (except orphaned ones) manually. This will
cause an unrecoverable loss of data.
315
15.5 System logger offload monitoring

System logger starts the process of moving, or offloading, log data to offload data sets when
log data reaches its defined high threshold setting. System logger has a function that allows
installations to monitor system logger offloads. If an offload appears to be taking too long or if
it hangs, system logger will issue message IXG312E, as shown in Figure 15-12.
IXG312E OFFLOAD DELAYED FOR logstream, REPLY "MONITOR", "IGNORE", "FAIL",
"AUTOFAIL", OR "EXIT".
Figure 15-12 Offload delayed message - IXG312E
There are several actions you can take to determine what could be inhibiting an offload:
Verify that there are no outstanding WTORs.
Determine whether there are inhibitors to offload processing by issuing commands such
as:
D
D
D
D
LOGGER,C,LSN=logstreamname
LOGGER,L,LSN=logstreamname
XCF,STRUCTURE,STRNAME=stucturename
GRS,C
Attempt to remedy any problems noted.

If the problem persists, respond or react to any allocation, catalog, or recall messages
such as IEF861I, IEF863I, or IEF458D.
If the problem persists, respond to message IXG312E. This can be used to stop the
offload processing for the log stream named in the message and allow it to run on another
system, if possible. It might also allow other work to run on the system that was attempting
the original offload.
The responses to message IXG312E are as follows:
MONITOR
IGNORE
FAIL
AUTOFAIL
EXIT
Continue monitoring this offload.

Stop monitoring this offload.
Fail the offload on this system.
Fail the offload on this system and continue this action for this log stream
for the duration of this connection.
Terminate system logger offload event monitor.
If message IXG115A is displayed, as shown in Figure 15-13, reply only after you have
attempted to remedy any delayed offloads by responding to the related IXG312E
messages. As a last resort, if you reply TASK=END to an IXG115A message, then system
logger will terminate all the log stream connections in the structure named in the message
on this system.
Review the complete description of messages IXG311I, IXG312E, IXG114I, and IXG115I
in z/OS MVS System Messages Volume 10 (IXC - IZP), SA22-7640, before responding to
any of these messages.
IXG115A CORRECT THE OFFLOAD CONDITION ON sysname FOR strname OR REPLY
TASK=END TO END THE STRUCTURE TASK.
Figure 15-13 Offload error condition - IXG115A
316
15.6 System logger ENQ serialization

System logger uses enqueues to serialize access to its resources. In case of a problem, you
can verify if there are any deadlock situations by issuing the following commands:
D GRS,C to check if there are any deadlock situations
D GRS,LATCH,JOBNAME=IXGLOGR to check for outstanding log streams latches
D GRS,RES=(SYSZLOGR,*) to check for ENQ contention major name is SYSZLOG and
minor is log stream name (or * for all)
If the command returns you an outstanding enqueue, reissue the command a few times over
several minutes. If the owner of the resource does not change, there may be a serialization
problem.
15.7 Handling a shortage of system logger directory extents

If your installation receives message IXG261E or IXG262A, as seen in Figure 15-14, it means
that the system logger has detected a shortage of log data set directory extent (DSEXTENT)
records in the active LOGR couple data set.
There is one DSEXTENT per log stream. There is a pool of spare offload data set directory
extents, and one extent is allocated whenever a log stream requires additional directory
extents. Each DSEXTENT will allow the log stream to use 168 offload data sets.
In this situation, log stream offloads may eventually fail if it is unable to obtain a log data set
directory extent required to process the offload.
IXG261E SHORTAGE OF DIRECTORY EXTENT RECORDS TOTAL numTotal
numInuse AVAILABLE: numAvail
IN USE:
IXG262A CRITICAL SHORTAGE OF DIRECTORY EXTENT RECORDS TOTAL numTotal IN

USE: numInuse AVAILABLE: numAvail
Figure 15-14 Shortage of Logger DSEXTENTS
The LOGR Couple Data Set, using the log data set directory extent, keeps track of all the
data sets in the log stream. Deleting them (with IDCAMS, for example) will not update the
LOGR Couple Data Set, and system logger will still think that the data set exists. It will report
missing data if an attempt is then made to access the data mapped in those offload data sets.
To resolve this situation, you can either try to understand which log stream is generating the
high number of offload data sets, or you can enlarge the DSEXTENT portion of the LOGR
couple data set.
To determine which log stream is using all the directory entries, you can run the IXCMIAPU
utility, as seen in Figure 15-10 on page 314,. Then take the corrective action against the log
stream, as described by the IXG261E or IXG262A messages.
If this does not solve the situation, here is a list of actions that may help the system
programmer resolve it:
Run the IXCMIAPU utility with the LIST option against the log streams to verify which log
streams are generating the high amount of offload data sets that are using all the directory
entries. Check if there is any anomaly in the definition of these log streams. A wrong
317
parameter may be the cause of the elevated number of offload data sets being created.
For example, a small value for LS_SIZE may be found. This means very small offload data
sets and if the log stream is generating a large amount of data, this can cause many
offload data sets being created, using all the available directory entries.
Define a new LOGR CDS with a bigger DSEXTENT value to allow new offload data sets
to be allocated and make this new LOGR Couple Data Set the active data set in the
sysplex.
Before allocating the new data set, you can display the current allocation with the
command D XCF,COUPLE,TYPE=LOGR, or you can run the IXCMIAPU and look for the
DSEXTENT field in the output display 1. This tells you how many extents are allocated in
the current LOGR Couple Data Set.
D XCF,COUPLE,TYPE=LOGR
LOGR COUPLE DATA SETS
PRIMARY
DSN: SYS1.XCF.LOGR01
VOLSER: #@$#X1
DEVN: 1D06
FORMAT TOD
MAXSYSTEM
12/11/2002 22:43:54
4
LOGR COUPLE DATA SET FORMAT LEVEL: HBB7705
LSR(200) LSTRR(120) DSEXTENT(10) 1
SMDUPLEX(1)
ALTERNATE DSN: SYS1.XCF.LOGR02
VOLSER: #@$#X2
DEVN: 1D07
FORMAT TOD
MAXSYSTEM
12/11/2002 22:43:58
4
LOGR COUPLE DATA SET FORMAT LEVEL: HBB7705
LSR(200) LSTRR(120) DSEXTENT(10) 1
SMDUPLEX(1)
LOGR IN USE BY ALL SYSTEMS
Figure 15-15 Display system logger Couple Data Set
The systems programmer can then use the IXCL1DSU utility to format a new LOGR Couple
Data Set, making sure the new Couple Data Set format has the appropriate number of LSRs,
LSTRRs, and a larger DSEXTENT.
After the new LOGR Couple Data Set is allocated, you can make it the alternate LOGR
Couple Data Set in your installation by issuing the command:
SETXCF COUPLE,ACOUPLE=(new_dsname),TYPE=LOGR
If the addition of this Couple Data Set is successful, then you can proceed and issue the
following command to switch control from the current primary to the new alternate Couple
Data Set:
SETXCF COUPLE,TYPE=LOGR,PSWITCH
A new alternate LOGR Couple Data Set will also need to be defined with the larger
DSEXTENT and allocated as the alternate Couple Data Set by issuing the command:
SETXCF COUPLE,ACOUPLE=(new_alternate_dsname),TYPE=LOGR
318
15.8 System logger structure rebuilds

This section applies to Coupling Facility log streams only. Here are some possible reasons to
rebuild a structure that contains log stream data:
Operator request because of the need to move allocated storage from one Coupling
Facility to another one.
Reaction to a failure.
15.8.1 Operator request

An operator can initiate the rebuild of the structure because of the need to: change the
configuration of the Coupling Facility; put the Coupling Facility offline due to a maintenance
request; or alter the size of the Coupling Facility structure. The rebuild operation can happen
dynamically while applications are connected to the log stream.
While the rebuild is in progress, system logger rejects any system logger service requests
against the log stream. Because this is only a temporary condition, most exploiters simply
report the failed attempt and redrive it.
To move the structure from one Coupling Facility to another, the structure needs an alternate
Coupling Facility in the preference list in the CFRM policy 1. The preference list of a structure
can be displayed using the D XCF,STR,STRNM=structure_name command, as seen in
Figure 15-16. The command also shows in which CF the structure currently resides 2.
D XCF,STR,STRNM=CIC_DFHLOG_001
STRNAME: CIC_DFHLOG_001
STATUS: ALLOCATED
. . .
DUPLEX
: DISABLED
ENFORCEORDER : NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 07/03/2007 07:09:23
CFNAME
: FACIL01
2
PARTITION: 00
CPCID: 00
ACTUAL SIZE
: 19200 K
. . .
# CONNECTIONS : 3
CONNECTION NAME ID VERSION SYSNAME JOBNAME
---------------- -- -------- -------- -------IXGLOGR_#@$1
03 00030039 #@$1
IXGLOGR
IXGLOGR_#@$2
01 00010055 #@$2
IXGLOGR
IXGLOGR_#@$3
02 0002003B #@$3
IXGLOGR
. . .
ASID
---0016
0016
0016
STATE
---------------ACTIVE
ACTIVE
ACTIVE
Figure 15-16 Display a CF preference list
319
To initiate a structure rebuild to an alternate Coupling Facility in the preference list, use this
command:
SETXCF START,REBUILD,STRNAME=structure_name
SETXCF START,REBUILD,STRNAME=CIC_DFHLOG_001
IXC521I REBUILD FOR STRUCTURE CIC_DFHLOG_001
HAS BEEN STARTED
CIC_DFHLOG_001 WAS ACCEPTED.
IXC526I STRUCTURE CIC_DFHLOG_001 IS REBUILDING FROM
INFO108: 00000013 00000013.
HAS BEEN COMPLETED
Figure 15-17 System Logger structure rebuild
15.8.2 Reaction to failure

The system logger structure is rebuilt if the following CF problems occur:
Damage to or failure of the Coupling Facility structure
Loss of connectivity to a Coupling Facility
A Coupling Facility becomes volatile
In all cases, the system logger initiates a rebuild to move the structure to another Coupling
Facility to avoid loss of data.
15.9 LOGREC logstream management

This section provides an example of the operational aspects of an exploiter of system logger.
LOGREC can use either a data set or a log stream for recording error and environmental
data. This section explains how to switch the recording of logrec data between a data set and
a log stream.
15.9.1 Displaying LOGREC status

Use the DISPLAY LOGREC command to see LOGREC status, as shown in Figure 15-18.
D LOGREC
IFB090I
00.23.01 LOGREC DISPLAY 845
CURRENT MEDIUM = DATASET
MEDIUM NAME = SYS1.#@$3.LOGREC
Figure 15-18 Display Logrec
Current Medium can be:

IGNORE
LOGSTREAM
DATASET
320
Recording of LOGREC error and environmental records is disabled.

The current medium for recording logrec error and environmental
records is a log stream.
The current medium for recording logrec error and environmental
records is a data set.
Medium Name can be:

SYSPLEX.LOGREC.ALLRECS The current medium is LOGSTREAM.
data set name
The current medium is DATASET.
The STATUS line is only displayed if the current medium is log stream. Its values can be:
CONNECTED
NOT CONNECTED
LOGGER DISABLED
If the STATUS shows as CONNECTED, then the log stream is connected and active.
15.9.2 Changing the LOGREC recording medium

Logrec recording can be changed dynamically via the following command:
SETLOGRC {LOGSTREAM | DATASET | IGNORE}
The operands indicate:
LOGSTREAM
The desired medium for recording logrec error and environment

records is a log stream. To use a log stream in your installation, the
logrec log stream must be defined.
DATASET
The desired medium for recording logrec error and environment

records is a data set. Setting the medium data set works only if the
system had originally been IPLed with a data set as the logrec
recording medium.
If the system was not IPLed with a data set logrec recording medium
and the attempt is made to change to DATASET, the system rejects
the attempt and maintains the current logrec recording medium.
IGNORE
This indicates that recording of logrec error and environmental records

is to be disabled. We recommend that you only use the IGNORE
option in a test environment.
D LOGREC
IFB090I
SETLOGRC LOGSTREAM
IFB097I LOGREC RECORDING MEDIUM CHANGED FROM DATASET TO LOGSTREAM
D LOGREC
IFB090I
CURRENT MEDIUM = LOGSTREAM
MEDIUM NAME = SYSPLEX.LOGREC.ALLRECS
STATUS = CONNECTED
SETLOGRC DATASET
IFB097I LOGREC RECORDING MEDIUM CHANGED FROM LOGSTREAM TO DATASET
D LOGREC
IFB090I
Figure 15-19 SETLOGRC command output
321
322
16
Chapter 16.
Network considerations in a
Parallel Sysplex
This chapter provides details of operational considerations to keep in mind for the network
environment about a Parallel Sysplex environment. It includes:
Virtual Telecommunications Access Method (VTAM) and its use of Generic Resources
(GR)
TCP/IP
Sysplex Distributor
Load Balancing Advisor (LBA)
IMS Connect
323
16.1 Introduction to network considerations in Parallel Sysplex

Mainframe architecture includes a variety of network capabilities. Some of these capabilities
include:
IP communications in a Parallel Sysplex
Communications using the TCP/IP suite of protocols, applications, and equipment
System Network Architecture (SNA) suite of protocols
The mainframe is usually connected to the outside world using an integrated LAN adapter
called the Open Systems Adapter OSA). The OSA is the equivalent of the network interface
card used in Windows and UNIX systems. It supports various operational modes and
protocols.
16.2 Overview of VTAM and Generic Resources

Virtual Telecommunications Access Method (VTAM) is a component of the z/OS
Communications Server (which is a part of z/OS) that is used to allow users to logon to SNA
applications.
Traditionally, when someone logged on to a SNA application, they would specify the APPLID
(VTAM name) of that application to VTAM. The APPLID would be unique within the network,
ensuring that everyone that logged on using a particular APPLID would end up in the same
instance of that application. Thus, even if you have 10 CICS regions, everyone who logged
on using an APPLID of CICPRD01 would end up in the only CICS region that had that
APPLID. Figure 16-1 illustrates a traditional SNA network.
Figure 16-1 SNA network connectivity
324
However, with the introduction of data sharing, it became necessary to have multiple SNA
application instances that could all access the same data. Taking CICS as an example, you
may have four CICS regions that all run the same applications and can access the same
data. To provide improved workload balancing and better availability, VTAM introduced a
function known as Generic Resources.
Figure 16-2 VTAM Generic Resource environment
Generic Resources allows an SNA application to effectively have two APPLIDs. One ID s
unique to that application instance. The other ID is shared with other SNA application
instances that share the same data or support the same business applications. The one that
is shared is called the generic resource name. Now, when an application connects to VTAM, it
can specify its APPLID and also request to join a particular generic resource group with the
appropriate generic resource name.
Note: VTAM Generic Resource can only be used by SNA applications, not by TCP/IP
applications.
There can be a number of generic resource groups. For example, there might be one for the
TSO IDs on every system, another for all the banking CICS regions, another for all the test
CICS regions, and so forth. When someone wants to logon to one of the banking CICS
regions, they can now specify the generic resource name, rather than the name of one
specific CICS region. As a result, if one of the CICS regions is down, the user will still get
logged on, and is not even aware of the fact that one of the regions is unavailable. This also
provides workload balancing advantages because VTAM, together with WLM, will now
ensure that the user sessions are spread across all the regions in the group.
Chapter 16. Network considerations in a Parallel Sysplex
325
VTAM uses a list structure in the Coupling Facility (CF) to hold the information about all the
generic resources in the Parallel Sysplex. In the structure, it keeps a list of all the active
generic resource groups, the APPLIDs of all the SNA applications that are connected to each
of those groups, a list of LUs that are in session with each APPLID, and counts of how many
sessions there are with each instance within each group. This information is updated
automatically each time a session is established or terminated. The default name for this
structure is ISTGENERIC. but you can override the default name by specifying a different
structure name on the VTAM STRGR start option (however, it must still begin with IST*).
For VTAM to use the CF, there must be an active CFRM policy defined for the Parallel
Sysplex, and the structure must be defined in that policy. All the VTAMs in the Parallel
Sysplex that are part of the same generic resource configuration must be connected to the
CF containing the structure, as well as all the other CFs indicated by the preference list for the
structure. When VTAM in a Parallel Sysplex is started, it automatically attempts to connect to
the CF structure, after first checking that the CFRM policy is active. When the first VTAM
become active in the Parallel Sysplex, XES will allocate the storage for the CF structure.
The structure disposition is specified as DELETE, which means when the last connector
disconnects from the structure, it is deallocated from CF storage.
The connection disposition is specified as KEEP, which means the connection is placed in a
failed-persistent state if it terminates. If the connection is failed-persistent, that usually means
that the VTAM that disconnected still has data out in the CF.
When one of the VTAMs in the sysplex disconnects from the structure, the remaining VTAMs
will normally clean up after that VTAM and remove the connection. If they detect any data that
was not deleted, they will leave the connection in a failed-persistent state. In that case, when
you issue the VARY NET,CFS command to get VTAM to disconnect from the structure, the other
VTAMs detect that the VTAM that disconnected is still active, and therefore do not actually
clean up any information relating to that VTAM, so the connection stays in failed-persistent
state.
On the other hand, when you actually stop VTAM, the other VTAMs know that it is not active
and clean up the entries related to that VTAM. As long as there are no persistent affinities,
they will delete the failing VTAM's connection.
The local copy of the generic resource information contained in the VTAM nodes is needed to
rebuild the VTAM structure.
When you stop VTAM normally, the connection to the structure will be deleted, unless it is the
last connector in the sysplex. If it is the last connector in the sysplex, it will go into a
failed-persistent state. This is because there might be persistent information in the structure
about affinities between certain applications and generic resources, so VTAM protects that
data by keeping the structure.
Because it impacts the availability of applications that use generic resources, you should be
aware that VTAM records affinities between application instances and sessions with those
instances for any application that has been using a generic resource name. The reason for
this is that, if the user initially logs on using a generic resource name, and is routed to CICSA,
any subsequent logon attempts should be routed to the same application instance (CICSA).
To be able to do this, VTAM sets a flag for any LU that is using an application that has
registered a generic resource name: on any subsequent logon attempt, VTAM checks that
flag to see if the logon should be routed to a particular instance. If the VTAM GR function
should become unavailable for some reason, VTAM is no longer able to check this
information. As a result, it will refuse to set up new sessions with those applications. This is
why it is important that you understand how VTAM GR works and how to manage it.
326
16.2.1 VTAM start options

Start options provide information about the conditions under which VTAM runs. They also
enable you to tailor VTAM to meet your needs each time VTAM is started. Many options can
have defaults specified as start options, thus reducing the amount of coding required. Many
start options can be dynamically modified and also displayed, as seen in Figure 16-3.
D NET,VTAMOPTS
IST097I DISPLAY ACCEPTED
IST1188I VTAM CSV1R8 STARTED AT 03:48:18
IST1349I COMPONENT ID IS 5695-11701-180
IST1348I VTAM STARTED AS END NODE
...
IST1189I CACHETI = 8
IST1189I CDRSCTI = 480S
IST1189I CDSREFER = ***NA***
IST1189I CMPMIPS = 100
IST1189I CNMTAB
= *BLANKS*
IST1189I COLD
= YES
IST1189I CONNTYPE = APPN
...
IST1189I NCPBUFSZ = 512
IST1189I NMVTLOG = NPDA
IST1189I NODELST = *BLANKS*
IST1189I NQNMODE = NAME
IST1189I NUMTREES = ***NA***
...
IST1189I SSEARCH = ***NA***
IST1189I STRMNPS = ISTMNPS
IST1189I SWNORDER = (CPNAME,FIRST)
...
ON 07/03/07
CDRDYN
CDSERVR
CINDXSIZ
CMPVTAM
CNNRTMSG
CONFIG
CPCDRSC
=
=
=
=
=
=
YES
***NA***
8176
0
***NA***
$3 1
= NO
NETID
NNSPREF
NODETYPE
NSRTSIZE
OSIEVENT
=
=
=
=
=
USIBMSC
NONE
EN
*BLANKS*
PATTERNS
STRGR
= ISTGENERIC 2
SUPP
= NOSUP
TCPNAME = *BLANKS*
Figure 16-3 Displaying VTAM options
1 ATCSTRxx VTAM start option member used

2 Default VTAM Generic Resource structure name being used
Be aware that some start options cannot be dynamically modified; they require that VTAM be
recycled. A complete list of start options is listed in z/OS V1R8.0 Communications Server:
SNA Resource Definition Reference, SC31-8778.
To use a start option list, create a member named ATCSTRxx and put it in the VTAMLST DD
partitioned data set that is referenced in the VTAM started procedure in your PROCLIB
concatenation. The xx value can be any two characters or numbers. This value allows you to
create different versions of the option list (ATCSTR00, ATCSTR01, ATCSTR02, and so forth)
and therefore different versions of VTAM start options.
VTAM is started from the z/OS console or during z/OS system startup with the command
START VTAM,LIST=xx. When VTAM initializes, LIST=xx determines which option list to use. For
example, if you specify LIST=01, VTAM uses ATCSTR01. VTAM always first attempts to
locate ATCSTR00, regardless of the option list chosen.
Tip: If your installation chooses to start VTAM with the SUB=MSTR option instead of under
JES2, you must ensure all data sets referenced in your VTAM started task (usually called
NET) are cataloged in the Master catalog or have a UNIT= and a VOL=SER= reference in
the started task.
327
When VTAM is restarted on your system, you see the message shown in Figure 16-4.
IXL014I IXLCONN REQUEST FOR STRUCTURE ISTGENERIC 461
WAS SUCCESSFUL. JOBNAME: NET ASID: 001B
CONNECTOR NAME: USIBMSC_#@$1M CFNAME: FACIL01
IST1370I USIBMSC.#@$1M IS CONNECTED TO STRUCTURE ISTGENERIC
Figure 16-4 VTAM connecting to the ISTGENERIC structure
16.2.2 Commands to display information about VTAM GR

You can use the commands shown here to display information relating to the VTAM GR:
Display resource statistics (Figure 16-5).
D NET,STATS,TYPE=CFS
IST350I DISPLAY TYPE = STATS,TYPE=CFS
IST1370I USIBMSC.#@$3M IS CONNECTED TO STRUCTURE ISTGENERIC 1
IST1797I STRUCTURE TYPE = LIST 2
IST1517I LIST HEADERS = 4 - LOCK HEADERS = 4
IST1373I STORAGE ELEMENT SIZE = 1024
IST924I ------------------------------------------------------------IST1374I
CURRENT
MAXIMUM PERCENT
IST1375I STRUCTURE SIZE
2560K
4096K
*NA*
IST1376I STORAGE ELEMENTS
4
77
5
IST1377I LIST ENTRIES
17
4265
0
IST314I END
Figure 16-5 Display resource statistics
1 VTAM GR default structure name

2 Structure type is LIST
Display resource statistics when the VTAM GR structure name (in our example,
ISTGENERIC) is known (Figure 16-6).
D NET,STATS,TYPE=CFS,STRNAME=ISTGENERIC
IST1797I STRUCTURE TYPE = LIST
IST924I ------------------------------------------------------------IST1374I
CURRENT
MAXIMUM PERCENT
2560K
4096K
*NA*
4
77
5
17
4265
0
IST314I END
Figure 16-6 Display ISTGENERIC
Display information about VTAM generic resource groups (Figure 16-7 on page 329).
328
D NET,RSCLIST,IDTYPE=GENERIC,ID=*
IST350I DISPLAY TYPE = RSCLIST
IST1417I NETID
NAME
STATUS
TYPE
IST1418I USIBMSC TSO$$
ACT/S
GENERIC
IST1418I USIBMSC SCSMCS$$ ACTIV
GENERIC
IST1418I USIBMSC #@$C1TOR ACTIV
GENERIC
IST1418I USIBMSC ITSOI#$# INACT
GENERIC
IST1418I USIBMSC #@$C1TOR *NA*
GENERIC
IST1418I USIBMSC ITSOI#$# *NA*
GENERIC
*NA*
GENERIC
IST1418I USIBMSC SCSMCS$$ *NA*
GENERIC
IST1454I 8 RESOURCE(S) DISPLAYED FOR ID=*
IST314I END
MAJNODE
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
RESOURCE
RESOURCE
RESOURCE
RESOURCE
USERVAR
USERVAR
USERVAR
USERVAR
Figure 16-7 Display generic resource groups
Display who is using a specific VTAM generic resource group (Figure 16-8).
D NET,ID=TSO$$,E
IST075I NAME = TSO$$, TYPE = GENERIC RESOURCE
IST1359I MEMBER NAME
OWNING CP
SELECTABLE
IST1360I USIBMSC.SC$2TS
#@$2M
YES
#@$3M
YES
#@$1M
YES
IST314I END
APPC
NO
NO
NO
Figure 16-8 Display users of a specific generic resource
Display affinity information for generic resources (Figure 16-9).

D NET,GRAFFIN,LU=*
IST350I DISPLAY TYPE = GENERIC AFFINITY
IST1358I NO QUALIFYING MATCHES
IST1454I 0 AFFINITIES DISPLAYED
IST314I END
Figure 16-9 Display affinity to generic resources
Disconnect from a VTAM Coupling Facility structure (Figure 16-10).

V NET,CFS,STRNM=strname,ACTION=DISCONNECT
IST097I VARY ACCEPTED
IST1380I DISCONNECTING FROM STRUCTURE ISTGENERIC 571
IST2167I DISCONNECT REASON - OPERATOR COMMAND
IST314I END
Figure 16-10 Disconnect from a structure
Note: When you have issued the command to disconnect from the structure and then
display the status of the structure, the connection will be in failed-persistent state, as
displayed in Figure 16-11.
329
D XCF,STR,STRNM=ISTGENERIC
STRNAME: ISTGENERIC
STATUS: ALLOCATED
...
---------------- -- -------- -------- -------- ---- ---------------USIBMSC_#@$1M
03 00030095 #@$1
NET
001B ACTIVE
USIBMSC_#@$2M
01 000100C3 #@$2
NET
001B ACTIVE
USIBMSC_#@$3M
02 0002008F #@$3
NET
001B FAILED-PERSISTENT 1
...
Figure 16-11 Failed-persistent state for structure
1 ISTGENERIC structure in a failed-persistent state,

Connect to a VTAM Coupling Facility structure.
V NET,CFS,STRNM=strname,ACTION=CONNECT
IST097I VARY ACCEPTED
IXL014I IXLCONN REQUEST FOR STRUCTURE ISTGENERIC 585
WAS SUCCESSFUL. JOBNAME: NET ASID: 001B
CONNECTOR NAME: USIBMSC_#@$3M CFNAME: FACIL01
Figure 16-12 Connect to a structure
16.3 Managing Generic Resources

With the Generic Resource function enabled in a sysplex you can achieve increased
availability, as well as the ability to balance the session workload. As mentioned, the GR
function allows the assignment of a unique name to a group of active application programs,
where they all provide the same function, for example CICS.
For the Generic Resource function to operate, each VTAM application must register itself to
VTAM under its generic resource name. This section examines CICS and TSO generic
resources.
Registration is performed automatically by CICS and TSO when they are ready to receive
logon requests. LUs initiate a logon request to the generic resource name and need not be
aware of which particular application is providing the function. Therefore, session workloads
are balanced and the session distribution is transparent to end users.
16.3.1 Determine the status of Generic Resources

You can issue the following VTAM and CICS commands to determine the status of Generic
Resources.
VTAM commands for Generic Resource management

You can issue the following command to determine if the Generic Resource function is
enabled for your system and the utilization of the structure. The strname is the name of the
GR structure as specified in your VTAM startup options. The default name is ISTGENERIC.
330
Figure 16-13 Display VTAM GR is active
Figure 16-14 illustrates sample output of the D NET,STATS command, displaying the status of
our GR structure.
IST1797I STRUCTURE TYPE = LIST
IST924I ------------------------------------------------------------IST1374I
CURRENT
MAXIMUM PERCENT 1
2560K
4096K
*NA*
4
77
5
17
4265
0
IST314I END
Figure 16-14 Output from D NET,STATS command
1 If any one of the entries is utilized more than 80%, contact your systems programmer to
review and possibly alter the size of the structure. This type of monitoring used by the system
in a Parallel Sysplex environment is called structure full monitoring.
Structure full monitoring adds support for the monitoring of objects within a Coupling Facility
structure. Its objective is to determine the level of usage for objects that are monitored within
a CF and to issue a warning message to the console if a structure full condition is imminent.
The default value for the monitoring threshold is 80%.
You can also issue the command in Figure 16-15 to display the structure full monitoring
threshold for the particular structure.
STRNAME: ISTGENERIC
STATUS: ALLOCATED
TYPE: SERIALIZED LIST
POLICY INFORMATION:
POLICY SIZE
: 4096 K
FULLTHRESHOLD : 80 1
. . .
Figure 16-15 Display structure full threshold value
1 Structure full threshold value for ISTGENERIC

To determine the generic resource group of your application systems, issue the command
shown in Figure 16-16 on page 332.
331
D NET,RSCLIST,IDTYPE=GENERIC,ID=*
IST350I DISPLAY TYPE = RSCLIST
IST1417I NETID
NAME
STATUS
TYPE 1
IST1418I USIBMSC ITSOI#$# INACT
GENERIC
ACT/S
GENERIC
IST1418I USIBMSC SCSMCS$$ ACTIV
GENERIC
IST1418I USIBMSC #@$C1TOR ACTIV
GENERIC
IST1418I USIBMSC #@$C1TOR *NA*
GENERIC
IST1418I USIBMSC ITSOI#$# *NA*
GENERIC
*NA*
GENERIC
IST1418I USIBMSC SCSMCS$$ *NA*
GENERIC
IST1454I 8 RESOURCE(S) DISPLAYED FOR ID=*
IST314I END
RESOURCE
RESOURCE
RESOURCE
RESOURCE
USERVAR
USERVAR
USERVAR
USERVAR
MAJNODE
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
**NA**
Figure 16-16 Display all generic resource names
1 List of all generic resource names

Issue the following commands, shown in Figure 16-17 and Figure 16-18, to determine which
application systems are connected to a specific generic name and the status of the sessions.
D NET,SESSIONS,LU1=TSO$$
IST350I DISPLAY TYPE = SESSIONS
IST1364I TSO$$ IS A GENERIC RESOURCE NAME FOR:
IST988I SC$2TS
SC$3TS
SC$1TS
IST924I ------------------------------------------IST172I NO SESSIONS EXIST
IST314I END
Figure 16-17 Displaying a specific generic resource name for TSO
To display a specific generic resource name for CICS, issue the command as displayed in
Figure 16-18.
D NET,SESSIONS,LU1=#@$C1TOR
IST1364I #@$C1TOR IS A GENERIC RESOURCE NAME FOR:
IST988I #@$C1T3A #@$C1T2A
IST924I ---------------------------------------------------IST878I NUMBER OF PENDING SESSIONS =
0
IST878I NUMBER OF ACTIVE SESSIONS =
1 1
IST878I NUMBER OF QUEUED SESSIONS =
0
IST878I NUMBER OF TOTAL
SESSIONS =
1
IST314I END
Figure 16-18 Displaying a specific generic resource name for CICS
1 There is 1 active session to the CICS Terminal Owning Region (TOR) named #@$C1TOR.
332
CICS commands for Generic Resource management

The INQUIRE VTAM command returns information about type and state of the VTAM
connection for the CICS system. stcname is the name of your CICS Terminal Owning Region
(TOR).
F stcname,CEMT I VTAM
+
Vtam
Openstatus( Open ) 1
Psdinterval( 000000 )
Grstatus( Registered ) 2
Grname(#@$C1TOR)3
RESPONSE: NORMAL TIME: 03.15.30
SYSID=1T3A APPLID=#@$C1T3A
DATE: 07.04.07
Figure 16-19 Sample output of the CEMT INQUIRE VTAM command
1 Openstatus returns the value indicating the communication status between CICS and
VTAM.
2 Grstatus returns one of the following, indicating the status of Generic Resource registration.
Blanks are returned if the Generic Resource function is disabled for the CICS system.
Deregerror
Deregistration was attempted but was unsuccessful and there has

been no attempt to register.
Deregistered
Deregistration was successfully accomplished.
Notapplic
CICS is not using the Generic Resource function.
Regerror
Registration was attempted but was unsuccessful and there has been
no attempt to deregister.
Registered
Registration was successful and there has been no attempt to

deregister.
Unavailable
VTAM does not support the generic resource function.
Unregistered
CICS is using the Generic Resource function but no attempt, as yet,

has been made to register.
3 Grname returns the Generic Resource under which this CICS system requests registration to
VTAM. Blanks are returned if the Generic Resource function is not enabled for the CICS
system.
16.3.2 Managing CICS Generic Resources

With the Generic Resource function enabled in a CICSPlex, you can achieve increased
availability, as well as the ability to balance the session workload. For example, if you have
implemented three CICS TORs and let them register to the same Generic Resource, VTAM
will distribute the incoming session requests among all three TORs based on
installation-defined criteria. If one of the TORs should become unavailable, users can still log
on to CICS, where VTAM now chooses between the two remaining TORs.
Removing a CICS region from a Generic Resource group

There might be times when you want to remove a particular CICS TOR from its generic
resource group. For example, you might want to take an z/OS image down for scheduled
service, so you would like to fence the CICS/VTAM on that system from accepting new
logons, and allow existing CICS/VTAM users on that system to continue working.
333
Issue the following command to deregister the CICS TOR, where stcname is the name of the
CICS TOR.
F stcname,CEMT SET VTAM DEREGISTER
Figure 16-20 Remove CICS from using a generic resource group
Figure 16-21 illustrates sample output of the deregister command CEMT SET VTAM
DEREGISTER.
F #@$C1T3A,CEMT SET VTAM DEREGISTER
+
Vtam
Openstatus( Open )
Psdinterval( 000000 )
Grstatus( Deregistered ) 1
Grname(#@$C1TOR)
NORMAL
RESPONSE: NORMAL TIME: 20.06.44 DATE: 07.04.07
SYSID=1T3A APPLID=#@$C1T3A
Figure 16-21 Sample output from the deregister command
1 The CICS TOR has been successfully deregistered from the generic resource group.
Refer to Figure 16-22 for sample output of the D NET,SESSIONS command. Notice that the
CICS TOR has been 1 removed from the generic resource group.
D NET,SESSIONS,LU1=#@$C1TOR
IST1364I #@$C1TOR IS A GENERIC RESOURCE NAME FOR:
IST988I #@$C1T2A 1
IST924I -------------------------------------------IST172I NO SESSIONS EXIST
IST314I END
Figure 16-22 Sample output from the D NET,SESSIONS command
Note: After you deregister the CICS TOR from the generic resource group, you must
restart the CICS TOR to register it to the generic resource group again.
16.3.3 Managing TSO Generic Resources

With the Generic Resource function enabled for TSO, you can obtain increased availability as
well as the ability to balance the TSO workload across the Parallel Sysplex.
Removing a TSO/VTAM from a generic resource group

There might be times when you want to remove a particular TSO/VTAM from its generic
resource group. For example, you might want to take an z/OS image down for scheduled
service, so you would like to fence the TSO/VTAM on that system from accepting new logons,
334
and allow existing TSO/VTAM users on that system to continue working. To accomplish this,
issue the following command, where stcname is the TSO name.
F stcname,USERMAX=0
Figure 16-23 Command to set TSO to user max of zero (0)
Figure 16-24 shows the output after the command was issued to system #@$2.
RO #@$2,F TSO,USERMAX=0
IKT033I TCAS USERMAX VALUE SET TO 0
IKT008I TCAS NOT ACCEPTING LOGONS
Figure 16-24 TSO usermax set to zero (0) on system #@$2
This causes the TSO/VTAM on the image (#@$2) to deregister and to stop accepting new
TSO logons.
Figure 16-25 shows the output of the D NET,SESSIONS command prior to the TSO/VTAM
generic resource group from system #@$2 being removed.
IST988I SC$2TS
SC$3TS
SC$1TS 1
IST924I -----------------------------------------IST172I NO SESSIONS EXIST
IST314I END
Figure 16-25 D NET,SESSIONS prior to removing TSO on system #@$2
1 TSO on z/OS systems #@$1, #@$2, and #@$3 are all connected to the TSO generic
resource group.
IST988I SC$3TS
SC$1TS 1
IST924I -------------------------------------------IST172I NO SESSIONS EXIST
IST314I END
Figure 16-26 D NET,SESSIONS after removing TSO on system #@$2
1 TSO on z/OS systems #@$1 and #@$3 are the only ones connected to the TSO generic
resource group.
Figure 16-26 shows the output of the D NET,SESSIONS command after we removed TSO on
system #@$2 from using the TSO generic resource group.
335
After you deregister TSO from the generic resource group, you can issue the F
stcname,USERMAX=nn command where nn is greater than zero (0), or you can restart the TSO
started procedure, to register it to the generic resource group again.
In Figure 16-27, we issue MVS commands to reset the 1 TSO usermax value to 30 on z/OS
system #@$2, and then 2 display the usermax setting to verify that the 3 value has been set
correctly.
RO #@$2,F TSO,USERMAX=30 1
RO #@$2,D TS,L 2
IEE114I 20.55.29 2007.185 ACTIVITY 512
JOBS
M/S
TS USERS
SYSAS
INITS
00001
00040
00001
00033
00016
HAIN
OWT
ACTIVE/MAX VTAM
00001/00030 3
OAS
00011
Figure 16-27 Reset TSO usermax value to 30 on system #@$2
Figure 16-28 shows the output of the D NET,SESSIONS command. Notice that 1 TSO (SC$2TS)
has been added to the generic resource group.
We issued the D NET,SESSIONS command shown in Figure 16-28 to verify that TSO on
system #@$2 has successfully been added to the generic resource group.
IST988I SC$2TS 1 SC$3TS
SC$1TS
IST924I ----------------------------------------------IST172I NO SESSIONS EXIST
IST314I END
Figure 16-28 System #@$2 added to generic resource group
1 TSO on system #@$2 is added to generic resource group.
16.4 Introduction to TCP/IP

The TCP/IP protocol suite is named for two of its most important protocols: Transmission
Control Protocol (TCP) and Internet Protocol (IP).
The main design goal of TCP/IP was to build an interconnection of networks, referred to as an
internetwork (or Internet), that provided universal communication services over
heterogeneous physical networks. The clear benefit of such an internetwork is the enabling of
communication between hosts on different networks, perhaps separated by a wide
geographical area.
The Internet consists of the following groups of networks:
Backbones, which are large networks that exist primarily to interconnect other networks.
Regional networks that connect, for example, universities and colleges.
Commercial networks that provide access to the backbones to subscribers, and networks
owned by commercial organizations for internal use that also have connections to the
Internet.
336
Local networks, such as campus-wide university networks.

Figure 16-29 provides a high level view of a TCP/IP network.
Figure 16-29 TCP/IP network
The TCP/IP started task is the engine that drives all IP-based activity on z/OS. The TCP/IP
profile data set controls the configuration of the TCP/IP environment.
Figure 16-30 is a sample of the TCP/IP started task. The DD statements PROFILE and SYSTCPD
refer to data sets that contain various configuration information that is used by TCP/IP.
//TCPIP
//*
//TCPIP
//
//
//SYSPRINT
//ALGPRINT
//SYSOUT
//CEEDUMP
//SYSERROR
//PROFILE
//SYSTCPD
PROC PARMS='CTRACE(CTIEZB00)'
EXEC PGM=EZBTCPIP,
PARM='&PARMS',
REGION=0M,TIME=1440
DD SYSOUT=*,DCB=(RECFM=FB,LRECL=137,BLKSIZE=137)
DD SYSOUT=*
DD DISP=SHR,DSN=SYS1.TCPPARMS(TCPPRF&SYSCLONE.)
DD DSN=SYS1.TCPPARMS(TCPDATA),DISP=SHR
Figure 16-30 TCP/IP started task
The TCP/IP profile member referred to by the PROFILE DD statement is read by TCP/IP
when it is started. If a change needs to be made to the TCP/IP configuration after it has been
started, TCP/IP can be made to reread the profile dynamically (or read a new profile
altogether) using the V TCPIP command. Additional information about the V TCPIP command
can be found in z/OS Communications Server: IP System Administration Commands,
SC31-8781.
337
16.4.1 Useful TCP/IP commands

Following are useful commands that can be issued from the z/OS console, from a z/OS UNIX
OMVS shell, or from ISHELL within ISPF. Refer to z/OS Communications Server: IP System
Administration Commands, SC31-8781 for detailed explanations about the commands and
associated parameters.
Note: Some of the commands have different names depending on the environment that
they are issued in. For example, the TSO TRACERTE command has a different name from
the UNIX TRACEROUTE command. Also, the UNIX version of the commands may be
prefixed with the letter O (for example, OTRACERT) which is a command synonym.
PING
NSLOOKUP
TRACERT
NETSTAT
HOMETEST
D TCPIP,,HELP
Find out if the specified node is active.

Query a NameServer.
Debug network problems.
Display network status of a local host.
Verify your host name and address configuration.
Obtain additional help for NETSTAT, TELNET, HELP, DISPLAY, VARY,
OMPROUTE, SYSPLEX, and STOR commands.
The following figures illustrate examples of the D TCPIP command.

Note: All of the following commands were issued from a z/OS console. The HELP
responses returned on certain commands may not be explained in detail, so you may need
to refer to z/OS Communications Server: IP System Administration Commands,
SC31-8781, for additional information and clarification.
D TCPIP,,HELP
EZZ0371I D...(NETSTAT|TELNET|HELP|DISPLAY|VARY|OMPROUTE|
EZZ0371I SYSPLEX|STOR)
Figure 16-31 D TCPIP command
Figure 16-32 shows the command for Help on the NETSTAT command.
D TCPIP,,HELP,NETSTAT
EZZ0372I D...NETSTAT(,ACCESS|ALLCONN|ARP|BYTEINFO|CACHINFO|
EZZ0372I CONFIG|CONN|DEVLINKS|HOME|IDS|ND|PORTLIST|ROUTE|
EZZ0372I SOCKETS|SRCIP|STATS|TTLS|VCRT|VDPT|VIPADCFG|VIPADYN)
Figure 16-32 Help on NETSTAT command
Figure 16-33 shows the command for Help on the TELNET command.
D TCPIP,,HELP,TELNET
EZZ0373I D...TELNET(,CLIENTID|CONNECTION|OBJECT|
EZZ0373I INACTLUS|PROFILE|WHEREUSED|WLM)
EZZ0373I V...TELNET(,ACT|INACT|QUIESCE|RESUME|STOP)
Figure 16-33 Help on TELNET command
Figure 16-34 on page 339 shows the command for Help on the VARY command.
338
D TCPIP,,HELP,VARY
EZZ0358I V...(,DATTRACE|DROP|OBEYFILE|OSAENTA|PKTTRACE|
EZZ0358I PURGECACHE|START|STOP|SYSPLEX|TELNET)
Figure 16-34 Help on VARY command
Figure 16-35 shows the command for Help on the OMPROUTE command.
D TCPIP,,HELP,OMPROUTE
EZZ0626I D...OMPROUTE(,GENERIC|GENERIC6|IPV6OSPF|IPV6RIP|
EZZ0626I OSPF|RIP|RTTABLE|RT6TABLE)
Figure 16-35 Help on OMPROUTE command
Figure 16-36 shows the command for Help on the SYXPLEX command.
D TCPIP,,HELP,SYSPLEX
EZZ0637I D...SYSPLEX,(GROUP|VIPADYN)
EZZ0637I V...SYSPLEX,(LEAVEGROUP|JOINGROUP|DEACTIVATE|REACTIVATE
|QUIESCE|RESUME)
Figure 16-36 Help on SYSPLEX command
Figure 16-37 shows the command for Help on the STOR command.
D TCPIP,,HELP,STOR
EZZ0654I D...STOR<,MODULE=XMODID>
Figure 16-37 Help on STOR command
Note: Although these commands are display only, some of the options returned have the
potential to impact your TCP/IP configuration. If you are unsure about the outcome, consult
your support staff.
16.5 Sysplex Distributor

Sysplex Distributor is a function of the z/OS IBM Communications Server. Using Sysplex
Distributor, workload can be distributed to multiple server instances within the sysplex without
requiring changes to clients or networking hardware and without delays in connection setup.
z/OS IBM Communications Server provides the way to implement a dynamic Virtual IP
Address (VIPA) as a single network-visible IP address for a set of hosts that belong to the
same sysplex cluster. Any client located anywhere in the IP network is able to see the sysplex
cluster as one IP address, regardless of the number of hosts that it includes.
With Sysplex Distributor, clients receive the benefits of workload distribution provided by
Workload Manager (WLM). In addition, Sysplex Distributor ensures high availability of the IP
applications running on the sysplex cluster, even if one physical network interface fails or an
entire IP stack or z/OS system is lost.
339
16.5.1 Static VIPA and dynamic VIPA overview

The concept of virtual IP address (VIPA) was introduced to remove the dependencies of other
hosts on particular network attachments to z/OS IBM Communications Server TCP/IP. Prior
to VIPA, other hosts were bound to one of the home IP addresses and, therefore, to a
particular network interface. If the physical network interface failed, the home IP address
became unreachable and all the connections already established with this IP address also
failed.
VIPA provides a virtual network interface with a virtual IP address that other TCP/IP hosts can
use to select an z/OS IP stack without choosing a specific network interface on that stack. If a
specific physical network interface fails, the VIPA address remains reachable by other
physical network interfaces. Hosts that connect to z/OS IP applications can send data to a
VIPA address via whatever path is selected by the dynamic routing protocol.
A VIPA is configured the same as a normal IP address for a physical adapter, except that it is
not associated with any particular interface. VIPA uses a virtual device and a virtual IP
address. The virtual IP address is added to the home address list. The virtual device defined
for the VIPA using DEVICE, LINK, and HOME statements is always active and never fails.
Moreover, the z/OS IP stack advertises routes to the VIPA address as though it were one hop
away and has reachability to it.
Dynamic VIPA was introduced to enable the dynamic activation of a VIPA as well as the
automatic movement of a VIPA to another surviving z/OS image after an z/OS TCP/IP stack
failure.
There are two forms of Dynamic VIPA, both of which can be used for takeover functionality:
Automatic VIPA takeover allows a VIPA address to move automatically to a stack (called a
backup stack) where an existing suitable application instance is already active. It also
allows the application to serve the client formerly going to the failed stack.
Dynamic VIPA activation for an application server allows an application to create and
activate VIPA so that the VIPA moves when the application moves.
Using D TCPIP commands, we can display the VIPA configuration on our test Parallel Sysplex
(#@$#PLEX) environment, which consists of three z/OS systems named #@$1, #@$2, and
#@$3; see Figure 16-38.
D TCPIP,,NETSTAT,VIPADYN
EZZ2500I NETSTAT CS V1R8 TCPIP 725
DYNAMIC VIPA:
IP ADDRESS
ADDRESSMASK
STATUS
201.2.10.11
255.255.255.192 ACTIVE
ACTTIME:
07/06/2007 00:17:23
201.2.10.12
255.255.255.192 ACTIVE
ACTTIME:
07/06/2007 00:17:20
201.2.10.13
255.255.255.192 ACTIVE
ACTTIME:
07/05/2007 22:12:40
3 OF 3 RECORDS DISPLAYED
ORIGINATION
VIPABACKUP 2
DISTSTAT
VIPABACKUP 3
VIPADEFINE 1
Figure 16-38 Dynamic VIPA configuration from z/OS system #@$3
1 Dynamic VIPA address for z/OS system #@$3.

2 and 3 backup Dynamic VIPA addresses for systems #@$1 and #@$2.
340
Figure 16-39 shows the Dynamic VIPA configuration for #@$#PLEX.

D TCPIP,,SYSPLEX,VIPADYN
EZZ8260I SYSPLEX CS V1R8 752
VIPA DYNAMIC DISPLAY FROM TCPIP
AT #@$3
IPADDR: 201.2.10.11 LINKNAME: VIPLC9020A0B
ORIGIN: VIPABACKUP
TCPNAME MVSNAME STATUS RANK ADDRESS MASK
-------- -------- ------ ---- --------------TCPIP
#@$3
ACTIVE
255.255.255.192
IPADDR: 201.2.10.12 LINKNAME: VIPLC9020A0C
ORIGIN: VIPABACKUP
-------- -------- ------ ---- --------------TCPIP
#@$3
ACTIVE
255.255.255.192
IPADDR: 201.2.10.13 LINKNAME: VIPLC9020A0D
ORIGIN: VIPADEFINE
-------- -------- ------ ---- --------------TCPIP
#@$3
ACTIVE
255.255.255.192
3 OF 3 RECORDS DISPLAYED
NETWORK PREFIX DIST

--------------- ---201.2.10.0
NETWORK PREFIX DIST

--------------- ---201.2.10.0
NETWORK PREFIX DIST

--------------- ---201.2.10.0
Figure 16-39 Dynamic VIPA configuration for sysplex #@$#PLEX
16.6 Load Balancing Advisor

The main function of the z/OS Load Balancing Advisor (LBA) is to provide external TCP/IP
load balancing solutions with recommendations about which TCP/IP applications and target
z/OS systems within a sysplex are best equipped to handle new TCP/IP workload requests.
These recommendations can then be used by the load balancer to determine how to route
the requests to the target applications and systems (that is, how many requests should be
routed to each target). The recommendations provided by the advisor are dynamic and can
change as the conditions of the target systems and applications change.
Note:
LBA is available in APAR PQ90032 for z/OS V1R4.
LBA is available in APAR PQ96293 for z/OS V1R5 and V1R6.
LBA is part of the base z/OS V1R7 Communications Server product and beyond.
Figure 16-40 on page 342 illustrates Load Balancing Advisor.
341
Figure 16-40 z/OS Load Balancing Advisor
In Figure 16-40, the load balancer is configured with a list of systems and applications that it
will balance. The load balancer tells the Load Balancing Advisor about the applications by
specifying an IP address, port, and protocol, or about the systems by specifying an IP
address. Note the following:
The advisor is configured with a list of authorized load balancers and a list of load
balancing agents with which it can gather data, and with a poll interval at which the agents
update the advisor's data.
Each agent gathers data on its own z/OS system about the TCP/IP stacks and
applications running on that system. The agent is configured with the information it needs
to contact the advisor.
The advisor consolidates the data from all its agents, and returns the data to the load
balancer to advise the load balancer about the status of the systems and applications.
16.7 IMS Connect

IMS Connect is an example of a TCP/IP-based application. In our Parallel Sysplex
environment, IMS Connect V9 was used.
We issued the NETSTAT command to display which ports are being used by IMS Connect, as
342
D TCPIP,,NETSTAT,ALLCONN
EZZ2500I NETSTAT CS V1R8 TCPIP 304
USER ID CONN
LOCAL SOCKET
#@$CWE3A 00000033 0.0.0.0..4445
BPXOINIT 00000012 0.0.0.0..10007
D#$3DIST 0000008B 0.0.0.0..33366
D#$3DIST 0000008C 0.0.0.0..33367
FTPD1
00000011 0.0.0.0..21
I#$3CON 000000AC 0.0.0.0..7302 1
I#$3CON 000000AB 0.0.0.0..7301 2
I#$3CON 000000AD 0.0.0.0..7303 3
FOREIGN SOCKET
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
0.0.0.0..0
STATE
LISTEN
LISTEN
LISTEN
LISTEN
LISTEN
LISTEN
LISTEN
LISTEN
Figure 16-41 NETSTAT command for IMS Connect
1, 2, and 3 display ports 7301, 7302, and 7303 as being used by IMS Connect.
Refer to Chapter 19, IMS operational considerations in a Parallel Sysplex on page 397, for
more information about IMS.
343
344
17
Chapter 17.
CICS operational considerations

This chapter provides an overview and background information about operational
considerations to keep in mind when CICS is used in a Parallel Sysplex. It covers the
following topics:
CICS and Parallel Sysplex
Multiregion operation (MRO)
CICS log and journal
CICS shared temporary storage
CICS Coupling Facility data table (CFDT)
CICS named counter server
CICS and ARM
CICSPlex System Manager (CPSM)
What is CICSPlex
345
17.1 Introduction to CICS

CICS, or Customer Information Control System, is a transaction processing (TP) monitor that
was developed to provide transaction processing for IBM mainframes. TP monitors perform
these functions:
System runtime functions
TP monitors provide an execution environment that ensures the integrity, availability, and
security of data, in addition to fast response time and high transaction throughput.
System administration functions
TP monitors provide administrative support that lets users configure, monitor, and manage
their transaction systems.
Application development functions
TP monitors provide functions for use in custom business applications, including functions
to access data, perform intercomputer communications, and design and manage the user
interface.
CICS controls the interaction between applications and users and allows programmers to
develop window displays without detailed knowledge of the terminals being used. It belongs
to the IBM online transaction processing (OLTP) family of products. It is sometimes referred
to as a DB/DC (database/data communications) system.
Typical CICS applications include bank ATM transaction processing, library applications,
student registration, airline reservations, and so on.
CICS has been called an operating system within an operating system, because it has a
dispatcher, storage control, task control, file control, and other features. It was designed to
allow application programmers to devote their time and effort to the application solution,
instead of dwelling on complex programming issues. CICS can be thought of as an interface
between its TP applications and the operating system.
17.2 CICS and Parallel Sysplex

A sysplex consists of multiple z/OS systems, coupled together by hardware elements and
software services. In a sysplex, z/OS provides a platform of basic multisystem services that
applications like CICS can exploit. As an installation's workload grows, additional MVS
systems can be added to the sysplex to enable the installation to meet the needs of the
greater workload.
To use XCF to communicate in a sysplex, each CICS region joins an XCF group called
DFHIR000 by invoking the MVS IXCJOIN macro using services that are provided by the
DFHIRP module. The member name for each CICS region is always the CICS APPLID
(NETNAME on the CONNECTION resource definition) used for MRO partners.
Each CICS APPLID must be unique within any sysplex, regardless of the MVS levels that are
involved. Within the sysplex, CICS regions can communicate only with members of the CICS
XCF group (DFHIR000). Figure 17-1 on page 347 illustrates CICS in a Parallel Sysplex
environment.
346
Figure 17-1 Overview of CICS in a Parallel Sysplex environment
17.3 Multiregion operation

CICS multiregion operation (MRO) enables CICS systems that are running in the same z/OS
image, or in the same z/OS sysplex, to communicate with each other.
The support within CICS that enables region-to-region communication is called interregion
communication (IRC). XCF is required for MRO links between CICS regions in different MVS
images of an MVS sysplex. It is selected dynamically by CICS for such links; SNA networking
facilities are not required for MRO.
Cross-system multiregion operation (XCF/MRO)

XCF is part of the z/OS base control program, providing high performance communication
links between z/OS images linked in a sysplex.
The IRC provides an XCF access method that makes it unnecessary to use VTAM to
communicate between z/OS images within the same z/OS sysplex. Using XCF services,
CICS regions join a single XCF group called DFHIR000. Members of the CICS XCF group
that are in different z/OS images select the XCF access method dynamically when they want
Chapter 17. CICS operational considerations in a Parallel Sysplex
347
to talk to each other, overriding the access method specified on the connection resource
definition.
Overview of transaction routing

CICS transaction routing allows terminals connected to one CICS system to run with
transactions in another connected CICS system. This means that you can distribute terminals
and transactions around your CICS systems and still have the ability to run any transaction
with any terminal.
Figure 17-2 Overview of transaction routing
Figure 17-2 shows a terminal connected to one CICS system running with a user transaction
in another CICS system. Communication between the terminal and the user transaction is
handled by a CICS-supplied transaction called the relay transaction.
The CICS system that owns the terminal is called the terminal-owning region (TOR). The
CICS system that owns the transaction is called the application-owning region (AOR). These
terms are not meant to imply that one system owns all the terminals and the other system all
the transactions, although this is a possible configuration.
The terminal-owning region and the application-owning region must be connected by MRO or
APPC links.
17.4 CICS log and journal

System logger
System logger is a set of services that allow an application to write, browse, and delete log
data. You can use system logger services to merge data from multiple instances of an
application, including merging data from different systems across a sysplex. It uses list
structures to hold the logstream data from exploiters of the system logger, such as CICS. For
more detail, refer to Chapter 15, z/OS system logger considerations on page 307.
CICS logs and journals

In CICS Transaction Server (TS), the CICS log manager uses the z/OS system logger for all
its logging and journaling requirements. Using services provided by the z/OS system logger,
the CICS log manager supports:
The CICS system log, used for transaction backout, emergency restart, and preserving
information for resynchronizing an in-doubt unit of work (UOW), even on a cold start. The
CICS system log is used for all transaction back-outs.
Forward recovery logs, auto-journals, user journals, and a log of logs. These are
collectively referred to as general logs, to distinguish them from system logs.
348
For CICS logging, you can use Coupling Facility-based logstreams, DASD only logstreams,
or a combination of both. Remember that all connections to DASD only logstreams must
come from the same z/OS image, which means you cannot use a DASD only logstream for a
user journal that is accessed by CICS regions executing on different z/OS images. For
CF-based logstreams, only place logstreams with similar characteristics (such as frequency
and size of data written to the logstream) in the same structure.
Monitoring and tuning CICS logstreams

For monitoring and tuning an existing CICS TS logstream, you can gather the SMF type 88
records and use the sample program SYS1.SAMPLIB(IXGRPT1J) to format the data.
17.4.1 DFHLOG
The DFHLOG logstream is the primary CICS log, often referred to as the CICS System Log.
DFHLOG contains transient data relating to an in-progress unit of work. The data contained
within the logstream is used for dynamic transaction backout (or backward recovery) and
emergency restart. CICS access to data in DFHLOG is provided by system logger.
When CICS is active, it writes information about its transactions to DFHLOG. Periodically
CICS tells the system logger to delete the DFHLOG records related to transactions that have
completed. If the log structure has been defined with enough space, it is unusual for data from
the DFHLOG logstream to be offloaded to a logger offload data set.
DFHLOG logstreams are used exclusively by a single CICS region. You have one DFHLOG
logstream per CICS region.
The log streams are accessible from any system that has connectivity to the CF containing
the logger structure that references the logstreams. If a z/OS system fails, it is possible to
restart the affected CICS regions on another z/OS system, which would still be able to access
their DFHLOG data in the CF.
DFHLOG is required for the integrity of CICS transactions. Failed transactions cannot be
backed out if no backout information is available in DFHLOG. CICS will stop working if it
cannot access the data in the DFHLOG logstream.
17.4.2 DFHSHUNT
The DFHSHUNT logstream is the secondary CICS log, which is also referred to as the CICS
SHUNT Log. It contains ongoing data relating to incomplete units of work; the data is used for
resynchronizing in-doubt UOWs (even on a cold start). Information or data about long-running
transactions is moved (or shunted) from DFHLOG to DFHSHUNT.
The status of a UOW defines whether or not it is removed (shunted) from DFHLOG to the
secondary system log, DFHSHUNT. If the status of a unit of work that has failed is at one of
the following points, it will be shunted from DFHLOG to DFHSHUNT pending recovery from
the failure:
While in doubt during a two-phase commit process.
While attempting to commit changes to resources at the end of the UOW.
While attempting to back out the UOW.
When the failure that caused the data to be shunted is fixed, the shunted UOW is
resolved. This means the data is no longer needed and is discarded.
349
DFHSHUNT logstreams are used exclusively by a single CICS region; you have one
DFHSHUNT logstream for EACH CICS region.
17.4.3 USRJRNL
The USERJRNL logstream contains recovery data for user journals where block writes are
not forced. A block write is several writes (each being a block) to a logstream that may get
grouped together and written as a group rather than being written immediately (block write
forced). The USERJRNL structure is optional and was designed primarily to be customized
and used by customers to manipulate their own data for other purposes.
17.4.4 General
The GENERAL logstream is another, more basic log that contains recovery data for forward
recovery, auto-journaling, and user journals.
17.4.5 Initiating use of the DFHLOG structure

Starting a CICS region will allocate the DFHLOG structure, as illustrated in Figure 17-3.
DFHLG0103I #@$C1T1A System log (DFHLOG) initialization has started.
DFHLG0104I #@$C1T1A 506
System log (DFHLOG) initialization has ended. Log stream
#@$C.#@$C1T1A.DFHLOG2 is connected to structure CIC_DFHLOG_001.
DFHLG0103I #@$C1T1A System log (DFHSHUNT) initialization has started.
Figure 17-3 CICS messages when allocating the DFHLOG structure
17.4.6 Deallocating the DFHLOG structure

Stopping the CICS region will deallocate the DFHLOG structure.
17.4.7 Modifying the size of DFHLOG

There may be a requirement to modify the size of the DFHLOG structure due to increased
activity resulting in a larger structure. Or, the original structure may be oversized and needs to
be decreased.
Perform the following steps using the appropriate z/OS system commands.
1. Check system loggers view of the DFHLOG structure and notice the association between
the structure name and logstream:
D LOGGER,STR,STRNAME=DFHLOG structure name
2. Check the structure's size and location:
D XCF,STR,STRNAME=DFHLOG structure name
3. Check that there is sufficient free space in the current Coupling Facility:
D CF,CFNAME=current CF name
4. Modify the structure size with the ALTER command:
SETXCF START,ALTER,STRNM=DFHLOG structure name,SIZE=new size
350
IXC530I SETXCF START ALTER REQUEST FOR STRUCTURE CIC_DFHLOG_001 ACCEPTED.

IXC533I SETXCF REQUEST TO ALTER STRUCTURE CIC_DFHLOG_001
COMPLETED. TARGET ATTAINED.
CURRENT SIZE:
18432 K TARGET:
18432 K
Figure 17-4 Extend DFHLOG structure
Observe the TARGET ATTAINED response.

5. Verify the results:
Check the ACTUAL SIZE value.
17.4.8 Moving the DFHLOG structure to another Coupling Facility

It may become necessary to move a structure from one CF to another to rebalance workload
between the CFs or to empty out a CF for maintenance. Perform the following steps using the
appropriate z/OS system commands.
1. Check the structure's size and location and that at least two CFs are specified on the
preference list:
2. Check that there is sufficient free space in the target Coupling Facility:
D CF,CFNAME=target CF name
3. Move the DFHLOG structure to the alternate CF. During the rebuild process, the data held
in this structure is not accessible:
SETXCF START,RB,STRNM=DFHLOG structure name,LOC=OTHER
HAS BEEN STARTED
CIC_DFHLOG_001 WAS ACCEPTED.
IXC526I STRUCTURE CIC_DFHLOG_001 IS REBUILDING FROM
INFO108: 00000064 00000064.
HAS BEEN COMPLETED
Figure 17-5 Move DFHLOG structure to alternate CF
4. Verify the results.

a. Check that CFNAME is pointing to the desired target CF.
b. It should still be connected to the same address spaces that it was before REBUILD
was issued. Check the details under CONNECTION NAME.
351
17.4.9 Recovering from a Coupling Facility failure

In the case of a CF failure containing the DFHLOG structure, the following recovery process
occurs:
All z/OS systems in the Parallel Sysplex detect that they have lost connectivity to the CF
and notify their system logger of the connectivity failure.
System loggers on all systems negotiate with each other to determine who will manage
the structure recovery. The system managing the structure recovery tells the other system
logger address spaces in the Parallel Sysplex to stop activity to the failed logger structure
and allocate a new logger structure in another CF in the preference list. System logger on
this system populates the new logger structure with the data from either staging data sets
or a data space.
The managing system logger instructs the other system logger address spaces in the
Parallel Sysplex to populate the new logger structure with data from either staging data
sets or a data space.
All system logger address spaces move their connections from the old logger structure to
the new structure and activity is resumed.
When recovery is complete, system logger moves all the log data into the offload data
sets.
During this time, CICS is unaware that the logger structure is being rebuilt. Consequently,
CICS will issue error messages to indicate it received a logger error and will continue to do so
until the rebuild is complete.
17.4.10 Recovering from a system failure

The action is nondisruptive to the DFHLOG structure on the surviving systems. Only the
failing system will be lost. If a z/OS system fails, it is possible to restart the affected CICS
regions on another z/OS system that is still able to access their DFHLOG data in the CF.
17.5 CICS shared temporary storage

CICS shared temporary storage uses list structures to provide access to non-recoverable
temporary storage queues from multiple CICS regions running on any image in a Parallel
Sysplex. CICS stores a set of temporary storage (TS) queues that you want to share in a TS
pool. Each TS pool corresponds to a Coupling Facility CICS TS list structure.
You can create a single TS pool or multiple TS pools within a single sysplex. For example,
you can create separate pools for specific purposes, such as a TS pool for production or a TS
pool for test and development.
The name of the list structure for a TS data sharing pool is created by appending the TS pool
name to the prefix DFHXQLS_, giving DFHXQLS_poolname.
17.5.1 Initiating use of a shared TS structure

Starting the CICS TS queue server will allocate the DFHXQLS structure; see Figure 17-6 on
page 353.
352
DFHXQ0101I Shared TS queue server initialization is in progress.

IXL014I IXLCONN REQUEST FOR STRUCTURE DFHXQLS_#@$STOR1 490
WAS SUCCESSFUL. JOBNAME: #@$STOR1 ASID: 0062
CONNECTOR NAME: DFHXQCF_#@$2 CFNAME: FACIL01
DFHXQ0401I Connected to CF structure DFHXQLS_#@$STOR1.
AXMSC0051I Server DFHXQ.#@$STOR1 is now enabled for connections.
DFHXQ0102I Shared TS queue server for pool #@$STOR1 is now active
Figure 17-6 Output from a successful start of the CICS TS queue server
Check the structures size, location, and connectors by issuing the z/OS command:
D XCF,STR,STRNAME=DFHXQLS_*
17.5.2 Deallocating a shared TS structure

To stop the use of the structure, use the z/OS modify command to perform an orderly
shutdown of the server or, if required, cancel the server.
Figure 17-7 demonstrates attempting an orderly shutdown when connections are still active:
DFHXQ0304I
DFHXQ0351I
DFHXQ0352I
DFHXQ0303I
DFHXQ0307I
DFHXQ0111I
AXMSC0061I
DFHXQ0461I
DFHXQ0112I
Attempt shutdown, F servername,STOP

Connections still active? 1
Active connections 2
Cancel Server, F servername,CANCEL 3
STOP command is waiting for connections to be closed. Number of active connections = 1.1
Connection: Job #@$C1A2A Appl #@$C1A2A Idle 00:00:04
2
Queue pool #@$STOR1 total active connections: 1.
DISPLAY command has been processed.
CANCEL command has been processed. Number of active connections = 1.
3
Shared TS queue server for pool #@$STOR1 is terminating.
Server DFHXQ.#@$STOR1 is now disabled for connections.
Disconnected from CF structure DFHXQLS_#@$STOR1.
Shared TS queue server has terminated, return code 8, reason code 307.
Figure 17-7 Shutting down shared TS structure
17.5.3 Modifying the size of a shared TS structure

There may be a requirement to modify the structure size due to increased use by your
applications resulting in a larger structure. Or, the original structure may be oversized and
need to be decreased. Perform the following steps using the appropriate z/OS system
commands:
1. Check the structure's size and location.
2. Check that there is sufficient free space in the current CF.
3. Modify the structure size with the ALTER command.
SETXCF START,ALTER,STRNM=TS structure name,SIZE=new size
353
IXC530I SETXCF START ALTER REQUEST FOR STRUCTURE DFHXQLS_#@$STOR1 ACCEPTED.

IXC533I SETXCF REQUEST TO ALTER STRUCTURE DFHXQLS_#@$STOR1
CURRENT SIZE:
22528 K TARGET:
22528 K
Figure 17-8 Extend TS structure

4. Verify the results
D XCF,STR,STRNAME=TS structure name
17.5.4 Moving the shared TS structure to another CF

It may become necessary to move a structure from one CF to another to rebalance workload
preference list:
2. Check that there is sufficient free space in the target CF:
3. Move the Shared TS structure to the alternate CF. During the rebuild process, the data
held in this structure is not accessible:
SETXCF START,RB,STRNM=TS structure name,LOC=OTHER
IXC570I SYSTEM-MANAGED REBUILD STARTED FOR STRUCTURE
DFHXQLS_#@$STOR1 IN COUPLING FACILITY FACIL01
PHYSICAL STRUCTURE VERSION: C0C745D6 18A0004A
LOGICAL STRUCTURE VERSION: C0C745D6 18A0004A
START REASON: OPERATOR-INITIATED
AUTO VERSION: C0D64E3E B59CC06E
DFHXQLS_#@$STOR1 WAS ACCEPTED.
IXC578I SYSTEM-MANAGED REBUILD SUCCESSFULLY ALLOCATED
STRUCTURE DFHXQLS_#@$STOR1.
OLD COUPLING FACILITY: FACIL01
OLD PHYSICAL STRUCTURE VERSION: C0C745D6 18A0004A
NEW COUPLING FACILITY: FACIL02
NEW PHYSICAL STRUCTURE VERSION: C0D64E41 6BC42F45
IXC577I SYSTEM-MANAGED REBUILD HAS
BEEN COMPLETED FOR STRUCTURE DFHXQLS_#@$STOR1
STRUCTURE NOW IN COUPLING FACILITY FACIL02
PHYSICAL STRUCTURE VERSION: C0D64E41 6BC42F45
Figure 17-9 Move TS structure to alternate CF
354
4. Verify the results

D XCF,STR,STRNAME=TS structure name
b. It should still be connected to the same address spaces that it was before the Rebuild
17.5.5 Recovery from a CF failure

In the event of a CF failure where there is no connectivity to the Shared TS Queue structure,
the server will terminate automatically, as displayed in Figure 17-10. The server may be
restarted, where it will attempt to connect to the original structure. If this should fail, it will
allocate a new structure in an alternate CF.
DFHXQ0424 Connectivity has been lost to CF structure 441
DFHXQLS_#@$STOR1. The shared TS queue server cannot continue.
DFHXQ0307I CANCEL RESTART=YES command has been processed. Number of
active connections = 0.
DFHXQ0111I Shared TS queue server for pool #@$STOR1 is terminating.
AXMSC0061I Server DFHXQ.#@$STOR1 is now disabled for connections.
Figure 17-10 Loss of CF connectivity with no duplexing of the structure
If System-Managed Duplexing were used, recovery would have been seamless. See 7.4,
Structure duplexing on page 112 for more details about that topic.
17.5.6 Recovery from a system failure

The action is nondisruptive to the shared TS structure on the surviving systems. Only the
failing system will be lost.
17.6 CICS CF data tables

Coupling Facility data tables (CFDT) provide a method of file data sharing without the need
for a file-owning region and without the need for VSAM RLS support. CICS Coupling Facility
data table support is designed to provide rapid sharing of working data across a sysplex, with
update integrity. Read/write access to CFDTs have similar performance, thus making this
form of table particularly useful for informal shared data. Informal shared data is
characterized as:
Data that is relatively short term in nature, which is either created as the application is
running, or is initially loaded from an external source
Data volumes that are not usually very large
Data that needs to be accessed fast
Data that commonly requires update integrity
355
Figure 17-11 Parallel Sysplex with CFDT servers
Data table pools are CF list structures that are defined in the CFRM policy. There must be a
definition statement in the CFRM policy for each list structure. These statements define
parameters such as maximum size of structure and initial size and preference list for the CFs.
Poolnames are defined in the CFDT server parms and must be in the form
DFHCFLS_user-given poolname.
Each z/OS image must have a CFDT server for each CFDT pool. A CFDT is created
automatically when the first file that names it is opened.
Typical uses might include sharing scratchpad data between CICS regions across a sysplex,
or sharing of files for which changes do not have to be permanently saved. Coupling Facility
data tables are particularly useful for grouping data into different tables, where the items can
be identified and retrieved by their keys. You could use a record in a Coupling Facility data
table to:
Maintain the next free order number for use by an order processing application
Look up tables of telephone numbers
Store data extracted from a larger file or database for further processing
17.6.1 Initiating use of the CFDT structure

Starting the CICS CFDT server will allocate the DFHCFLS structure.
356
DFHCF0101I CF data table server initialization is in progress.

DFHCF0401I Connected to CF structure DFHCFLS_#@$CFDT1.
IXL014I IXLCONN REQUEST FOR STRUCTURE DFHCFLS_#@$CFDT1 095
WAS SUCCESSFUL. JOBNAME: #@$CFDT1 ASID: 0060
CONNECTOR NAME: DFHCFCF_#@$3 CFNAME: FACIL01
AXMSC0051I Server DFHCF.#@$CFDT1 is now enabled for connections.
DFHCF0102I CF data table server for pool #@$CFDT1 is now active.
Figure 17-12 Messages issued when starting the CFDT server
5. Check the structures size, location and connectors by issuing the z/OS command:
D XCF,STR,STRNAME=DFHCFLS_*
17.6.2 Deallocating the CFDT structure

Figure 17-13 demonstrates attempting an orderly shutdown when connections are still active:
1.
2.
3.
4.
Attempt shutdown, F servername,STOP

Connections still active? 1
Active connections 2
Cancel Server, F servername,CANCEL 3
DFHCF0304I STOP command is waiting for connections to be closed.1

Number of active connections = 1.
DFHCF0351I Connection: Job #@$C1A3A Applid #@$C1A3A Idle 00:00:00 2
DFHCF0352I Total connections to this server: 1.
DFHCF0303I DISPLAY command has been processed.
DFHCF0461I Disconnected from CF structure DFHCFLS_#@$CFDT1.3
DFHCF0112I CF data table server has terminated, return code 8, reason
code 307.
IEF352I ADDRESS SPACE UNAVAILABLE
$HASP395 #@$CFDT1 ENDED
Figure 17-13 CFDT server shutdown
17.6.3 Modifying the size of the CFDT structure

applications resulting in a larger structure, or the original structure may be oversized and
need to be decreased.
2. Check that there is sufficient free space in the current CF:
SETXCF START,ALTER,STRNM=CFDT structure name,SIZE=new size
357
IXC530I SETXCF START ALTER REQUEST FOR STRUCTURE DFHCFLS_#@$CFDT1 ACCEPTED.

IXC533I SETXCF REQUEST TO ALTER STRUCTURE DFHCFLS_#@$CFDT1
CURRENT SIZE:
6144 K TARGET:
6144 K
Figure 17-14 Extend CFDT structure

D XCF,STR,STRNAME=CFDT structure name
17.6.4 Moving the CFDT structure to another CF

It may become necessary to move a structure from one CF to another, to rebalance workload
preference list:
3. Move the CFDT structure to the alternate CF. During the rebuild process the data held in
this structure is not accessible:
SETXCF START,RB,STRNM=CFDT structure name,LOC=OTHER
STRUCTURE DFHCFLS_#@$CFDT1.
OLD PHYSICAL STRUCTURE VERSION: C0C745DA B7AD3A4E
BEEN COMPLETED FOR STRUCTURE DFHCFLS_#@$CFDT1
Figure 17-15 Partial output from the CFDT structure rebuild process

D XCF,STR,STRNAME=CFDT structure name
17.6.5 Recovering CFDT after CF failure

In the event of a CF failure where there is no connectivity to the CFDT structure, the server
will terminate automatically as displayed in Figure 17-16 on page 359. The server may be
358
restarted, where it will attempt to connect to the original structure. If this should fail, it will
allocate a new structure in an alternate CF.
DFHCF0424 Connectivity has been lost to CF structure 445
DFHCFLS_#@$CFDT1. The CF data table server cannot continue.
DFHCF0307I CANCEL RESTART=YES command has been processed. Number of
DFHCF0111I CF data table server for pool #@$CFDT1 is terminating.
AXMSC0061I Server DFHCF.#@$CFDT1 is now disabled for connections.
Figure 17-16 Loss of CF connectivity to the CFDT structure
Recovery would have been seamless if System-Managed Duplexing had been used. See 7.4,

The action is nondisruptive to the CFDT Structure on the surviving systems. Only the failing
system will be lost.
17.7 CICS named counter server

CICS provides a facility for generating unique sequence numbers for use by application
programs in a Parallel Sysplex environment. This is provided by a named counter server,
which generates each sequence of numbers using a named counter (where the counter
name is an identifier of up to 16 characters). Each time a sequence number is assigned, the
corresponding named counter is incremented automatically.
A named counter is stored in a named counter pool, which resides in a list structure in the
coupling facility. The list structure name is of the form DFHNCLS_poolname. Different pools
can be created to suit your needs. You could, for example, have a pool for use by production
CICS regions and others for test and development regions.
A named counter pool name can be any valid identifier of up to 8 characters, but by
convention pool names should normally be of the form DFHNCxxx. The default named
counter options table assumes that when an application specifies a pool selector of this form,
it is referring to that physical named counter pool. Any other pool selector for which there is no
specific option table entry is mapped to the default named counter pool for the current region,
or to the standard default pool name DFHNC001 if there is no specific default set for the
current region.
This means that different applications can use their own logical pool names to refer to their
named counters, but the counters will normally be stored in the default pool unless the
installation specifically adds an option table entry to map that logical pool name to a different
physical pool.
The structure size required for a named counter pool depends on the number of different
named counters you need. The minimum size of 256 KB should be enough for most needs,
because it holds hundreds of counters. However you can, if necessary, allocate a larger
structure which can hold many thousands of counters.
359
17.7.1 Initiating use of the NCS structure

Starting the CICS NCS server will allocate the DFHNCLS structure.
DFHNC0101I Named counter server initialization is in progress.
IXL014I IXLCONN REQUEST FOR STRUCTURE DFHNCLS_#@$CNCS1 810
WAS SUCCESSFUL. JOBNAME: #@$CNCS1 ASID: 0050
CONNECTOR NAME: DFHNCCF_#@$2 CFNAME: FACIL01
DFHNC0401I Connected to CF structure DFHNCLS_#@$CNCS1.
AXMSC0051I Server DFHNC.#@$CNCS1 is now enabled for connections.
DFHNC0102I Named counter server for pool #@$CNCS1 is now active.
Figure 17-17 Messages issued when starting the NCS server
1. Check the structures size, location and connectors by issuing the z/OS command:
D XCF,STR,STRNAME=DFHNCLS_*
17.7.2 Deallocating the NCS structure

1. Attempt shutdown: F servername,STOP.
2. If there are still address spaces with active connections, then these will be displayed.
3. To remove these connections, cancel the server: F servername,CANCEL.
17.7.3 Modifying the size of the NCS structure

applications resulting in a larger structure, or the original structure may be oversized and
need to be decreased. Perform the following steps using the appropriate z/OS system
commands.
1. Check the structure's size and location
SETXCF START,ALTER,STRNM=NCS structure name,SIZE=new size
360
Figure 17-18 shows extending the NCS structure.
IXC530I SETXCF START ALTER REQUEST FOR STRUCTURE DFHNCLS_#@$CNCS1

ACCEPTED.
IXC533I SETXCF REQUEST TO ALTER STRUCTURE DFHNCLS_#@$CNCS1 757
CURRENT SIZE:
25088 K TARGET:
25088 K
IXC534I SETXCF REQUEST TO ALTER STRUCTURE DFHNCLS_#@$CNCS1 758
CURRENT SIZE:
25088 K TARGET:
25088 K
CURRENT ENTRY COUNT:
80415
TARGET:
80415
CURRENT ELEMENT COUNT:
0
TARGET:
0
CURRENT EMC COUNT:
0
TARGET:
0
DFHNC0417I Alter request completed normally for CF structure
DFHNCLS_#@$CNCS1.
Figure 17-18 Extend NCS structure

D XCF,STR,STRNAME=NCS structure name
17.7.4 Moving the NCS structure to another CF

It may become necessary to move a structure from one CF to another, to rebalance workload
between the CFs or to empty out a CF for maintenance.
preference list:
3. Move the NCS structure to the alternate CF. During the rebuild process the data held in
this structure is not accessible.
SETXCF START,RB,STRNM=NCS structure name,LOC=OTHER
DFHNCLS_#@$CNCS1 WAS ACCEPTED.
STRUCTURE DFHNCLS_#@$CNCS1.
OLD PHYSICAL STRUCTURE VERSION: C0D7A1C2 1D8E68CE
NEW PHYSICAL STRUCTURE VERSION: C0D7A26A DFA8B64A
BEEN COMPLETED FOR STRUCTURE DFHNCLS_#@$CNCS1
Figure 17-19 Partial output from the NCS Structure rebuild process
361

D XCF,STR,STRNAME=NCS structure name
17.7.5 Recovering NCS after a CF failure

In the event of a CF failure where there is no connectivity to the NCS structure, the server will
terminate automatically, as displayed in Figure 17-20. The server may be restarted, where it
will attempt to connect to the original structure. If this fails, it will allocate a new structure in an
alternate CF.
DFHNC0424 Connectivity has been lost to CF structure 449
DFHNCLS_#@$CNCS1. The named counter server cannot continue.
DFHNC0307I CANCEL RESTART=YES command has been processed. Number of
DFHNC0111I Named counter server for pool #@$CNCS1 is terminating.
AXMSC0061I Server DFHNC.#@$CNCS1 is now disabled for connections.
Figure 17-20 Loss of CF connectivity to the NCS structure
Recovery would have been seamless if System-Managed Duplexing had been used. See 7.4,

The action is nondisruptive to the NCS Structure on the surviving systems. Only the failing
system will be lost.
17.8 CICS and ARM

This section describes how CICS uses the Automatic Restart Manager (ARM) component of
MVS to increase the availability of your systems. The main benefits of the MVS Automatic
Restart Manager are that it:
Enables CICS to preserve data integrity automatically in the event of any system failure.
Eliminates the need for operator-initiated restarts, or restarts by automation, thereby:
Improving emergency restart times
Reducing errors
Reducing complexity.
Provides cross-system restart capability. It ensures that the workload is restarted on MVS
images with spare capacity, by working with the MVS workload manager.
Allows all elements within a restart group to be restarted in parallel. Restart levels (using
the ARM WAITPRED protocol) ensure the correct starting sequence of dependent or
related subsystems.
362
Automatic restart of CICS data-sharing servers

All three types of CICS data-sharing server, temporary storage, Coupling Facility data tables,
and named counters, support automatic restart using the services of Automatic Restart
Manager. The servers also have the ability to wait during startup, using an event notification
facility (ENF) exit, for the Coupling Facility structure to become available if the initial
connection attempt fails.
Data-sharing server ARM processing

During data-sharing initialization, the server unconditionally registers with ARM, except when
starting up for unload or reload. A server will not start if registration fails with return code 8 or
above.
If a server encounters an unrecoverable problem with the Coupling Facility connection,
consisting either of lost connectivity or a structure failure, it cancels itself using the server
command CANCEL RESTART=YES. This terminates the existing connection, closes the
server and its old job, and starts a new instance of the server job.
You can also restart a server explicitly using either the server command CANCEL
RESTART=YES, or the MVS command CANCEL jobname,ARMRESTART
By default, the server uses an ARM element type of SYSCICSS, and an ARM element
identifier of the form DFHxxnn_poolname where xx is the server type (XQ, CF or NC) and nn
is the one- or two-character &SYSCLONE identifier of the MVS image. You can use these
parameters to identify the servers for the purpose of overriding automatic restart options in
the ARM policy. See Chapter 6, Automatic Restart Manager on page 83 for more details.
17.9 CICSPlex System Manager

CICSPlex System Manager (CPSM) is an IBM strategic tool for managing multiple CICS
systems in support of the on demand environment. This component is provided free as part of
CICS TS.
CICS systems and CICSPlexes have become more complex and challenging to manage.
CPSM provides a facility to logically group and manage large numbers of CICS regions.
Figure 17-21 on page 364 illustrates an overview of CPSM.
363
Figure 17-21 CPSM overview
17.10 What is CICSPlex

A CICSPlex is commonly described as a set of interconnected CICS regions that process
customer workload. A typical CICSPlex is the set of interconnected TORs, AORs, FORs, and
so on. For CICSPlex SM purposes, a CICSPlex is any logical grouping of CICS regions that
you want to manage and manipulate as a whole. CICSPlexes are typically groups of systems
that are logically related by usage, for example, test, quality assurance, or production
CICSPlexes.
A CICSPlex managed by CICSPlex SM has the following attributes:
The CICSPlex is the largest unit you can work with. That is, you cannot group CICSPlexes
and manipulate such a group as a single entity.
CICSPlexes are mutually exclusive, so no CICS region can belong to more than one
CICSPlex.
CPSM enables you to define subsets of a CICSPlex, which are known as system groups.
CICS system groups are not mutually exclusive, and can reference the same CICS
regions, or other system groups. CICS system groups are typically used to represent
region types such as the set of TORs or AORs, a physical location such as the CICS
regions on a z/OS image, or a set of CICS regions processing a workload.
364
17.10.1 CPSM components

CPSM consists of the following components.
CICSPlex System Manager

CICSPlex System Manager (CMAS) a CICS region dedicated solely to the CICSPlex SM
function responsible for managing and reporting on all CICS regions and resources within the
defined CICSPlex or CICSPlexes. The CMAS interacts with CICSPlex SM agent code
running on each managed CICS region (MAS) to define events or conditions of interest, and
collect information. A CMAS region is not part of a managed CICSPlex.
Coordinating Address Space

Coordinating Address Space (CAS) is used to set up the CICSPlex SM component topology,
and to support the MVS/TSO ISPF end-user interface (EUI) to CPSM.
Web User Interface

The Web User Interface (WUI) offers an easy-to-use interface that you can use to carry out
operational and administrative tasks necessary to monitor and control CICS resources.
CPSM allows you, via the Web browser interface, to manipulate CICS regions with a single
command.
ISPF interface
An ISPF interface is available to carry out operational and administrative tasks.
Environment Services System Services

Environment Services System Services (ESSS) is an address space that is started
automatically upon startup of the CMAS. It provides MVS system services to the CPSM
components.
Real Time Analysis

Real Time Analysis (RTA) provides system-wide monitoring and problem resolution.
Performance thresholds can be set on all aspects of the system critical to maintaining
performance and throughput.
RTA provides a System Availability Monitor (SAM), which monitors the health of all systems
within the CICSPlex. If a threshold is exceeded anywhere within the CICSPlex, it triggers an
event pinpointing the system and threshold instantly. You are alerted and are able to use
CPSM to diagnose and repair the problem from a single point of control. There is also a
provision to configure automatic recovery from specific error conditions. For example if a DB2
connection is lost RTA reports this and CPSM can take remedial action to restore the
connection without user intervention.
Business Application Services

Business Application Services (BAS) allows you to replicate entire applications and their
resources from a single point of control. BAS is similar to CICS Resource Definition Online
(RDO) with some differences:
A resource can belong to more than one group.
It allows the creation of resource definitions for all supported releases of CICS.
It provides Logical Scoping, which means that resources can be grouped according to the
application they are used in.
SYSLINK construct simplifies connection installation.
365
Workload Manager
The Workload Manager (WLM) component of CPSM provides for dynamic workload
balancing. WLM routes transactions to regions based upon predefined performance criteria. If
one region reaches a performance threshold, either through volume of work or because of
some problem, WLM stops routing work to it until the workload has reduced. WLM, therefore,
ensures optimum capacity usage and throughput, and guards against any system in its
cluster becoming a single point of failure.
Note that with WLM, work is not balanced in a round-robin fashion. WLM selects the system
most likely to meet specified criteria by using either the QUEUE algorithm or GOAL algorithm.
QUEUE algorithm
The QUEUE algorithm uses the following selection criteria:
Selects the system with shortest queue of work relative to system MAXTASKS
The system least likely to be affected by Short on Storage, SYSDUMP, and TRANDUMP
conditions
The system least likely to cause the transaction to abend
Standardizes response times across a CICSPlex
Accommodates differences in processor power and MAXTASK values, asymmetric region
configuration and unpredictable workloads.
GOAL algorithm
The GOAL algorithm uses the following selection criteria:
Selects system least likely to be affected by SOS, SYSDUMP, and TRANDUMP
conditions
The system least likely to cause the transaction to abend
Most likely to meet average MVS WLM response time goals
17.10.2 Coupling Facility structures for CPSM

CPSM does not currently use any Coupling Facility structures.
366
18
Chapter 18.
DB2 operational considerations

This chapter introduces DB2 Data Sharing and provides an overview of operational
considerations when it is implemented in a Parallel Sysplex.
367
18.1 Introduction to DB2

DB2 is a relational database manager that controls access to data from a connecting
application. This data is stored in the form of pages, which are kept in tables. A group of
tables comprise a tablespace. DB2 uses a locking mechanism to control access to the data to
ensure integrity. DB2's Intersystem Resource Lock Manager (IRLM) is both a separate
subsystem and an integral component of DB2.
18.1.1 DB2 and data sharing

Data sharing allows for read and write access to DB2 data concurrently from more than one
DB2 subsystem residing on multiple z/OS systems in a Parallel Sysplex. DB2 subsystems
that share data must belong to a DB2 data sharing group, which is a collection of one or more
DB2 subsystems that access shared DB2 data.
Each DB2 subsystem in a data sharing group is referred to as a member of that group; see
Figure 18-1. All members of the group use the same shared DB2 catalog and DB2 directory,
and must reside in the same Parallel Sysplex. Members of a data sharing group are
independent of the applications that support nondisruptive scalable growth and workload
balancing.
Figure 18-1 Members of a data sharing group
Group attachment
Each DB2 member in a data sharing group must have a unique subsystem name. To facilitate
this, the Group Attachment Name was created. This is a common name that can be used by
batch jobs, utilities, IMS BMPs, and CICS TS to connect to any DB2 subsystem within the
data sharing group. The Group Attachment Name is specified in the IEFSSNxx member of
PARMLIB, or created dynamically via the SETSSI command.
368
18.2 DB2 structure concepts

Members of the DB2 data sharing group use areas of storage in the CF called structures to
communicate and move data among the members. There are three types of structures:
Lock
List
Cache
Lock structure
There is one lock structure per data sharing group. This is used by IRLM to serialize the
resources used by the associated data sharing group. The naming convention for the lock
structure is DB2-data-sharing-groupname_LOCK1.
List structure
There is one list structure per data sharing group used as a Shared Communication Area
(SCA) for the members of the group. The SCA contains all database exception status
conditions, copies of the Boot Strap Data Sets (BSDS), and other information. The naming
convention for the SCA structure is DB2-data-sharing-groupname_SCA.
Cache structure
Group Buffer Pools (GBPs) are used to cache data in the CF and to maintain the consistency
of data across the buffer pools of members of the group by using a cross-invalidating
mechanism. Cross Invalidation is used to notify a member when its local buffer pool contains
an out-of-date copy of the data. The next time the DB2 member tries to use that data, it must
get the current data from either the GBP or DASD.
One GBP is used for all local buffer pools of the same name in the DB2 group that contain
shared data. For example, each DB2 must have a local buffer pool named BP0 to contain the
catalog and directory table spaces. Therefore, you must define a GBP0 in a CF that maps to
local buffer pool BP0.
How GBP works

The solution that is implemented in a Parallel Sysplex is for each database manager to tell
the Coupling Facility (CF) every time it adds a record to its local buffer. The CF then knows
which instances have a copy of any given piece of data. Each instance also tells the CF every
time it updates one of those records. The CF knows who has a copy of each record, and it
also knows who it has to tell when a given record is updated. This is the process of Cross
Invalidation, and it is handled automatically by the database managers and the CF.
18.3 GBP structure management and recovery

If only one member of the DB2 data sharing group is started, then no GBP structures will be
allocated because no other DB2 is using any of the databases. As soon as another member
of the DB2 data sharing group is started and any database is being updated, the GBP0
structure will immediately be allocated. A CF for the Lock and DB2 SCA structures is still
required, even when all members of the data sharing group have been stopped.
In general, if a DB2 member in a data sharing group terminates normally, the connection to
the GBP is deleted. When all the DB2 members in the data sharing group terminate in a
normal fashion, then all connections to the GBP will be deleted and the GBP structure is
deallocated. The structure is deallocated because its structure disposition is DELETE. A
Chapter 18. DB2 operational considerations in a Parallel Sysplex
369
disposition of DELETE implies that the structure will be deallocated when all the connectors
are gone.
Assumptions
In this chapter the following environment is used for our examples:
DB2 Version 8.1
We will assume a DB2-data-sharing-groupname of D#$#
3 x DB2 Subsystem names of D#$1, D#$2, and D#$3 in our DB2 data sharing group
2 x Coupling Facility, FACIL01 and FACIL02
18.3.1 Stopping the use of GBP structures

Check which DB2 systems are active in the group using the DB2 command:
-D#$1 DIS GROUP
DSN7100I -D#$1 DSN7GCMD 421
*** BEGIN DISPLAY OF GROUP(D#$#
) GROUP LEVEL(810) MODE(N)
PROTOCOL LEVEL(2) GROUP ATTACH NAME(D#$#)
-------------------------------------------------------------------DB2
DB2 SYSTEM
IRLM
MEMBER
ID SUBSYS CMDPREF
STATUS
LVL NAME
SUBSYS IRLMPROC
-------- --- ----------- -------- --- -------- ----------D#$1
1 D#$1
-D#$1
ACTIVE
810 #@$2
DR$1
D#$1IRLM
D#$2
2 D#$2
-D#$2
QUIESCED 810 #@$2
DR$2
D#$2IRLM
D#$3
3 D#$3
-D#$3
QUIESCED 810 #@$3
DR$3
D#$3IRLM
-------------------------------------------------------------------SCA
STRUCTURE SIZE:
4096 KB, STATUS= AC,
SCA IN USE:
3 %
LOCK1 STRUCTURE SIZE:
4096 KB
NUMBER LOCK ENTRIES:
1048576
NUMBER LIST ENTRIES:
5408, LIST ENTRIES IN USE:
0
*** END DISPLAY OF GROUP(D#$#
)
Figure 18-2 Display the status of members of the DB2 data sharing group
Determine if any GBP structures are allocated using the z/OS XCF command:
D XCF,STR
STRNAME
ALLOCATION TIME
D#$#_GBP0
--D#$#_GBP1
--D#$#_GBP32K
--D#$#_GBP32K1
--D#$#_LOCK1
06/20/2007 03:32:15
D#$#_SCA
06/20/2007 03:32:10
STATUS
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
ALLOCATED
ALLOCATED
Figure 18-3 Status display of all DB2 structures
370
TYPE
LOCK
LIST
1
2
18.3.2 Deallocate all GBP structures

Stopping all DB2 members of the data sharing group will remove the allocation of DB2 GBP
structures, as displayed in Figure 18-3. Note that 1 the LOCK1 and 2 SCA structures are still
allocated although all members of the DB2 data sharing group have been stopped.
18.4 DB2 GBP user-managed duplexing

The following definitions highlight the major differences between user-managed duplexing
and system-managed duplexing:
The connector is responsible for constructing the new

instance and maintaining the duplicate data for a duplexed
structure.
System-managed duplexing The system, not the connector, is responsible for

propagating data from the old instance to the new instance
and, for a duplexed structure, maintaining duplicate data.
One method of achieving higher availability for your GBP structures in case of planned or
unplanned outages of the CF hosting the GBP structure is to run your GBP in duplex mode.
When DB2 duplexes a GBP, the same structure name will be used for both structures. A
duplexed structure requires just a single CFRM policy definition with one structure name. For
duplexed GBPs, there is only one set of connections from each member.
Each GBP structure must be in a different CF.
One instance is called the primary structure. DB2 uses the primary structure to keep the
page registration information that is used for cross-invalidation.
The other instance is called the secondary structure.
Changed pages are written to both the primary structure and the secondary structure.
Changed pages are written to the primary structure synchronously and to the secondary
structure asynchronously. When a GBP is duplexed, only the primary structure is used for
castout processing; pages are never read back from the secondary structure. When a set of
pages has been written to DASD from the primary structure, DB2 deletes those pages from
the primary and secondary structures.
When planning for storage, you must allow for the same amount of storage for the primary
and secondary structures. When using duplexing the secondary structure uses the storage
that would normally be reserved for the primary structure should you have to rebuild it.
There are two ways to start duplexing for a group buffer pool:
Activate a new CFRM policy with DUPLEX(ALLOWED) for the structure.
This allows the GBP structures to be duplexed. However, the duplexing must be initiated
by a SETXCF command; otherwise, the system will not automatically duplex the structure.
This option would normally be used while you are testing duplexing, before deciding
whether you want to use it on a permanent basis.
Activate a new CFRM policy with DUPLEX(ENABLED) for the structure.
If the group buffer pool is currently allocated, then XCF will automatically initiate the
process to establish duplexing as soon as you activate the policy. If the group buffer pool
is not currently allocated, then the duplexing process will be initiated automatically the
next time the group buffer pool is allocated.
371
This option would normally be used when you have finished testing duplexing and have
decided that you want to use it on a permanent basis.
When considering duplexing, note the following to avoid confusion:
The primary structure is referred to as the OLD structure in many of the displays and the
more recently allocated secondary structure is called the NEW structure.
When moving a structure to another Coupling Facility or implementing a duplex structure,
it is important to ensure that enough storage capacity is available to create every other
structure that could be allocated in that CF. The z/OS system programmer is responsible
for ensuring that there is sufficient storage capacity available in the Coupling Facility to
allow every structure that needs to be allocated in a failure scenario.
Be aware that if a structure is manually moved or duplexed, this has to be included in the
calculation. Otherwise, another structure allocation may fail or a duplex recovery may fail
due to CF storage not being available.
18.4.1 Preparing for user-managed duplexing

To display the status of the GBP, DB2-data-sharing-groupname_GBP1, use the following
XCF command after all required DB2 subsystems have initialized:
Figure 18-4 on page 373 displays the output of the XCF view of the GBP structure.
372
D XCF,STR,STRNM=D#$#_GBP1
STRNAME: D#$#_GBP1
STATUS: ALLOCATED 1
TYPE: CACHE
POLICY INFORMATION:
POLICY SIZE : 8192 K
FULLTHRESHOLD : 80
ALLOWAUTOALT : NO
DUPLEX : ALLOWED 2
ENFORCEORDER : NO
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/29/2007 01:07:414
CFNAME : FACIL02 5
ACTUAL SIZE : 4096 K 6
ENTRIES: IN-USE: 18 TOTAL: 2534, 0% FULL
ELEMENTS: IN-USE: 18 TOTAL: 506, 3% FULL
PHYSICAL VERSION: C0932A63 30E5DD94
LOGICAL VERSION: C0932A63 30E5DD94
DISPOSITION : DELETE 7
ACCESS TIME : 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3 8
---------------- -- -------- -------- -------- ---- ---------DB2_D#$1 02 00020021 #@$1 D#$1DBM1 0044 ACTIVE 8
DB2_D#$2 01 0001001D #@$2 D#$2DBM1 003A ACTIVE
DB2_D#$3 03 00030019 #@$3 D#$3DBM1 003C ACTIVE
Figure 18-4 Output of the XCF view of the GBP structure
Figure 18-4 indicates:

1 Structure has a status of allocated.
2 Duplexing is permitted but can only be initiated by SETXCF START,REBUILD command.
3 The preference list includes CF FACIL01 and FACIL02.
4 Date and time the structure was allocated.
5 Currently allocated in CF FACIL02.
6 Actual size of structure, 4096 K.
7 The structure disposition is DELETE.
8 The number of connections, including DB2 subsystem information.
Before initiating duplexing, use the following command to verify that there is enough space
available in the target CF to allocate the second copy of GBP:
D CF,CFNM=FACIL01
373
Note that the primary (OLD) structure is allocated in FACIL02

D CF,CFNM=FACIL01
NAMED FACIL01
ALLOCATED SPACE
STRUCTURES:
209920 K
DUMP SPACE:
2048 K
TABLE COUNT:
FREE SPACE:
511488 K 1
FREE DUMP SPACE:
TOTAL SPACE:
723456 K
TOTAL DUMP SPACE:
VOLATILE:
YES
CFLEVEL:
14
BUILT ON 03/26/2007 AT 17:58:00
IN USE
CONTROL SPACE:
211968 K
NON-CONTROL SPACE:
0 K
FREE
511488 K
0 K
0
0
2048
2048
0
256
K
K
K
K
K
TOTAL
723456 K
0 K
Figure 18-5 Display CF information
1 displays the free space available in the target CF.

Comparing the actual size of the GBP structure detailed in Figure 18-4 on page 373, we have
ascertained that there is adequate space in our target CF, FACIL01, for a duplicate copy.
18.4.2 Initiating user-managed duplexing

At this point, sufficient information has been provided by the displays for us to start the
duplexing process for the GBP structure by using the following command:
SETXCF START,REBUILD,DUPLEX,STRNAME=D#$#_GBP1
374
IXC521I REBUILD FOR STRUCTURE D#$#_GBP1

HAS BEEN STARTED
D#$#_GBP1 WAS ACCEPTED.
DSNB740I -D#$1 DSNB1RBQ ATTEMPTING TO ESTABLISH
DUPLEXING FOR
GROUP BUFFER POOL GBP1
REASON = OPERATOR
DSNB740I -D#$2 DSNB1RBQ ATTEMPTING TO ESTABLISH
DUPLEXING FOR
REASON = OPERATOR
IXC529I DUPLEX REBUILD NEW STRUCTURE D#$#_GBP1
IS BEING ALLOCATED IN COUPLING FACILITY FACIL01 1 .
OLD STRUCTURE IS ALLOCATED IN COUPLING FACILITY FACIL02 2 .
REBUILD START REASON: OPERATOR INITIATED.
INFO108: 00000002 00000000.
DSNB302I -D#$2 DSNB1RBC GROUP BUFFER POOL GBP1-SEC IS
ALLOCATED IN A VOLATILE STRUCTURE
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE D#$#_GBP1
WAS SUCCESSFUL. JOBNAME: D#$2DBM1 ASID: 0085
CONNECTOR NAME: DB2_D#$2 CFNAME: FACIL01
DSNB332I -D#$2 DSNB1PCD THIS MEMBER HAS COMPLETED
CASTOUT OWNER WORK FOR GROUP BUFFER POOL GBP1
PAGES CAST OUT FROM ORIGINAL STRUCTURE = 0
PAGES WRITTEN TO NEW STRUCTURE = 0 3
DSNB332I -D#$1 DSNB1PCD THIS MEMBER HAS COMPLETED
CASTOUT OWNER WORK FOR GROUP BUFFER POOL GBP1
PAGES CAST OUT FROM ORIGINAL STRUCTURE =
PAGES WRITTEN TO NEW STRUCTURE = 0
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE D#$#_GBP1
WAS SUCCESSFUL. JOBNAME: D#$1DBM1 ASID: 0037
CONNECTOR NAME: DB2_D#$1 CFNAME: FACIL01
IXL015I REBUILD NEW STRUCTURE ALLOCATION INFORMATION FOR
STRUCTURE D#$#_GBP1, CONNECTOR NAME DB2_D#$2
CFNAME ALLOCATION STATUS/FAILURE REASON
FACIL02 RESTRICTED BY REBUILD OTHER
FACIL01 STRUCTURE ALLOCATED
HAS REACHED THE DUPLEXING ESTABLISHED PHASE
DSNB333I -D#$2 DSNB1GBR FINAL SWEEP COMPLETED FOR
PAGES WRITTEN TO NEW STRUCTURE = 0
DSNB742I -D#$1 DSNB1GBR DUPLEXING HAS BEEN
SUCCESSFULLY ESTABLISHED FOR
Figure 18-6 Output from user initiated structure duplexing
From the output in Figure 18-6, note that the new structure has been allocated in 1 CF
FACIL01 while the OLD structure still exists in 2 CF FACIL02. During the duplexing process,
changed pages are copied from the primary to the secondary structure. The message 3
DSNB332I shows that nothing in the primary structure has been copied to the secondary
structure. This could be because the duplexing process was started before the jobs got a
chance to write any data into the GBP.
You may find that one of the systems will copy all the pages while the other DB2 Group
members do not copy anything. This is because the castout owner is the one that is
375
responsible for copying pages for the page sets that it owns from the primary structure to the
secondary one.
The castout owner is generally the DB2 subsystem that first updates a given page set or
partition. In this instance, DB2 must cast out any data that will not fit in the secondary
structure. If this happens, you will get a non-zero value in the PAGES CAST OUT FROM
ORIGINAL STRUCTURE field in the DSNB332I message, and this should be treated as an
indicator of a possible problem.
If the secondary structure is smaller than the primary one, DB2 will treat both structures as
though they were the same size as the smaller of the two, resulting in wasted CF storage and
degraded performance.
18.4.3 Checking for successful completion

When you receive the DSNB742I message, the duplexing process should be complete. To
confirm this, display the duplexed structure by issuing the following XCF command:
376

STRNAME: D#$#_GBP1
STATUS: REASON SPECIFIED WITH REBUILD START:
OPERATOR INITIATED
DUPLEXING REBUILD 1
METHOD : USER-MANAGED
REBUILD PHASE: DUPLEX ESTABLISHED 2
TYPE: CACHE
POLICY INFORMATION:
POLICY SIZE : 8192 K
POLICY INITSIZE: 4096 K 3
FULLTHRESHOLD : 80
ALLOWAUTOALT : NO
DUPLEX : ALLOWED 4
ENFORCEORDER : NO
DUPLEXING REBUILD NEW STRUCTURE 5
------------------------------ALLOCATION TIME: 06/29/2007 01:07:44 6
CFNAME : FACIL01 7
...
DISPOSITION : DELETE
ACCESS TIME : 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3
DUPLEXING REBUILD OLD STRUCTURE 11
------------------------------ALLOCATION TIME: 06/28/2007 23:07:41 9
CFNAME : FACIL02 10
...
MAX CONNECTIONS: 32
# CONNECTIONS : 3
---------------- -- -------- -------- -------- ---- ---------------DB2_D#$1 02 00020021 #@$1 D#$1DBM1 0044 ACTIVE NEW,OLD 12
DB2_D#$2 01 0001001D #@$2 D#$2DBM1 003A ACTIVE NEW,OLD
DB2_D#$3 03 00030019 #@$3 D#$3DBM1 003C ACTIVE NEW,OLD
Figure 18-7 Output of the XCF view of the GBP structure after user-initiated duplexing
Figure 18-7 displays the status after the user-initiated duplexing of the GBP structure:
The information in the STATUS field tells you that a rebuild has been initiated. It also
indicates that the rebuild was a duplexing rebuild 1 and that the duplex pair has been
established 2.
The structure INITSIZE as defined in the CFRM policy 3. This value is used when
allocating the secondary structure. If the size of the primary structure was altered prior to
377
the start of the duplexing, then the primary structure will be a different size than the
secondary structure.Check to see that the POLICY INITSIZE is the same as the ACTUAL
SIZE for each of the structure instances 8. If it is not, then inform your DB2 systems
programmer as soon as possible.
The DUPLEX option defined in the CFRM policy is ALLOWED 4. This means that
duplexing must be started by the operator instead of the system starting it automatically,
as would happen if DUPLEX was set to ENABLED.
The information relating to the secondary (NEW) structure 5 indicates that, because this is
a duplexed structure, then the primary (OLD) structure will not be deleted. For the
secondary structure, notice that ALLOCATION TIME 6 for this structure is later than the
ALLOCATION TIME 9 for the primary structure.
The CF that this structure allocated in 7 must be a different CF from the primary structure
in 10.
The same information is provided for the primary (OLD) structure 11.
Looking at the list of connections, notice that three lines are still provided, one for each of
the DB2 subsystems, with new information in the STATE field 12. For a simplex structure,
this usually says ACTIVE or possibly FAILED-PERSISTENT. Now, however, it displays
the status ACTIVE and tells which structure instance each DB2 has a connection to. In our
example, all three DB2s have an ACTIVE connection to both the OLD and NEW structure
instances.
Displaying the DB2 view of the structure

Figure 18-8 on page 379 displays the DB2 view of the structure after issuing the following
DB2 command:
DB2-command prefix DIS GBPOOL(GBP1)
378
DSNB750I -D#$1 DISPLAY FOR GROUP BUFFER POOL GBP1 FOLLOWS

DSNB755I -D#$1 DB2 GROUP BUFFER POOL STATUS 359
CONNECTED = YES
CURRENT DIRECTORY TO DATA RATIO = 5
PENDING DIRECTORY TO DATA RATIO = 5
CURRENT GBPCACHE ATTRIBUTE = YES
PENDING GBPCACHE ATTRIBUTE = YES
DSNB756I -D#$1 CLASS CASTOUT THRESHOLD = 10% 360
GROUP BUFFER POOL CASTOUT THRESHOLD = 50%
GROUP BUFFER POOL CHECKPOINT INTERVAL = 8 MINUTES
RECOVERY STATUS = NORMAL
AUTOMATIC RECOVERY = Y
DSNB757I -D#$1 MVS CFRM POLICY STATUS FOR D#$#_GBP1 = NORMAL
MAX SIZE INDICATED IN POLICY = 8192 KB
DUPLEX INDICATOR IN POLICY = ALLOWED
CURRENT DUPLEXING MODE = DUPLEX 1
ALLOCATED = YES
DSNB758I -D#$1 ALLOCATED SIZE = 2048 KB362
VOLATILITY STATUS = VOLATILE
REBUILD STATUS = DUPLEXED 2
CFNAME = FACIL02
FLEVEL = 8
DSNB759I -D#$1 NUMBER OF DIRECTORY ENTRIES = 1809 363
NUMBER OF DATA PAGES = 360
NUMBER OF CONNECTIONS = 3
DSNB798I -D#$1 LAST GROUP BUFFER POOL CHECKPOINT 364
03:05:45 JUN 29, 2007
GBP CHECKPOINT RECOVERY LRSN = B5FB0016F8EB
STRUCTURE OWNER = D#$2 3
DSNB799I -D#$1 SECONDARY GBP ATTRIBUTES 365
ALLOCATED SIZE = 4096 KB
VOLATILITY STATUS = VOLATILE
CFNAME = FACIL01
CFLEVEL = 8
NUMBER OF DIRECTORY ENTRIES = 1809
NUMBER OF DATA PAGES = 360
DSNB790I -D#$1 DISPLAY FOR GROUP BUFFER POOL GBP1 IS COMPLETE
DSN9022I -D#$1 DSNB1CMD '-DIS GBPOOL' NORMAL COMPLETION
Figure 18-8 Output from DIS GBPOOL(GBP1)
1 DUPLEX is the current duplexing mode.

2 REBUILD STATUS is DUPLEXED.
3 D#$2 is the structure owner, as indicated by message DSNB798I. Be aware that this is not
the same as being the castout owner. Castout owners are at the page set or partition level,
and you could potentially have a number of castout owners for a given structure, depending
on how many page sets (or partitions) use that buffer pool. The structure owners are
responsible for structure-level activities such as rebuild processing and initiating activities
when thresholds are reached.
The XCF view of the Coupling Facility

To display the XCF view of the contents of both CFs. issue the system command:
D XCF,CF,CFNAME=*
379

CFNAME: FACIL01
COUPLING FACILITY : SIMDEV.IBM.EN.0000000CFCC1
SITE : N/A
POLICY DUMP SPACE SIZE: 2000 K
ACTUAL DUMP SPACE SIZE: 2048 K
CONNECTED SYSTEMS:
#@$1 #@$2 #@$3 1
STRUCTURES:
D#$#_GBP1(NEW) 2 D#$#_LOCK1 D#$#_SCA
IRRXCF00_P001 IXC_DEFAULT_2 SYSTEM_LOGREC
SYSTEM_OPERLOG
CFNAME: FACIL02
COUPLING FACILITY : SIMDEV.IBM.EN.0000000CFCC2
SITE : N/A
POLICY DUMP SPACE SIZE: 2000 K
ACTUAL DUMP SPACE SIZE: 2048 K
CONNECTED SYSTEMS:
#@$1 #@$2 #@$3 1
STRUCTURES:
D#$#_GBP0 D#$#_GBP1(OLD) 2 IGWLOCK00
IRRXCF00_B001 ISGLOCK ISTGENERIC
IXC_DEFAULT_1
Figure 18-9 Displaying XCF information about the CFs
Figure 18-9 shows the structure D#$#_GBP1 2 is in both FACIL01 and FACIL02. The process
would be exactly the same for any of the other GBP structures. Note that the OLD structure is
the Primary and the NEW structure is the secondary. Updates are made to both OLD and
NEW structures, but reads are only done from the OLD (primary) structure.
1 identifies systems connected to the Coupling Facility.
18.5 Stopping DB2 GBP duplexing

You are only likely to stop duplexing if you want to take one of the Coupling Facilities offline.
If you want to empty a CF, you would use the structure rebuild function to move any
structures in that CF to an alternate CF. For duplexed structures, however, if you want to
empty a CF you would stop duplexing, keeping the structure instance that is not in the CF that
you want to stop.
The following steps are required:
When you need to stop duplexing DB2 GBP structures, you must decide which of the
structure instances is to remain as the surviving simplex GBP structure.
If you simply want to stop duplexing, keep the primary GBP because the primary GBP
structure is the one that contains the page registration information.
If you keep the secondary instance, which does not contain the page registration
information, all pages in all the associated local buffer pools will be invalidated, causing a
performance impact until the local buffers are repopulated from DASD or the GBP.
380
The following example demonstrates how to keep the primary GBP as the surviving one. The
process is almost the same regardless of which instance you are keeping.
Checking duplex status

Assuming that all the DB2 subsystems are started, and that the GBP1 structure is in duplex
mode:
1. Check the status of your DB2 subsystems with the following DB2 command:
DB2-command prefix DIS GROUP.
2. Check that the GBP1 structure is in duplex mode by issuing the following DB2 command:
DB2-command prefix DIS GBPOOL(GBP1).
3. The response should indicate CURRENT DUPLEXING MODE as DUPLEX.
Stopping duplexing
Issue the command to stop duplexing, specifying that you want to keep the primary (OLD)
instance of the structure:
SETXCF STOP,REBUILD,DUPLEX,STRNAME=D#$#_GBP1,KEEP=OLD
Specifying KEEP=OLD tells XCF that you want to stop duplexing and retain the primary
structure. To continue using the secondary structure, and delete the primary, you would
specify KEEP=NEW. You may want do this if you had to remove the CF containing the
primary structure, but be aware of the performance impact.
Figure 18-10 on page 382 displays the messages issued while duplexing is being stopped.
381
SETXCF STOP,REBUILD,DUPLEX,STRNAME=D#$#_GBP1,KEEP=OLD
IS BEING STOPPED TO FALL BACK TO THE OLD STRUCTURE DUE TO
REQUEST FROM AN OPERATOR 1
IXC367I THE SETXCF STOP REBUILD REQUEST FOR STRUCTURE
D#$#_GBP1 WAS ACCEPTED.
DSNB742I -D#$2 DSNB1GBR DUPLEXING HAS BEEN
SUCCESSFULLY ESTABLISHED FOR
DSNB743I -D#$1 DSNB1GBR DUPLEXING IS BEING STOPPED
FOR GROUP BUFFER POOL GBP1
FALLING BACK TO PRIMARY
REASON = OPERATOR 2
DB2 REASON CODE = 00000000
IXC579I NORMAL DEALLOCATION FOR STRUCTURE D#$#_GBP1 IN
HAS BEEN COMPLETED.
PHYSICAL STRUCTURE VERSION: B48FCF37 01F2E844
INFO116: 13089180 01 2800 00000004
TRACE THREAD: 00001A1B.
DSNB743I -D#$2 DSNB1GBR DUPLEXING IS BEING STOPPED
FOR GROUP BUFFER POOL GBP1
FALLING BACK TO PRIMARY
REASON = OPERATOR
DB2 REASON CODE = 00000000
HAS BEEN STOPPED
DSNB745I -D#$1 DSNB1GBR THE TRANSITION BACK TO
SIMPLEX MODE HAS COMPLETED FOR
GROUP BUFFER POOL GBP1 3
Figure 18-10 Stopping structure duplexing
Two messages confirm that the secondary GBP structure is being deallocated from CF
FACIL01 and that duplexing is stopped.
The first message, IXC522I, is from XCF and it states that the rebuild has stopped and is
falling back to the OLD structure 1. This is due to the operator requested change.
Message DSNB743I is from DB2 (you will get one of these from every DB2 in the data
sharing group). It advises that it is falling back to the primary structure, and that the change
was requested by an operator 2.
After the rebuild has stopped, you receive message IXC579I indicating that the structure has
been deleted from FACIL01.The last message, DSNB745I, indicates that the processing
related to switching back to simplex mode has completed 3.
Check for successful completion by confirming that GBP duplexing has stopped:
The following should be true:
The GBP structure should only be located in one Coupling Facility.
In SIMPLEX mode, there is only one structure, which should have a state of ACTIVE. This
was the primary structure previously called the OLD structure when duplexed.
382
Displaying the GBP structure status in DB2

Display the DB2 view of the structure by using the following DB2 command:
DB2-command prefix DIS GBPOOL(GBP1)
The output displayed in Figure 18-11 now shows that the 1 CURRENT DUPLEXING MODE is
SIMPLEX and the location of this structure 2 is FACIL02.
-D#$3 DISPLAY FOR GROUP BUFFER POOL GBP1 FOLLOWS
-D#$3 DB2 GROUP BUFFER POOL STATUS
CONNECTED
= NO
CURRENT DIRECTORY TO DATA RATIO
= 6
PENDING DIRECTORY TO DATA RATIO
= 6
CURRENT GBPCACHE ATTRIBUTE
= YES
PENDING GBPCACHE ATTRIBUTE
= YES
-D#$3
CLASS CASTOUT THRESHOLD
= 10%
GROUP BUFFER POOL CASTOUT THRESHOLD
= 50%
GROUP BUFFER POOL CHECKPOINT INTERVAL
= 8 MINUTES
RECOVERY STATUS
= NORMAL
AUTOMATIC RECOVERY
= Y
-D#$3 MVS CFRM POLICY STATUS FOR D#$#_GBP1
= NORMAL
MAX SIZE INDICATED IN POLICY
= 8192 KB
DUPLEX INDICATOR IN POLICY
= ENABLED
CURRENT DUPLEXING MODE
= SIMPLEX
1
ALLOCATED
= YES
-D#$3
ALLOCATED SIZE
= 4096 KB
VOLATILITY STATUS
= VOLATILE
REBUILD STATUS
= NONE
CFNAME
= FACIL02
2
CFLEVEL - OPERATIONAL
= 14
CFLEVEL - ACTUAL
= 14
-D#$3
NUMBER OF DIRECTORY ENTRIES
= 3241
NUMBER OF DATA PAGES
= 538
NUMBER OF CONNECTIONS
= 1
-D#$3 LAST GROUP BUFFER POOL CHECKPOINT
18:14:20 JUL 5, 2007
GBP CHECKPOINT RECOVERY LRSN
= C0D9BAC2655E
STRUCTURE OWNER
=
-D#$3 DISPLAY FOR GROUP BUFFER POOL GBP1 IS COMPLETE
-D#$3 DSNB1CMD '-DIS GBPOOL' NORMAL COMPLETION
Figure 18-11 Output from DISPLAY GBPOOL command
18.6 Modifying the GBP structure size

There are two methods for changing the size of GBP structures, static and dynamic. Using
the static method, the size of the structure in the CFRM policy is modified and then rebuilt
with the SETXCF START,REBUILD command. An alternative MVS command, SETXCF
START,REALLOCATE, will resolve all pending changes from the activation of a new policy. (Be
aware, however, that the SETXCF START,REALLOCATE command may reposition structures that
other people are working with.)
The main characteristics of the static method are:
It is permanent. If the structure is deleted and allocated again, it will still have the new size.
If the actual size of the structure has already reached the SIZE value specified in the
CFRM policy, this is the only way to make it larger.
383
The main characteristics of the dynamic method are:

You can use the SETXCF START,ALTER command to change the size of the structure
dynamically, without having to make any changes to the CFRM policy, as long as the
target size is equal to or less than the SIZE specified in the CFRM policy. The advantage
of this method is that it is faster than having to update the CFRM policy.
The disadvantage is that if the structure is deleted and reallocated, then the change that
you made will be lost and the structure will be allocated again using the INITSIZE value
specified in the CFRM policy.
Any time a structure is altered using the dynamic method, and you intend the change to be
permanent, the CFRM policy must be updated to reflect the change.
There may be a time when you want to increase the structure size because it is currently too
small. If the GBP is too small, you may observe that the threshold for changed pages is
reached more frequently, causing data to be cast out to DASD sooner than is desirable and
impacting performance. If the structure is too small, you may have too few directory entries in
the GBP, resulting in directory reclaims which in turn cause unnecessary buffer invalidation
and a corresponding impact on DB2 performance.
If the GBP is defined too large, this will result in wasted CF storage; for example, if table
spaces get moved from one buffer pool to another.
The SETXCF START,ALTER command can be used to increase the structure up to the SIZE
parameter defined in CFRM policy. If MINSIZE is specified in the CFRM policy, then you
cannot ALTER the structure to be smaller than that value. However, it is possible to change
the size to be smaller than INITSIZE, if appropriate.
Note that, for a duplexed GBP, when you change the size of the primary structure, both the
primary and secondary will get adjusted to the same size, assuming there is sufficient space
in both CFs.
18.6.1 Changing the size of a DB2 GBP structure

Perform the following steps using the appropriate system commands:
1. Check the current GBP1 structure's size and location.
...
POLICY INFORMATION:
POLICY SIZE : 8192 K 1
POLICY INITSIZE: 4096 K2
...
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/29/2007 01:01:45
CFNAME : FACIL02 4
...
Figure 18-12 Output from D XCF,STR,STRNAME
1 POLICY SIZE is the largest size that the structure can be increased to, without updating the
CFRM policy.
384
2 POLICY INITSIZE is the initial size of the structure, as defined in the CFRM policy.
3 ACTUAL SIZE is the size of the structure at this time.
4 CFNAME: FACIL02 details where the structure currently resides.
D CF,CFNAME=FACIL02
The field FREE SPACE: displays the amount of space available.
3. Extend the structure size with the ALTER command.
SETXCF START,ALTER,STRNM=D#$#_GBP1,SIZE=8192
SETXCF START,ALTER,STRNAME=D#$#_GBP1,SIZE=8192
IXC530I SETXCF START ALTER REQUEST FOR STRUCTURE D#$#_GBP1 ACCEPTED.
IXC533I SETXCF REQUEST TO ALTER STRUCTURE D#$#_GBP1 064
CURRENT SIZE: 8192 K TARGET: 8192 K
IXC534I SETXCF REQUEST TO ALTER STRUCTURE D#$#_GBP1 065
CURRENT SIZE: 8192 K TARGET: 8192 K
CURRENT ENTRY COUNT: 6142 TARGET: 6142
CURRENT ELEMENT COUNT: 1228 TARGET: 1228
CURRENT EMC COUNT: 0 TARGET: 0
Figure 18-13 Output from SETXCF ALTER

-D#$1 DIS GBPOOL(GBP1)
DSNB750I -D#$1 DISPLAY FOR GROUP BUFFER POOL GBP1 FOLLOWS
DSNB755I -D#$1 DB2 GROUP BUFFER POOL STATUS 287
CONNECTED = YES
CURRENT DIRECTORY TO DATA RATIO = 5
PENDING DIRECTORY TO DATA RATIO = 5
CURRENT GBPCACHE ATTRIBUTE = YES
PENDING GBPCACHE ATTRIBUTE = YES
DSNB757I -D#$1 MVS CFRM POLICY STATUS FOR D#$#_GBP1 = NORMAL
289
MAX SIZE INDICATED IN POLICY = 8192 KB
DUPLEX INDICATOR IN POLICY = ALLOWED
CURRENT DUPLEXING MODE = SIMPLEX
ALLOCATED = YES
DSNB758I -D#$1 ALLOCATED SIZE = 8192 KB
Figure 18-14 Output from DISPLAY GBPOOL command
The DSNB758I message shows that the new size has been allocated.
To make these changes permanent, modify the structure size definition in the CFRM policy
and then rebuild the structure.
18.6.2 Moving GBP structures

There may be occasions when you have to remove GBP structures from their current CF for
maintenance purposes or to balance workload across all CFs
385
Note that you cannot rebuild a duplexed structure. A duplexed structure already has a copy in
both CFs. If you want to free up one CF, you would revert to simplex mode, deleting the
structure that is in the CF that you wish to free up.
If you have a number of duplexed structures in a CF that you want to empty out, you can
issue the command SETXCF STOP,REBUILD,DUPLEX,CFNAME=Target CF to revert all the
structures to simplex mode, and to delete whichever structure instance (primary or
secondary) might be in the named CF.
To move a GBP structure to another CF using REBUILD, follow these steps.
1. Check the current GBP1 structure size, location, and connectors and the preference list:
2. Check that enough free space is available in the new location:
D CF,CFNAME=Target CF name
3. The structure must be allocated in the alternate CF; this will be the next CF in the
preference list. Perform the rebuild:
SETXCF START,RB,STRNM=D#$#_GBP1,LOC=OTHER
4. All structure data is copied from the old structure to the new structure.
5. All the connections are moved from the original structure to the new structure.
6. Activity is resumed.
7. The original structure is deleted.
8. Check the current GBP1 structure size, location and connectors:
The GBP structure should now be allocated in the target Coupling Facility and all DB2
systems should still have ACTIVE connections to the structure.
18.6.3 GBP simplex structure recovery after a CF failure

A CF failure does not cause all members in a DB2 data sharing group to fail, but it can mean
a temporary availability impact for applications that depend on the data in that GBP. To
minimize that impact, the structure must be recovered as quickly as possible.
If a CF fails or the structure is damaged, all the systems connected to the CF structure detect
the failure. The first DB2 subsystem to detect the failure initiates the structure rebuild
process, which results in the recovery of the contents of the affected structure or structures.
All the DB2 subsystems in the sysplex that are connected to the structure participate in the
process of rebuilding the contents in a new structure in one of the other CFs contained in the
preference list. DB2 recovers the structure by recreating the structure contents from
in-storage information from each DB2 that was connected to the structure.
DB2 members will continue to operate without the use of the GBP. However, the requests
needing access to data in an affected GBP are rejected with a -904 SQL return code. This
indicates that a request failed because a required resource was unavailable at the time of the
request.
When all members of a data sharing group lose connectivity to the CF, DB2 does the
following:
1. The DB2 member that detects the problem puts the GBP in Damage Assessment Pending
(DAP) status.
386
2. DB2 adds entries to the Logical Page List (LPL), if required.

3. The Damage Assessment process determines which page sets were
group-buffer-pool-dependent and places them in Group Recovery Pending (GRECP)
status. As long as the resource remains in GRECP status, it is inaccessible to any
application program.
4. START DATABASE commands are automatically issued by DB2 to recover all table
spaces from the GRECP status. If this does not cause successful recovery, you may have
to issue the START DATABASE command manually.
5. The GBP Damage Assessment Pending (DAP) status is reset.
DB2 terminology for DAP status

The DB2 terminology used for Damage Assessment Pending status is explained here:
Damage Assessment Pending (DAP)
The GBP uses information in the lock structure and SCA to determine which databases
must be recovered. The Shared Communication Area (SCA) is a list structure used by all
data sharing group members to pass control information back and forth. The SCA is also
used to provide recovery of the data sharing group.
Logical Page List (LPL)
Some pages were not read from or written to the GBP because of a failure, such as
complete loss of link connectivity between the GBP and the processor, or some pages
could not be read from or written to DASD because of a DASD problem. Typically, only
write problems result in LPL pages and the LPL list is kept in the database exception table
(DBET) in the Shared Communication Area.
GBP Recovery Pending (GRECP)
This indicates that the GBP failed, and the changes that are recorded in the log must be
applied to the page set. When a page set is placed in the GRECP status, DB2 sets the
starting point for the merge log scan to the Log Record Sequence Number (LSRN) of the
last complete GBP checkpoint.
After you have recovered the failed CF, move all the structures that normally reside there
back to that CF by using the system command:
SETXCF START,REBUILD,POPCF=Target CF
The recovery process for a duplexed structure is different and significantly faster. In that
case, DB2 simply reverts to simplex mode and continues processing using the surviving
structure instance.
18.6.4 GBP duplex structure recovery from a CF failure

To minimize the impact from a CF failure, structures can be duplexed. Duplexing will speed
up the recovery process because the secondary structure will already be allocated and will
contain updated data.
Note that a secondary structure is not an exact copy of the primary. Although it does contain
updated data, if the primary structure fails, there still has to be a rebuild phase to the
secondary structure.
387
18.7 SCA structure management and recovery

This section discusses the management and recovery of the list structure.
18.7.1 SCA list structure

There is one list structure per data sharing group used as the shared communications area
(SCA) for the members of the group. The SCA contains information about databases in an
exception condition and recovery information for each data sharing group member. This
information may include:
Logical Page List (LPL)

Database Exception Table (DBET)
Boot Strap Data Set (BSDS) - names of all members
Log data set - names of all members
LRSN delta
Copy Pending
Write Error ranges
Image copy data for certain system data spaces
This information is available to all members of the data sharing group. Other DB2s have the
ability to recover information from a failure during a group-wide restart if one of the DB2 is not
restarted. This is known as a Peer Restart of the Current Status Rebuild phase of DB2
restart.
The DB2 member performing the Peer Restart needs to know the log information about the
non-starting member. It finds this out by reading the SCA structure. The first connector to the
structure is responsible for building the structure if it does not exist.
The SCA structure supports REBUILDPERCENT and automatic rebuild, if the SFM policy is
defined with CONNFAIL(YES) specified. Manual rebuild is supported using the SETXCF
START,REBUILD command without stopping the data sharing group members.
18.7.2 Allocating the SCA structure

DB2 rebuilds the SCA structure from information in the bootstrap data set BSDS, and also
from information in the storage of any connected DB2s. If you lose the SCA and one of the
DB2s, the structure cannot be rebuilt so you must perform a group restart. To avoid a group
restart for this type of condition, you can exploit System Managed CF Duplexing for the SCA
structure.
When DB2 starts, the SCA structure called Group attachment name_SCA gets built by the
first connector to allocate it, if it does not already exist. The preference list defined in the
active CFRM policy determines which CF is chosen.
18.7.3 Removing the SCA structure

After stopping all DB2 members of a data sharing group normally, all connections to the SCA
structure will be terminated. The structure itself, however, will not be deleted because its
disposition is KEEP.
388
Note: We do not recommend deleting the SCA structure (even though it is possible to do
so). If there is a catastrophic failure and all DB2s in the data sharing group come down
without resource cleanup, then the recovery information needed to perform a peer
recovery is still available. When a DB2 restarts, it will have the names of the logs of the
other DB2s, what the current location in the logs are, where the oldest unit of recovery is,
and other recovery information in the SCA.
To remove the structure from a Coupling Facility to perform maintenance on it, use the SETXCF
START,REBUILD command to move it to the other CF. Although not recommended, it is
possible to remove the SCA structure using the command:
SETXCF FORCE,STR,STRNAME=Group attachment name_SCA
18.7.4 Altering the size of a DB2 SCA

There are two methods of altering the size of the SCA, static or dynamic. The static method
entails changing the size of the structure in the CFRM policy and then rebuilding using SETXCF
START,REBUILD command. This could be used if:
The currently allocated size of the structure has reached the maximum size defined in the
SIZE parameter of the CFRM policy.
The initial size of a structure is not correct and always has to be adjusted. In this case, a
static change to INITSIZE would be required.
The dynamic method uses the SETXCF START,ALTER command or the autoalter function
defined in the CFRM to change the size of the SCA, but only if the currently allocated size of
the structure is less than the maximum size as defined in the SIZE parameter of the CFRM
policy. An SCA that is too small may cause DB2 to crash.
18.7.5 Moving the SCA structure

It may be necessary to move a structure from one CF to another for workload rebalancing, or
to move all the structures out of a CF for maintenance purposes. This can be done
dynamically.
During the move, all activity to the SCA structure is temporarily stopped while a new structure
is allocated on the target CF and the contents of the old structure are copied to the new one.
The connections are established to the new structure, and finally the old structure is
deallocated. This is all handled by the system after you initiate the SETXCF START,REBUILD
command.
Note: The rebuild of the SCA structure is dynamic and nondisruptive to DB2.
If the current size of the SCA structure is different from the INITSIZE (the structure size was
changed using the SETXCF ALTER command), the rebuild will attempt to allocate the structure
in the new CF with the same size that was in the previous CF.
18.7.6 SCA over threshold condition

When an unusually large amount of exception type or recovery-related conditions occur, the
SCA structure may need to be increased in size. If the SCA structure size is not increased in
a timely fashion, it may cause the DB2 subsystems to crash. Structure monitoring will
389
produce highlighted warning messages on the MCS console when the structure reaches its
threshold The default threshold is 80% full for the SCA structure. If the SCA is too small, DB2
may crash.
18.7.7 SCA recovery from a CF failure

In the event of a single CF failure, if a dual CF is used, a dynamic rebuild of the SCA structure
would be initiated and it will be possible to recover from the failure. If a single CF were used,
then the DB2 members using the SCA structure would come down.
To minimize the impact of CF failures:
Have an active SFM policy.

Use dual Coupling Facilities.
Have enough space on each CF to back up the structures on the other CF.
Use duplexing for the GBPs, SCA, and Lock structures.
Use dual paths to help prevent connectivity failures.
Even though the DB2 system remains operational, processes that require access to the SCA
structures are queued while DB2 is rebuilding the SCA. This is the same for Lock and GBP
structures.
18.7.8 SCA recovery from a system failure

In the event of a DB2 subsystem or system failure, the SCA structure will remain active and
usable to the surviving DB2 members. See 18.9.6, DB2 restart with Restart Light on
page 393 for more detailed information about this topic.
18.8 How DB2 and IRLM use the CF for locking

IRLM is shipped with DB2. Each DB2 subsystem must have its own instance of IRLM. Note
that IRLM has its own version and release numbering system, which does not correspond
one-to-one to DB2 version and release numbering.
You cannot share IRLM between DB2s or between DB2 and IMS. If you are running a DB2
data sharing group, there is a corresponding IRLM group.
IRLM works with DB2 to serialize access to the data. DB2 requests locks from IRLM to
ensure data integrity when applications, utilities, commands, and so on attempt to access the
same data. There is one IRLM instance per system on which a DB2 member resides.
IRLM will have entries in storage, as well as duplicate entries in the structure. There is only
one IRLM lock structure per data sharing group.
In a data sharing environment, DB2 uses global locks so each member knows about the other
members' locks. Local locks still exist but they are only for use by a single member for data
that is not being shared.
There are two types of global Locks:
Physical locks (P-locks)
The mechanism that DB2 uses to track Inter-DB2 Read/Write interest in an object is a
global lock called a physical lock. P-Locks are:
Issued for every table open.
390
Initiated by DB2, not by transactions.

Negotiable.
Page P-Locks are used to preserve integrity when two systems are updating the same
page as with Row Level Locking. They are also used on the Space Map Page and Index
leaf pages.
Logical locks (L-locks)
Logical locks are also known as transaction locks. They are used to serialize access to
data and are owned by the transaction. L-locks are controlled by each members' IRLM.
P-locks and L-locks work independently of each other, although information about them is
stored in both the lock table and the Modified Retained List (MRL) parts of the lock structure.
The CF lock structure contains lock information that all group members need to share. The
lock structure is comprised of two independent parts:
Lock table entry (LTE)
The LTE keeps track of global lock contentions. This is where each DB2 member of a data
group registers their interest in a resource (for example, table spaces, an index, a
partition.) It is often referred to as the hash table.
IRLM
IRLM uses a hashing algorithm to assign lock entries in the lock table to a resource
requested by more than one member. These lock table entries are also known as hash
classes.
If two different resources hash to the same lock entry, it is referred to as false contention.
False contention causes overhead because IRLM must determine whether the contention
is real or false. Extra XCF signalling between the systems is a consequence of both real
and false contention. A lock table must be of sufficient size to minimize false contention.
18.9 Using DB2 lock structures

Before starting DB2 for data sharing, you must have defined one lock structure in the CFRM
policy. This policy determines how and where the structure resources are allocated.
18.9.1 Deallocating DB2 lock structures

When all the DB2 members of a data sharing group terminate normally by DB2 command, all
connections to the lock structure will end normally. The connections will be broken normally
but the last DB2 system in the group to close down will stay in FAILED-PERSISTENT status.
This is a completely normal circumstance (the connection is waiting to be reestablished.) The
restart of any group member will clear that status when it can establish a new connection with
the lock structure. This is called current status rebuild Peer Restart.
It is possible that one of the members of the DB2 data sharing group has abended. In this
case, the connection would also be in FAILED-PERSISTENT status.
18.9.2 Altering the size of a DB2 lock structure

The size of the DB2 Lock structure may need to be changed due to increased use of DB2,
resulting in a requirement for a larger structure. Conversely, possibly the original structure
was oversized and needs to be decreased.
391
When you add new applications or workload to your DB2 data sharing group, you must take
into account the size of your lock structure because the number of locks may increase. If the
lock structure is too small, IRLM warns you that it is running out of space by issuing the
DXR170I message when storage begins to fill.
It is extremely important to avoid having the structure reach 100% full. This condition can
cause DB2 locks to fail and lead to severe performance degradation due to false contention.
DB2 will continue, but transactions might begin failing with resource unavailable errors and
SQL CODE -904. Use of the lock structure should be monitored by using a performance
monitor such as the RMF Structure Activity Report.
Perform the following steps to alter the size of a DB2 lock structure using the appropriate
z/OS system commands.
D XCF,STR,STRNAME=D#$#_LOCK1
SETXCF START,ALTER,STRNM=D#$#_LOCK1,SIZE=new size
D XCF,STR,STRNAME=D#$#_LOCK1
18.9.3 Moving DB2 lock structures

It may become necessary to move a structure from one CF to another to perform
maintenance or some other type of reconfiguration. Before proceeding, verify that enough
free space is available in the target location by issuing the system command:
D CF,CFNAME=Target CF name
You can then dynamically move the structure by issuing the system command:
SETXCF START,REBUILD,STRNAME=D#$#_LOCK1,LOC=OTHER
This command rebuilds the structure in an alternate CF in accordance with the current
preference list in the CFRM policy.
18.9.4 DB2 lock structures and a CF failure

In the event of a CF failure, the lock structure will be automatically rebuilt to another CF. The
rebuild takes place whether System Failure Management (SFM) is active or not, because the
failure is classified as a structure failure. When rebuild occurs, the information used to
recover the lock structure is contained in DB2's virtual storage (not in the logs.)
If the rebuild of the lock structure fails, all DB2 members in the group terminate abnormally on
a S04F abend with a 00E30105 reason code. DB2 must then perform a group restart.
The rebuild of the lock structure may fail because:
There is no alternate CF specified in the CFRM policy preference list.
There is not enough storage in the alternate CF.
392
A group restart is distinguished from a normal restart by the act of recovering from the logs of
all members that were lost from the lock structure. A group restart does not necessarily mean
that all DB2s in the group start up again, but information from all non-starting DB2s must be
used to rebuild the lock structure.
18.9.5 Recovering from a system failure

When a member of the data sharing group fails, any modified global locks held by that DB2
become retained locks. This can affect the availability of data to other members of the group.
At DB2 (or z/OS system failure) abend time, the information from the modify resource list is
used to create retained locks. The information about retained locks in the lock structure is
stored until they are released during the DB2 restart. These retained locks are held because
DB2 was using the data when z/OS or DB2 failed and the data may require to be updated or
rolled back depending on whether a unit of work (UOW) completed.
The DB2 log has to be processed by the failed DB2 system performing a system recovery to
determine whether a UOW should be completed or rolled back. If the RETLWAIT (the
retained lock timeout field) parameter in DSNZPARM is zero (0), then the lock request will be
immediately rejected and a resource unavailable condition will be returned to the
application. If the RETLWAIT is non-zero, then DB2 will wait for the retained lock to become
available.
To keep data available for all members of the group, you must restart all of your failed
members as quickly as possible, either on the same z/OS system or on another z/OS system.
The purpose of the DB2 restart is to clear the retained locks in the CF. You may also have to
restart CICS or IMS to clear in-doubt locks. After the retained locks have been cleared, you
may take DB2 down again.
18.9.6 DB2 restart with Restart Light

There is a special type of DB2 restart designed for the system recovery situation called
Restart Light. To recover the retained locks held by a member, use the LIGHT(YES) clause of
the START DB2 command to restart the member in light mode.
Restart Light allows a DB2 data sharing member to restart with a minimal storage footprint,
and then to terminate normally after DB2 frees retained locks. By reducing storage
requirements, restart for recovery may be possible for more resource-constrained systems.
Restart Light mode does the following:
Minimizes the overall storage required to restart the member.
Removes retained locks as soon as possible, except for the following:
Locks that are held by postponed abort units of recovery.
IX mode page set P-locks. These locks do not block access by other members;
however, they do block drainers, such as utilities.
Terminates the member normally after forward and backward recovery is complete. No
new work is accepted.
In DB2 V8 and above, if in-doubt units of recovery (URs) exist at the end of restart
recovery, DB2 will remain running so that the in-doubt URs can be resolved. After all the
in-doubt URs have been resolved, the DB2 member that is running in LIGHT(YES) mode
will shut down and can be restarted normally.
393
Note that a data sharing group started with the Light option is not registered with the
Automatic Resource Manager (ARM). Therefore, ARM will not automatically restart a
member that has been started with LIGHT(YES).
You are able to restart any DB2 on any system in the DB2 data-sharing group if the command
prefix scope is already specified as S or X in the system definition statements of the
IEFSSNxx member using the DB2 command START DB2,LIGHT(YES).
This same DB2 can later be recovered back on the original failing system as soon as it is
available.
18.10 Automatic Restart Manager

Automatic Restart Manager (ARM) provides a quick restart capability without any operator
action. Locks held by the failed members, called retained locks, will be released as soon as
the member is restarted. Releasing retained locks quickly is very important to providing high
data availability to applications running on other members, while maintaining data integrity.
ARM can rebuild CICS regions associated with the DB2 to resolve in-doubt units of work.
To have DB2 restarted in Light mode (Restart Light), you must have an ARM policy for the
DB2 group that specifies LIGHT(YES) within the RESTART_METHOD(SYSTERM) keyword for the
DB2 element name. For example:
RESTART_METHOD(SYSTERM,STC,'cmdprfx STA DB2,LIGHT(YES)')
For more information about ARM, refer to Chapter 6, Automatic Restart Manager on
page 83.
18.11 Entering DB2 commands in a sysplex

This section describes routing commands and command scope.
Routing commands
You can control operations on an individual member of a data sharing group from any z/OS
console by entering commands prefixed with the appropriate command prefix.
For example, assuming that you chose -D#$1 as the command prefix for member D#$1, you
can start a DB2 statistics trace on that member by entering this command at any z/OS
console in the Parallel Sysplex:
-D#$1 START TRACE (STAT)
Command routing requires that the command prefix scope is registered as S or X on the
IEFSSNxx PARMLIB member. You can also control operations on certain objects by using
commands or command options that affect an entire group. These can also be entered from
any z/OS console.
For example, assuming that D#$1 is active, you can start database XYZ by entering this
command at any z/OS console in the Parallel Sysplex:
-D#$1 START DATABASE (XYZ)
394
Command scope
The breadth of a command's impact is called the scope of that command.
Many commands that are used in a data sharing environment affect only the member for
which they are issued. For example, a STOP DB2 command stops only the member identified
by the command prefix. Such commands have member scope.
Other commands have group scope because they affect an object in such a way that all
members of the group are affected. For example, a STOP DATABASE command, issued from
any member of the group, stops that database for all members of the group.
395
396
19
Chapter 19.
IMS operational considerations

Information Management System (IMS) is the IBM transaction and hierarchical database
management system. This chapter provides an overview of operational considerations when
IMS is used in a Parallel Sysplex.
IMS is an exploiter of Parallel Sysplex. When IMS sysplex data sharing is implemented,
operators need the following knowledge:
A basic understanding of IMS
How IMS uses Parallel Sysplex facilities
The role of the various IMS components when running within a Parallel Sysplex
The function of the various IMS structures in the Coupling Facility
IMS support of Automatic Restart Manager (ARM)
Familiarity with the processes for starting and stopping IMS within a Parallel Sysplex
IMS recovery procedures
The implication of entering commands on certain consoles and systems in the sysplex
397
19.1 Introduction to Information Management System

Information Management System (IMS) is both a transaction manager and a hierarchical
database management system. It is designed to provide an environment for applications that
require very high levels of performance, throughput, and availability. IMS consists of two
major components:
IMS Database Manager (IMS DB)
IMS Transaction Manager (IMS TM)
Each component can be configured to run together or independently, depending on the client
requirement.
19.1.1 IMS Database Manager

IMS DB is a database management system that helps you manage your business data with
program independence and device independence. IMS DB provides all the data integrity,
consistency, and recoverability for an environment with many databases, both large and
small, and with many concurrent updaters.
Unlike DB2, which uses relational tables to manage its data, IMS uses a hierarchical data
implementation to manage the IMS data.
IMS DB can also be used as a database manager for CICS transactions when CICS is used
as the transaction manager.
19.1.2 IMS Transaction Manager

IMS TM is a message-based transaction manager. It provides services to process messages
received from the network (input messages) and messages created by application programs
(output messages). It also provides an underlying queuing mechanism for handling these
messages.
IMS TM can use either IMS DB or DB2 as a database manager.
19.1.3 Common IMS configurations

There are three major configurations available to IMS:
Database Control (DBCTL)
Data Communications Control (DCCTL)
IMS DB/DC
Note: In each of these scenarios, a complete set of IMS regions is required, depending on
the functions needed. The IMS DB or IMS TM boxes illustrated in the following figures are
indicative only of what functionality has been implemented. Where both IMS DB and IMS
TM are listed, only a single set of IMS address spaces is required.
DBCTL
DBCTL is an environment where only IMS DB is implemented, and CICS is used as the only
transaction manager. In this model, the application may also access DB2 data. An example of
this type of configuration is shown in Figure 19-1 on page 399.
398
CICSterminal
CICS
DB2
IMS DB
RECON
DB2
table
FAST PATH
DATABASE
IMS logs
(OLDS & WADS)
FULL
FUNCTION
DATABASE
Figure 19-1 IMS DBCTL configuration
DCCTL
DCCTL is an environment where only IMS TM is implemented and DB2 is used as the only
database manager.
Note: The DC in DCCTL refers to Data Communications. There may still be IMS
documentation that refers to IMS/DC, but this has been replaced by IMS/TM.
An example of this type of configuration is shown in Figure 19-2 on page 400.
Chapter 19. IMS operational considerations in a Parallel Sysplex
399
CICSterminal
IMSTerminal
CICS
IMS TM
Message
Queues
DB2
RECON
IMS logs
(OLDS & WADS)
DB2
table
Figure 19-2 IMS DCCTL configuration
IMS DB/DC
IMS DB/DC is an environment where both IMS DB and IMS TM are implemented. Here, IMS
can process transactions submitted by users logged on to terminals connected to IMS, and
trigger application programs running in IMS that access IMS databases.
In this model, CICS can coexist as another transaction manager and also access the IMS
data, as per the DBCTL model. This configuration also supports access to DB2 databases by
IMS applications.
An example of this type of configuration is shown in Figure 19-3 on page 401.
400
CICSterminal
IMSterminal
CICS
IMS TM
Message
Queues
DB2
IMS DB
RECON
DB2
table
FAST PATH
DATABASE
IMS logs
(OLDS & WADS)
FULL
FUNCTION
DATABASE
Figure 19-3 IMS DB/DC IMS configuration
19.1.4 Support of IMS systems

Most components of the IMS environment are managed by IMS systems programmers,
together with the z/OS and network systems programmers. However, most of the planning for
the database-related functions would be the responsibility of the IMS systems programmers
and the IMS database administrators.
19.1.5 IMS database sharing

To explain the different components of IMS and how they work in relation to each other, there
are four IMS data sharing configurations to discuss. They are:
IMS single system, where the IMS databases are not being shared with any other IMS
system.
Local IMS Data Sharing, where multiple IMS subsystems running in the same z/OS image
and sharing a single IRLM are sharing a set of databases.
Global IMS Data Sharing, where multiple IMS subsystems running on more than one z/OS
image are sharing a set of databases.
IMS Data Sharing with Shared Queues is a logical extension of IMS Local or Global data
sharing, and allows multiple IMS subsystems to share the transaction queue as well as the
databases.
401
In addition, there are several communication components of IMS that can be incorporated
into an IMSplex environment. They are:
VTAM Generic Resources
Rapid Network Reconnect (RNR), which comes in two varieties:
SNPS: Single Node Persistent Sessions
MNPS: Multi Node Persistent Sessions
For a brief discussion of these components, refer to 19.4.2, VTAM Generic Resources on
page 411, and 19.4.3, Rapid Network Reconnect on page 412.
19.2 IMS system components

Before conceptualizing IMS in a sysplex, it is helpful to understand basic IMS address spaces
and important data sets that exist in all IMS environments. Figure 19-4 shows a diagram of a
simple IMS system (one that is not exploiting any sysplex functions).
In this case, IMS shares the data between the different applications, but only within the same
IMS system. When IMS runs in this mode (that is, it is not sharing its databases with any
other IMS systems), data is serialized at the segment level and serialization information is
kept in the IMS address space. This is known as Program Isolation (PI) mode.
IMS Control Region
WADS
OLDS
DBRC
DLISAS
MPP
IFP
JMP
BMP
JBP
RECON
FULL
FUNCTION
DATABASE
FAST PATH
DATABASE
MESSAGE
QUEUES
Figure 19-4 IMS structure - simple system
The following sections briefly describe the various components of IMS that are shown in
Figure 19-4.
IMS control region (IMSCTL)

The IMS control region (IMSCTL) provides the central point for an IMS subsystem (an IMS
subsystem is the set of all the related address spaces that provide the IMS service). It
provides the interface to all network services for Transaction Manager functions and the
402
interface to z/OS for controlling the operation of the IMS subsystem. It also controls and
dispatches the application programs running in the various dependent regions.
The control region provides all logging, restart, and recovery functions for the IMS
subsystems. The terminals, message queues, and logs are all attached to this region. If Fast
Path is used, the Fast Path database data sets are also allocated by the control region
address space.
DLISAS (IMSDLI)
The DLI Separate Address Space (DLISAS) has all the full function IMS database data sets
allocated to it, and it handles most of the data set access functions. It contains some of the
control blocks associated with database access and the database buffers used for accessing
the full function databases. Although it is not required to use the DLISAS address space, its
use is recommended. If you specify that you wish to use DLISAS, this address space is
automatically started when IMS is started.
DBRC (IMSDBRC)
The DataBase Recovery and Control (DBRC) address space contains the code for the DBRC
component of IMS. It processes all access to the DBRC recovery control data sets (RECON).
It also performs all generation of batch jobs for DBRC; for example, for archiving the online
IMS log. All IMS control regions have a corresponding DBRC address space because it is
needed, at a minimum, for managing the IMS logs. This address space is automatically
started when IMS is started.
Message Processing Regions (MPR)

An IMS Message Processing Program (MPP) runs in a Message Processing Region (MPR)
and is used to run applications that process messages input to the IMS Transaction Manager
component (that is, online programs). The MPRs are usually started by issuing the IMS
command /STA REG xxx.
Integrated Fast Path (IFP) regions

Integrated Fast Path regions also run application programs to process messages for
transactions, but in this case it is transactions that have been defined as Fast Path
transactions. The applications are broadly similar to the programs that run in an MPR. Like
MPRs, the IFP regions are started by the IMS control region as a result of an IMS command.
The difference with IFP regions is in the way IMS loads and dispatches the application
program, and handles the transaction messages.
Batch Message Processing (BMP) region

Unlike the other types of application-dependent regions, Batch Message Processing regions
are not started by the IMS control region, but rather by submitting a batch job. The batch job
then connects to an IMS control region identified in the execution parameters. BMPs do not
normally process online transactions, but are designed for larger bulk processing of data.
Java Message Processing (JMP) region

A Java Message Processing region is similar to an MPP, except that it is used for Java
programs.
Java Batch Processing (JBP) regions

A Java Batch Processing region is similar to a BMP, except that it is used for batch Java
programs.
403
RECON data sets

The Recovery Control (RECON) data sets are a set of three VSAM files used by DBRC to
hold all the IMS system and database recovery information. Two of these are used at any one
time, with the third one available as a spare. For more information about this topic, refer to
IMS Database Recovery Control (DBRC) Guide and Reference Version 9, SC18-7818.
OLDS data sets

All IMS log records, database update information, and other system-related information are
written to the Online Log Data Sets (OLDS) to enable any database or IMS system recovery.
In addition, the OLDS can be post-processed for debugging or accounting purposes.
WADS data sets

The Write Ahead Data Sets (WADS) are also used for logging, and are designed for
extremely fast writes using a very small blocksize of 2 K or 4 K. WADS are used for events
that cannot wait for partially filled OLDS buffers to be written out. For example, when an IMS
transaction completes, the log data must be externalized to ensure any updates are
recoverable. If the OLDS buffer is only partially full, it is written to the WADS to avoid writing
partial OLDS buffers to the OLDS data set.
On high volume systems, these data sets are critical to performance.
Message queues
All messages and transactions that come into IMS are placed on the IMS message queue,
and are then scheduled to be processed by an online dependant region (for example, an
MPR). In a non-sysplex environment, the message queue is actually kept in storage buffers in
the IMS control region. In a sysplex environment, you have the option to place the message
queue in the Coupling Facility. This is known as IMS Shared Queues.
19.2.1 Terminology
This section defines terminology used later in this chapter.
IMSplex
An IMSplex is one or more IMS subsystems that work together as a unit. Typically (but not
always), these address spaces:
Share either databases or resources or message queues (or a combination of these)
Run in a z/OS Parallel Sysplex environment
Include an IMS Common Service Layer
The address spaces that can participate in the IMSplex are:
Control region address spaces
IMS manager address spaces (Operations Manager, Resource Manager, Structured Call
Interface)
IMS server address spaces (Common Queue Server (CQS))
An IMSplex allows you to manage multiple IMS systems as though they were one system (a
single-system perspective). An IMSplex can exist in a non-sysplex environment, or it can
consist of multiple IMS subsystems (in data or queue sharing groups) in a sysplex
environment.
404
IMS data sharing

It is possible for any IMS control region or batch job running in a z/OS system to share access
to a set of IMS databases. This requires the use of a separate feature, the Internal Resource
Lock Manager (IRLM), to manage the IMS locks on the database (instead of using Program
Isolation (PI), as would be used in a single-IMS-subsystem environment).
This type of database sharing is also known as block-level data sharing. In block-level data
sharing, IMS locks the databases for the application at the block level. By comparison, PI
mode locking is done at the segment level; a block will typically contain a number of
segments. Because of this coarser level of locking, there is an increased risk of deadlocks
and contention between tasks for database records.
Shared queues
IMS provides the option for multiple IMS systems in a sysplex to share a single set of
message queues. The set of systems that share the message queue is known as an IMS
Queue Sharing Group.
Common Service Layer

The IMS Common Service Layer (CSL) is a collection of IMS manager address spaces that
provide the infrastructure needed for IMS systems management tasks. The CSL address
spaces include the Operations Manager (OM), the Resource Manager (RM), and the
Structured Call Interface (SCI).
Full Function databases

Full Function databases (otherwise known as IMS databases, DL/I databases, or DL/1
databases) provide a hierarchically-structured database that can be accessed directly,
sequentially, or by any other predefined method based on a predefined index.
Traditionally, these data bases were limited to 4 GB or 8 GB in size, but they can now be
made much larger by exploiting the High Availability Large Databases (HALDB) function
available since IMS Version 7.
These physical databases are based on two different access methods, VSAM or OSAM.
VSAM databases
Virtual Sequential Access Method (VSAM) is used by many IMS and non-IMS applications,
and comes in two varieties:
Entry Sequenced Data Sets (ESDS) for the primary data sets
Key Sequenced Data Sets (KSDS) for index databases
These data sets are defined using the IDCAMS utility program.
OSAM databases
The Overflow Sequential Access Method (OSAM) is unique to IMS. It is delivered as part of
the IMS product. It consists of a series of channel programs that IMS executes to use the
standard operating system channel I/O interface. The data sets are defined using JCL
statements. As far as the operating system is concerned, an OSAM data set looks like a
physical sequential data set (DSORG=PS).
405
Fast Path databases

Fast Path databases were originally available only as part of a separately priced, optional
feature of IMS. This resulted in the documentation and code being separate from that for the
Full Function (FF) databases. There are two types of Fast Path databases:
Data Entry Databases (DEDBs)
Main Storage Databases (MSDBs)
DEDBs
The Data Entry Database (DEDB) was designed to support particularly intensive IMS
database requirements, primarily in the banking industry, for larger databases, high
transaction workloads, improved availability, and reduced I/O.
MSDBs
The Fast Path database access method, Main Storage Database (MSDB), has functionality
that has been superseded by the Virtual Storage Option (VSO) of the DEDB, so it is not
described in this book, and you are advised not to use it.
19.3 Introduction to IMS in a sysplex

This section describes the components of IMS and how they make up an IMSplex.
19.3.1 Local IMS data sharing

As discussed in 19.1.5, IMS database sharing on page 401, IMS supports two ways to
share databases across IMS subsystems. Both of these are referred to as Block Level Data
Sharing, because the data is shared between the IMS systems at the block level. When two
or more IMS subsystems in the same z/OS system access the same database, this is known
as local IMS data sharing. It uses the IRLM address space to maintain tables of the locks
within IRLM.
An example of this can be found in Figure 19-5 on page 407, which shows a two-way IMSplex
running within a single system sharing the IRLM address space as well as the RECON and
databases. Note that each system still has its own message queues.
406
IRLM
IMS Control Region 1
RECON
DATABASE
MESSAGE
QUEUES
MESSAGE
QUEUES
Figure 19-5 Simple IMS 2-way local data sharing in a single z/OS system
The following additional address spaces are required in this scenario:
IRLM
Internal Resource Lock Manager (IRLM) is required by IMS when running block-level data
sharing. It is used to externalize all the database locks to enable data sharing. When the IMS
subsystems are all within the same z/OS system, then IRLM maintains the database locks
within its own address space.
IRLM was originally known as the IMS Resource Lock Manager, and you may find it referred
to by this name in older publications. It is now also used by DB2.
Important: IRLM is provided and shipped along with IMS as well as with DB2, but you
cannot share IRLM between IMS and DB2. Ensure that the IRLM instance running to
support IMS is managed along with the IMS product, and that the IRLM instance running to
support DB2 is managed along with the DB2 product.
19.3.2 Global IMS data sharing

Global IMS data sharing is simply the extension of local IMS data sharing, where a number of
IMS systems are connected in a sysplex, but running in different z/OS systems.
With the databases and RECONs on DASD shared by the sysplex, it is possible for IMS
control regions and batch jobs to run on any of these z/OS images and share access to the
databases. To do this, an IRLM address space must be running on each z/OS image that the
IMS address spaces are running on. The IRLMs perform the locking as in the previous case;
however, instead of holding details of the locks in the IRLM address space, the lock tables
and IMS buffers are stored in shared structures in the Coupling Facility.
407
An example of this can be found in Figure 19-6, which shows a two-way IMSplex across two
systems. This can be extended to many more IMS systems across many more systems (up to
32 in total), all using the same shared RECONs, databases, and Coupling Facility structures.
Note that in this scenario, the IMS message queues are still unique to each IMS system and
not shared.
IRLM
IRLM
COUPLING
FACILITY
Lock Structures
DB Structures
System 2
System 1
RECON
MESSAGE
QUEUES
MESSAGE
QUEUES
DATABASE
Figure 19-6 Simple two-way global IMS data sharing environment across two systems
The following additional or changed items are required in this scenario:
IRLM
(Internal Resource Lock Manager (IRLM) is required by IMS when running in BLDS or data
sharing mode, and is still used to externalize all the database locks to enable data sharing, as
explained in the preceding example.
In this case, because the IMS systems are running on different systems, each system
requires an IRLM address space to be active and IRLM uses the Coupling Facility to store
and share its lock information and database buffers.
Coupling Facility structures

To utilize the features of the Parallel Sysplex, IMS stores data in many different types of
Coupling Facility structures. For a list of the IMS structures held in the Coupling Facility, refer
to 19.6, IMS structures on page 413.
For additional information about the Coupling Facility in general, refer to Chapter 7, Coupling
Facility considerations in a Parallel Sysplex on page 101.
408
19.3.3 Global IMS data sharing with shared queues

In addition to sharing the IMS databases, IMS provides the facility for multiple IMS systems in
a sysplex to share a single set of message queues. This function is known as IMS shared
queues.
Instead of the messages being held within buffers in IMS storage, backed up by Message
Queue data sets, the messages are held in structures in a Coupling Facility. All the IMS
subsystems in the sysplex can share a common set of queues for all non-command
messages (that is, input, output, message switch, and Fast Path messages). A message that
is placed on a shared queue can be processed by any of the IMS subsystems in the shared
queues group as long as the IMS has the resources to process the message.
Using shared queues provides the following benefits:
Automatic workload balancing across all IMS subsystems in a sysplex
Increased availability to avoid both scheduled and unscheduled outages
Figure 19-7 shows both sysplex data sharing and shared queues in a simple two-way IMS
sysplex. This can also be extended to many more IMS systems across many more systems.
The ideal configuration is to have every IMS subsystem able to run any transaction. This
makes adding or removing a subsystem to the IMSplex a relatively simple and transparent
process.
We refer to this scenario in the rest of this chapter.
CQS
CQS
SCI
SCI
COUPLING
FACILITY
IRLM
IRLM
MsgQStructures
Lock Structures
DB Structures
Resource Struc.
SCI
SCI
SCI
SCI

SCI
COMMUNICATIONS
System 1
System 2
SCI
SCI
SCI
SCI
RECON
OM
SCI
RM
SCI
SCI
FDBR1
FDBR2
DATABASE
SCI
OM
SCI
RM
SCI
Figure 19-7 IMS database data sharing with shared queues
The additional address spaces and structures not previously described are listed here:
409
Common Queue Server

The Common Queue Server (CQS) is a generalized server that manages data objects in a
Coupling Facility on behalf of multiple clients. It is used by:
IMS Shared Queues, to provide access to the IMS shared queue structures, which replace
the IMS messages queues in the IMS control region storage and the message queue data
sets on DASD in a non-shared queues environment
The Resource Manager address space, to access the resource manager structure
If using shared queues, this address space is automatically started when IMS is started.
To shut down IMS in this environment, it is recommended that you enter /CHECKPOINT FREEZE.
CQS may automatically shutdown, based on the status of the /CQS command, as mentioned
in 19.11.4, CQS shutdown on page 452. If the DUMPQ or PURGE option is used, IMS does not
dump the message queues during shutdown.
Warning: If CQS abnormally ends, it needs to be restarted as soon as possible to avoid
having to cold start IMS if two structure checkpoints should occur before this is done.
CQS automatically registers with the Automatic Restart Manager; see Chapter 6, Automatic
Restart Manager on page 83 for more information. It does not have to be manually restarted
after a CQS failure. CQS does not have to be restarted after an IMS failure because CQS and
IMS are known as separate subsystems to ARM. If CQS is not available, IMS will not shut
down using /CHECKPOINT FREEZE but only with the z/OS MODIFY command.
Recovery of the IMS shared queue structures is handled by the master CQS. This involves
repopulating the structure using the CQS Structure Recovery Data sets (SRDS), and then
applying updates using the message queue log stream. In the case of a structure not
recovering, then forcing the structure may be required. If this is done, it is similar to an IMS
cold start and the IMS system programmers should be involved.
More information can be found in the IMS manuals and other IBM Redbooks publications
about the IMS sysplex.
Structured Call Interface

The Structured Call Interface (SCI) address space provides for standardized intra-IMSplex
communications between members of an IMSplex. It also provides security authorization for
IMSplex membership, and SCI services to registered members.
The structured call interface services are used by SCI clients to register and deregister as
members of the IMSplex and to communicate with other members.
When running in an IMSplex environment, one SCI address space is required on each z/OS
image with IMS sysplex members, and it needs to be started prior to IMS starting.
Resource Manager
The Resource Manager (RM) address space maintains global resource information for clients
using a Resource structure in the Coupling Facility
It can contain IMSplex global and local member information, resource names and types,
terminal and user status, global process status, and resource management services. It also
handles sysplex terminal management and global online change.
One or more RM address spaces are required per IMSplex in IMS V8. IMS V9 allows for zero
RMs in the IMSplex.
410
Operations Manager
The Operations Manager (OM) address space provides an API allowing single point of
command entry into IMSplex. It will become the focal point for operations management and
automation. Command responses from multiple IMS systems are consolidated.
One or more OM address spaces are required per IMSplex.
Fast Database Recovery

Fast Database Recovery (FDBR) is an optional function of IMS. It is designed to quickly
release locks held by an IMS subsystem if that subsystem fails. It would normally run on an
alternate z/OS system within the sysplex, in case the system IMS is running on abends. For a
detailed description of Fast Database Recovery (which was made available along with IMS
Version 6), refer to IMS/ESA Version 6 Guide, SG24-2228.
19.4 IMS communication components of an IMSplex

The previous sections of this chapter refer to all the address spaces and components specific
to both data sharing and message queue sharing. However, there are several other
components of IMS that can also have an impact within an IMSplex, as discussed here.
19.4.1 IMS Connect

IMS Connect is used to communicate between TCP/IP clients and IMS.
Prior to IMS Version 9, IMS Connect was available as a separately-orderable product. It may
also be referred to by its original name, IMS TCP/IP OTMA Connector (ITOC). Since IMS
Version 9, IMS Connect is delivered as part of IMS.
Although IMS Connect does not impact how an IMSplex functions, and therefore was not
included in the previous figures, it has been included here because of its growing importance
in customer IMS configurations.
IMS Connect runs as a separate address space on one or more of the systems within the
sysplex. It listens for incoming TCP requests on predefined TCP/IP ports. Further information
about IMS Connect can be found in the publication IMS Connect Guide and Reference,
Version 9, SC18-9287.
19.4.2 VTAM Generic Resources

VTAM Generic Resources (VGR) is a service provided by VTAM that allows multiple
instances of a server application (such as IMS) to be accessed using a single VTAM resource
name. This minimizes the information that the user needs to know to logon to IMS. Each IMS
in the sharing group joins a Generic Resource Group, and the user simply logs on using the
VTAM application name for this sharing group. VTAM then selects which IMS in the group the
user will be logged on to, transparently to the user. In addition to making the process simpler
for the user, this also assists with workload balancing across the IMSplex and masks the
unavailability of a single IMS subsystem from the user.
The information about which members are in the Generic Resource Group, and their status,
is stored in a structure in the Coupling Facility. For more information, refer to Chapter 16,
Network considerations in a Parallel Sysplex on page 323.
411
19.4.3 Rapid Network Reconnect

Rapid Network Reconnect (RNR) is an optional function. It represents the IMS support for
VTAM persistent sessions. RNR can eliminate session cleanup and restart when an IMS
subsystem or a z/OS system failure occurs. There are two kinds of persistent session
support:
Single node persistent session (SNPS) provides support only for IMS failures. With SNPS,
the VTAM instance must not fail. Following an IMS failure and emergency restart, the
users session is automatically given to the restarted IMS. The user does not have to logon
again, but if signon is required, they will need to sign on again.
Multi node persistent session (MNPS) provides support for all types of host failures,
including IMS, z/OS, VTAM, or the processor. The session data is stored in a Coupling
Facility structure, so following any sort of host failure, when IMS is emergency restarted
the users session is automatically given to the restarted IMS as per an SNPS session.
19.5 IMS naming conventions used for this book

This section describes the environment we used to provide all the examples presented in this
book. For a high-level description of this environment, refer to 1.4, Parallel Sysplex test
configuration on page 10. The IMS systems used for these examples are running IMS
Version 9.1, and the IMSplex name for this IMSplex is I#$#.
There are three systems in this test sysplex and three IMS systems. Although each IMS can
run on any system, assuming that there is one IMS on each system, then the layout of IMS
address spaces would typically look as shown in Table 19-1.
Table 19-1 Address space names for the test IMS used for all examples
First system
Second system
Third system
System Name
#@$1
(known as System1)
#@$2
(known as System2)
#@$3
(known as System3)
IMS ID
I#$1
(known as IMS1)
I#$2
(known as IMS2)
I#$3
(known as IMS3)
IMS CTL
IMS control region
I#$1CTL
I#$2CTL
I#$3CTL
DLI SAS
(DLI Separate Address Space)
I$#1DLI
I#$2DLI
I#$3DLI
DBRC
(Database Recovery & Control)
I#$1DBRC
I#$2DBRC
I#$3DBRC
CQS
(Common Queue Server)
I#$1CQS
I#$2CQS
I#$3CQS
IRLM
(Internal Resource Lock Manager)
I#$#IRLM
I#$#IRLM
I#$#IRLM
SCI
(Structured Call Interface)
I#$#SCI
I#$#SCI
I#$#SCI
OM
(Operations Manager)
I#$#OM
I#$#OM
I#$#OM
RM
(Resource Manager)
I#$#RM
I#$#RM
I#$#RM
412
First system
Second system
Third system
IMS Connect
(TCP/IP Communication)
I#$1CON
I#$2CON
I#$3CON
FDBR
(Fast Database Recovery)
I#$3FDR
I#$1FDR
I#$2FDR
General parameter settings

The following parameters are used by the IMS plex:
GRNAME=I#$#XCF
GRSNAME=ITSOI#$#
IRLMNM=IR#I
This parameter specifies the name of the XCF group that will be
used by OTMA.
This parameter specifies the name of the VTAM Generic Resource
group that this IMS subsystem will use.
This parameter specifies the name of the z/OS subsystem that
IRLM will use.
IMS Connect definitions

Our IMS Connect is set up to listen on three TCP/IP ports, and it can communicate with any
of the available IMS systems in this sysplex. An example of the configuration file is shown in
Figure 19-8.
HWS
TCPIP
(ID=I#$1,RACF=N)
(HOSTNAME=TCPIP,PORTID=(7101,7102,7103),ECB=Y,MAXSOC=1000,
EXIT=(HWSCSLO0,HWSCSLO1))
DATASTORE (ID=I#$1,GROUP=I#$#XCF,MEMBER=I#$1TCP1,TMEMBER=I#$1OTMA)
IMSPLEX
(MEMBER=I#$1CON,TMEMBER=I#$#)
Figure 19-8 IMS Connect configuration for an IMSplex
VTAM Generic Resources

The VTAM Generic Resource name is defined by the GRSNAME parameter and is set to
ITSOI#$#.
Other connections
There are no DB2 systems, MQ systems, or CICS DBCTL systems connected to IMS, and
APPC is not enabled in this test environment.
19.6 IMS structures

This section describes the CF structures that may be used by IMS in a data sharing and
queue sharing environment. For more information about IMS structures, refer to these other
IBM Redbooks publications:
IMS in the Parallel Sysplex Volume I: Reviewing the IMSplex Technology, SG24-6908
IMS in the Parallel Sysplex Volume II: Planning the IMSplex, SG24-6928
IMS in the Parallel Sysplex Volume III: IMSplex Implementation and Operations,
SG24-6929
IMS/ESA Data Sharing in a Parallel Sysplex, SG24-4303
413
IMS/ESA Version 6 Shared Queues, SG24-5088

IMS/ESA Sysplex Data Sharing: An Implementation Case Study, SG24-4831
Coupling Facility structures

The Coupling Facility structures used by the IMS systems used for the examples in this book
are described in Table 19-2.
Table 19-2 IMS structures available
Description of the structure
Structure name in this

example
Type of structure
IRLM lock structure
I#$#LOCK1
LOCK structure
VSAM buffer structure
I#$#VSAM
CACHE structure
OSAM buffer structure
I#$#OSAM
CACHE structure
DEDB VSO structures
I#$#VSO1DB1
I#$#VSO1DB2
I#$#VSO2DB1
I#$#VSO2DB2
CACHE structures
IMS Shared Message Queue

structure
I#$#MSGQ
LIST structure
IMS Shared Message Queue

overflow structure
I#$#MSGQOFLW
LIST structure
IMS Shared Expedited

Message Handler structure
I#$#EMHQ
LIST structure
IMS Shared Expedited

Message Handler overflow
structure
I#$#EMHQOFLW
LIST structure
Resource structure
I#$#RM
LIST structure
IRLM lock structures

IMS uses the IRLM to manage locks in a block level data sharing environment. When running
in a sysplex data sharing configuration, the IRLM address space keeps a copy of the locks
that it is holding in a lock structure in the CF. From the perspective of the lock structure, the
connected user is IRLM, and the resource being protected is data in an IMS database (for
example, a record or a block).
VSAM and OSAM cache structures

Every time an IMS subsystem in a data sharing group accesses a piece of data in a shared
database, it informs the Coupling Facility about this. The information is stored in an OSAM or
VSAM cache structure, and allows the CF to keep track of who has an interest in a given
piece of data. Then, if one of the IMS subsystems in the data sharing group updates that
piece of data, the CF can inform all the other IMS systems that still have an in-storage copy
that their copy is outdated and that they should invalidate that buffer.
Data Entry Data Base Virtual Storage Option (DEDB VSO)

DEDB databases using VSO use what is known as a store-in cache structure to store data.
When an update is committed, the updated data is written to the cache structure. Some time
after that, the updates will then be written to DASD. During the intervening period, however,
414
the updated data is only available in the cache structure. If that structure is lost for some
reason, the updates must be recovered by reading back through the IMS logs.
Message Queue (MSGQ) structure

The Message Queue structure is a list structure that contains the IMS shared message
queues for full function IMS transactions.
Expedited Message Handler Queue (EMHQ) structure

The Expedited Message Handler Queue structure is a list structure that contains the IMS
expedited message handler queues for fast path IMS transactions.
With IMS Version 8 and earlier, this structure is required whether or not Fast Path is
implemented. From IMS version 9, this structure is optional (and is not used) if Fast Path is
not required.
MSGQ and EMHQ overflow structures

The shared queue overflow structures are both list structures that contain selected messages
from the shared queues when the MSGQ or EMHQ primary structures reach an
installation-specified overflow threshold. The overflow structures are optional.
Message Queue and EMHQ log streams

The Message Queue and EMHQ log streams are shared System Logger log streams that
contain all CQS log records from all CQSs in the shared queues group. This log stream is
important for recovery of shared queues, if necessary. Each shared queue structure pair has
an associated log stream.
Resource structure
The Resource structure is a CF list structure that contains information about uniquely-named
resources that are managed by the resource manager address space. This structures takes
on a greater role with the introduction of Dynamic Resource Definition in IMS Version 10.
Checkpoint data sets

A local data set that contains CQS system checkpoint information. There is one set of these
data sets per CQS address space.
Structure Recovery Data Sets

The Structure Recovery Data Sets (SRDS) are data sets that contain structure checkpoint
information for shared queues on a structure pair. All CQS address spaces in the queue
sharing group share the one set of SRDS data sets. Each structure pair has two associated
SRDSs.
19.6.1 IMS structure duplexing

As you can see, the CF structures contain information that is critical to the successful
functioning of the IMSplex. To ensure that IMS can deliver the expected levels of availability,
therefore, it is vital that all the structure contents can be quickly recovered in case of a failure
of the CF containing those structures.
One way to address this requirement is to maintain duplex copies of the structures, so that if
one CF fails, all the structure contents are immediately available in the other CF. In this case,
the system automatically swaps to the duplexed structure with minimal interruption. There are
different types of structure duplexing available to IMS for its various structures.
415
One option is to tell the operating system (actually the XES component of the operating
system) that you want the structure to be duplexed by the system. In this case, XES creates
two copies of the selected structures (with identical names) and ensures that all updates to
the primary structure are also applied to the secondary one.
To enable this function for a structure, you must add the DUPLEX keyword to the structure
definition in the CFRM policy. For more information about system-managed duplexing, refer
to 7.4.4, Enabling system-managed CF structure duplexing on page 115.
Another option provided by XES is what is known as user-managed duplexing. However, IMS
does not support user-managed duplexing for its CF structures.
IMS duplexing for VSO structures

The third possibility is where the structure owner (IMS, for example) maintains two structures,
with unique names, and has responsibility for keeping them synchronized and for failover
following a failure. This option can be used for VSO DEDB AREAs, and is independent of any
system-manages or user-managed duplexing.
When you define the DEDB AREA to DBRC, you specify the name (or names) of the structure
(or structures) to be used for that area. The first structure used for an AREA is defined by the
CFSTR1 option on the INIT.DBDS command. The second structure is defined by the CFSTR2
option. Both structures must be previously defined in the CFRM policy.
19.6.2 Displaying structures

To obtain a list of all structures defined and whether or not they are in use, use the command
D XCF,STR without any other parameters, as shown in Figure 19-9 (this display was modified
to show only the IMS structures in this example).
D XCF,STR
STRNAME
ALLOCATION TIME
. . .
I#$#EMHQ
06/25/2007 23:17:13
I#$#EMHQOFLW
--I#$#LOCK1
07/03/2007 19:35:22
I#$#LOGEMHQ
07/04/2007 00:47:01
I#$#LOGMSGQ
07/04/2007 00:46:53
I#$#MSGQ
06/25/2007 23:17:15
I#$#MSGQOFLW
--I#$#OSAM
07/04/2007 00:54:29
I#$#RM
07/03/2007 19:34:45
I#$#VSAM
07/04/2007 00:54:27
I#$#VSO1DB1
07/04/2007 21:00:03
I#$#VSO1DB2
07/04/2007 21:00:05
I#$#VSO2DB1
--I#$#VSO2DB2
---
STATUS
ALLOCATED
NOT ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
NOT ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
NOT ALLOCATED
NOT ALLOCATED
Figure 19-9 List of all structures defined (excluding non-IMS structures)
More detailed information about an individual structures can be obtained using the
D XCF,STR,STRNAME=structure_name command. A subset of the response from this command
416
is shown in Figure 19-10. For a complete description, refer to Appendix B, List of structures
on page 499.
-D XCF,STR,STRNAME=I#$#RM
STRNAME: I#$#RM
STATUS: ALLOCATED
TYPE: SERIALIZED LIST
POLICY INFORMATION:
...
DUPLEX
: DISABLED
ENFORCEORDER
: NO
2
3
...
ENTRIES: IN-USE:
ELEMENTS: IN-USE:
LOCKS:
TOTAL:
110 TOTAL:
12 TOTAL:
256
5465,
5555,
2% FULL
0% FULL
...
MAX CONNECTIONS: 32
# CONNECTIONS : 3
---------------- -- -------- -------- -------- ---- ---------------CQSS#$1CQS
03 0003000F #@$1
I#$1CQS 0024 ACTIVE
CQSS#$2CQS
01 00010019 #@$2
I#$2CQS 0047 ACTIVE
CQSS#$3CQS
02 00020010 #@$3
I#$3CQS 0043 ACTIVE
Figure 19-10 Example of a D XCF,STR,STRNAME=I#$#RM command
1 Shows the type of structure, such as LIST

2 Shows that system-managed duplexing of this structure is disabled
3 Shows the Coupling Facility preference list
4 Shows the address spaces connected to the structure
19.6.3 Handling Coupling Facility failures

The different Coupling Facility structures used by IMS all handle the recovery from failures in
different ways.
A number of different scenarios are listed here. In most cases, operator intervention is not
required (this depends on the site configuration), but the information is included here for
reference.
Failure of a CF containing an IRLM Lock Structure

Structure failures can occur if the Coupling Facility fails, or if structure storage is corrupted.
417
If a loss of the IRLM structures or the Coupling Facility that contain the IRLM structure occurs,
then:
IMS batch data sharing jobs end abnormally with a U3303 abend code on the system with
the loss of connectivity. Backout is required for updaters. All the batch data sharing jobs
must be restarted later.
Although the online system continues operating, data sharing quiesces, and transactions
making lock requests are suspended until the lock structure is automatically rebuilt. Each
IRLM participating in the data sharing group is active in the automatic rebuild of the IRLM
lock structure to the alternate CF.
When the rebuilding is complete, transactions that were suspended have their lock
requests processed.
To invoke automated recovery, a second Coupling Facility is required and the CFRM policy
must specify an alternate CF in the preference list.
The target CF structure is repopulated with active locks from the IRLMs. Given that IMS and
IRLM will rebuild the structure1, using system-managed duplexing for the lock structure will
not provide any additional recovery, but it may speed up the recovery.
Failure of a CF containing an OSAM or VSAM cache structure

If a CF containing the IMS OSAM or VSAM cache structures fails, then:
All local OSAM buffers are invalidated if the CF contains an OSAM cache structure. This
means that buffers cannot be used and any future request for blocks requires a read from
DASD. This impacts performance but not availability.
All VSAM buffers are invalidated if the CF contains a VSAM cache structure. The process
for the VSAM buffer set is the same as for the OSAM buffers.
The online subsystems continue, but some transactions may get lock reject status when
receiving an error return code from an operation to the OSAM or VSAM structure. This
results in an abend U3303 code or return of status codes BA or BB, depending on how the
applications have been coded. After an application receives the error return code, no other
programs will begin a new operation to the structure.
All transactions that would initiate an operation that will result in an access to the structure
are placed into a wait. When the structure is rebuilt, they are taken out of this wait state.
IMS batch data sharing jobs abend, and backout is required for updates. All of the batch
data sharing jobs must be restarted later.
All IMS subsystems participate in the rebuild process because the architecture requires this
for all connectors during a rebuild. The contents of the buffer pools are not used in this
rebuild; the OSAM and VSAM structures are rebuilt but empty. No operator involvement is
necessary and the time required for structure rebuild is measured in seconds, rather than
minutes.
Integrity of the data in the databases is not compromised, because this is a store-through
cache structure. Everything that was in the structure has already been written to DASD.
Fast Path DEDB VSO structure read and write errors

Although the following seven situations do not all result in the loss of access to the VSO
cache structure, two of them cause IMS to unload the data from the cache.
418
This assumes that only the lock structure was impacted by the failure. If the lock structure and one or more
connected IRLMs are impacted, all IRLMs in the data sharing group will abend and need to be restarted.
Read errors from the CF structure

If only one VSO cache structure is defined, then the application is presented with a status
code and a DFS2830I message displayed. If two structures are defined for the DEDB VSO
area, then the read request is reissued to the second structure.
When four read errors occur with a single structure

The area is unloaded from the structure to DASD by means of an internal /VUNLOAD
command, resulting in what is known as castout processing. Only changed control intervals
are written, and if the control interval with the read error has been modified, an Error Queue
Element (EQE) is created. The area is stopped on that IMS, but processing continues from
DASD.
When four read errors occur with dual structures

There are still only three errors allowed. The fourth read error causes a disconnect from the
offending structure and a continuation using the second structure. If both structures become
unavailable for use, then the area is stopped on that IMS system. For VSO (non-preload), if
the control interval is not in the Coupling Facility, it is read from DASD and written to the CF.
Write errors to a single CF structure

If an error occurs writing to the VSO cache structure, the control interval is deleted from the
CF structure and written to DASD. If the delete fails, a notification is shipped to the data
sharing IMS subsystems to delete the entry:
If the sharers cannot delete the entry, an EQE is created and propagated to all sharing
subsystems. Because shared VSO does not support I/O toleration, this situation is treated
as though an I/O error on DASD occurred.
If the sharing subsystems can delete the entry, the buffer is written to DASD and
subsequent access to that control interval is from DASD.
Write errors in a dual CF structure

If multiple structures are defined, and a write request to one of the structures fails, the entry is
deleted from the structure by this local IMS or one of the sharing partners. The write is then
done to one of the other structures. If one of the writes is successful, then it is considered to
be a completed write. If the write fails on both structures, the control interval is deleted by one
of the sharing partners in both structures and then written to DASD. The next request for that
control interval will be satisfied from DASD.
When four write errors occur with a single structure

The area is unloaded by means of an internal /VUNLOAD command from the CF using castout
processing and the area is stopped on the detecting IMS. Only changed control intervals are
written and if the control interval with the read error has been modified, an EQE is created.
Processing continues from DASD.
When four write errors occur with dual structures

If the fourth request fails, and multiple structures are defined, IMS disconnects from the
structure in error and continues processing with the one structure remaining.
CQS Shared Message Queue structure failures

CQS takes regular snapshots of the content of the shared queue structures to the SRDS data
sets. If the CF containing the shared queue structure fails, and the structure is not duplexed,
CQS allocates a new structure in an alternate CF (one of the ones listed in the PREFLIST in
the CFRM policy). It populates it with the snap shot from the SRDS data set, and applies all
the log records from the log stream, bringing the structure back up to date as of the time of
the failure.
419
19.6.4 Rebuilding structures

IRLM, OSAM, and VSAM structures can be rebuilt automatically (for example, as a result of a
POPULATECF command or a REALLOCATE command) or directly, as a result of a SETXCF
START,REBUILD command.
There are several reasons why a structure may have to be rebuilt. Most commonly, you want
to move the structure from one CF to another, possibly to allow the CF to be stopped for
maintenance. An example of such a rebuild is shown in Figure 19-11.
-SETXCF START,REBUILD,STRNAME=I#$#OSAM,LOCATION=OTHER
IXC521I REBUILD FOR STRUCTURE I#$#OSAM
HAS BEEN STARTED
I#$#OSAM WAS ACCEPTED.
IXC526I STRUCTURE I#$#OSAM IS REBUILDING FROM
COUPLING FACILITY FACIL02 TO COUPLING FACILITY FACIL01.1
INFO108: 00000003 00000003.
IXC521I REBUILD FOR STRUCTURE I#$#OSAM
HAS BEEN COMPLETED
Figure 19-11 Rebuilding a structure in place
The response to the command shows 1 that the structure was rebuilt to the alternate Coupling
Facility.
Another reason for rebuilding a structure might be to implement a change to the maximum
size for the structure. This is known as a rebuild-in-place because the structure is not
moving to a different CF. An example is shown in Figure 19-12.
-SETXCF START,REBUILD,STRNAME=I#$#VSAM
IXC521I REBUILD FOR STRUCTURE I#$#VSAM
HAS BEEN STARTED
I#$#VSAM WAS ACCEPTED.
IXC526I STRUCTURE I#$#VSAM IS REBUILDING FROM
INFO108: 00000003 00000003.
IXC521I REBUILD FOR STRUCTURE I#$#VSAM
HAS BEEN COMPLETED
2
3
4
Figure 19-12 Rebuilding a structure on the alternative Coupling Facility
In the response to the command:

1 Indicates the structure being rebuilt.
2 Shows the Coupling Facility in use and where it is being rebuilt.
3 The reason why the rebuild was initiated.
4 Rebuild has now completed.
420
Most structures can be rebuilt without impacting the users of those structures. However, in the
case of a rebuild of an IMS OSAM structure, if batch DL/1 jobs using shared OSAM
databases were running at the time of the rebuild, those jobs will abend and will need to be
restarted. To avoid this, we recommend only rebuilding these structures at a time when no
IMS batch DL/1 jobs are running.
19.7 IMS use of Automatic Restart Manager

This section describes the impact of the z/OS facility Automatic Restart Manager (ARM) has
on IMS and its various address spaces. It also provides examples showing what will occur in
various scenarios.
IMS is able to use the function of ARM, and this section summarizes that from an IMS
perspective. Refer to Chapter 6, Automatic Restart Manager on page 83 for further details
about ARM in general.
IBM provides policy defaults for Automatic Restart Management. You can use these defaults,
or you can define your own ARM policy to specify how the various IMS address spaces
should be restarted.
19.7.1 Defining ARM policies

The ARM policies define which jobs are to be restarted by ARM, and some detail about how
this will occur. For further detail about how to define ARM policies, refer to MVS Setting Up A
Sysplex, SA22-7625, under Automatic Restart Management Parameters for Administrative
Data Utility.
The RESTART_ORDER parameter can specify the order in which certain jobs within a
group are started.
The RESTART_GROUP can group a number of different jobs so they always get acted on
together.
The TARGET_SYSTEM can indicate which system you want the tasks restarted on. If
nothing is specified, the restart will occur on any appropriate system, optionally dependent
on CSA by specifying the FREE_CSA parameter.
19.7.2 ARM and the IMS address spaces

The ARM configuration should be closely coordinated with any automation products to avoid
duplicate startup attempts.
Only those jobs or started tasks which register to ARM are eligible to be restarted by ARM.
IMS control regions, IRLMs, FDBR, and CQS may do this registration. For control regions,
FDBR, and CQS, it is optional. For the IRLM, it always occurs when ARM is active.
IMS dependent regions (MPR, BMP, and IFP), IMS batch jobs (DLI and DBB), IMS utilities,
and the online DBRC and DL/I SAS regions do not register to ARM.
There is no need for ARM to restart online DBRC and DLISAS regions, because they are
started internally by the control region.
421
ARM and IRLM

IRLM will always use ARM, if ARM is active.
If IRLM abends, ARM will always restart IRLM on the same system.
ARM element name for IRLM

The ARM element name for IRLM can be determined as:
For data sharing environments, it is a concatenation of the IRLM group name, the IRLM
subsystem name, and the IRLM ID.
For non-data sharing environments, it is the IRLM subsystem name and the IRLM ID. In
our example, this equates to #$#IR#I001 as shown by Figure 19-13.
D XCF,ARMS,DETAIL,ELEMENT=I#$#IR#I001
0
1
0
0
0
1
200
RESTART GROUP:IMS
PACING :
0
FREECSA:
0
01
ELEMENT NAME :I#$#IR#I001
JOBNAME :I#$#IRLM STATE
:AVAILABLE
2
CURR SYS :#@$1
JOBTYPE :STC
ASID
:0055
3
INIT SYS :#@$1
4
EVENTEXIT:DXRRL0F1
ELEMTYPE:SYSIRLM LEVEL
:
0
TOTAL RESTARTS :
0
INITIAL START:07/03/2007 01:49:25
RESTART THRESH :
0 OF 3
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
IST663I INIT OTHER REQUEST FAILED, SENSE=087D0001 511
IST664I REAL OLU=USIBMSC.#@$CCC$1
ALIAS DLU=USIBMSC.#@$CCC$2
IST889I SID = E61318508EB7A1B3
IST1705I SORDER = APPN FROM START OPTION
IST1705I SSCPORD = PRIORITY FROM START OPTION
IST894I ADJSSCPS TRIED FAILURE SENSE
ADJSSCPS TRIED FAILURE SENSE
IST895I
ISTAPNCP
08400007
IST314I END
Figure 19-13 Output from D XCF,ARMS,DETAIL,ELEMENT=I#$#IR#I001
1 Shows that this element is defined in the restart group IMS.

2 Shows the element name, jobname and status.
3 Shows the current system the job is active on.
4 Shows the original system the job was started on.
ARM and the CSL Address Spaces (SCI, RM, OM)

The CSL address spaces will register with ARM unless they have been deliberately disabled
via the ARMRST=N parameter.
ARM element name for CSL

When ARM is enabled, the CSL address spaces register to ARM with an ARM element name.
The element names can be found in Table 19-3 on page 423. The
OMNAME/RMNAME/SCINAME values can be found in the CSL startup parameter for each
address space, or overridden in the JCL parameters.
422
Table 19-3 ARM Element names for IMS CSL address spaces
CSL address space name
ARM element name
OM
CSL + omname + OM
RM
CSL + rmname + RM
SCI
CSL + sciname + SC
Given that the CSL address space names are the same on each system, there is no need to
have ARM active for them, and so ARMRST=N has been coded. This is because they are
already started on the other system. Given this, the ARM policies for these address spaced
do not show up in any displays. If they were active, then they would be displayed by
commands like D XCF,ARMS,DETAIL,ELEMENT=CSLOM1OM.
ARM and the IMS control region

The IMS control region will register with ARM unless it has been deliberately disabled via the
ARMRST=N parameter.
When ARM restarts an IMS online system, IMS will always use AUTO=Y. This is true even if
AUTO=N is specified. The use of AUTO=Y eliminates the need for the operator to enter the
/ERE command.
When ARM restarts an IMS online system, IMS specifies the OVERRIDE parameter when
appropriate. This eliminates the requirement for an operator action during automatic restarts.
ARM will NOT restart IMS if any of the following IMS abends occur:
U0020 - MODIFY
U0028 - /CHE ABDUMP
U0604 - /SWITCH SYSTEM
U0758 - QUEUES FULL
U0759 - QUEUE I/O ERROR
U2476 - CICS TAKEOVER
IMS abends before it completes restart processing. This is used to avoid recursive abends
because another restart would presumably also abend.
ARM element name for IMS control region

The ARM Element name is the IMSID, and it may be used in ELEMENT(imsid) and
ELEMENT_NAME(imsid) in the ARM policy. The Element Type is SYSIMS, and it may be
used in ELEMENT_TYPE(SYSIMS) in ARM policy. In our example, Figure 19-14 on page 424
shows a display of the IMS ARM policy.
423
D XCF,ARMS,DETAIL,ELEMENT=I#$1
0
1
0
0
0
1
200
RESTART GROUP:IMS
PACING :
0
FREECSA:
0
0
ELEMENT NAME :I#$1
JOBNAME :I#$1CTL STATE
:AVAILABLE
1
CURR SYS :#@$1
JOBTYPE :STC
ASID
:0056
INIT SYS :#@$1
EVENTEXIT:*NONE*
ELEMTYPE:SYSIMS
LEVEL
:
1
TOTAL RESTARTS :
0
INITIAL START:07/03/2007 01:49:56
RESTART THRESH :
0 OF 3
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
Figure 19-14 Output from the command D XCF,ARMS,DETAIL,ELEMENT=I#$1
1 Shows the element name of the IMS name I#$1, and the jobname of I#$1CTL.
ARM and CQS

CQS will register with ARM unless it has been deliberately disabled via the ARMRST=N
parameter.
Module CQSARM10 contains a table of CQS abends for which ARM restarts will not be done.
Users may modify this table. The table is shipped with the following abends:
0001 - ABEND during CQS initialization
0018 - ABEND during CQS restart
0020 - ABEND during CQS restart
Because a CQS must execute with its IMS system, it should be included in a restart group
with its IMS.
ARM element name for CQS

When ARM is enabled, CQS registers to ARM with an ARM element name of CQS + cqsssn
+ CQS. Use this ARM element name in the ARM policy to define the ARM policy for CQS.
Note that cqsssn is the CQS name. It can be defined either as a CQS execute parameter, or
with the SSN= parameter in the CQSIPxxx IMS PROCLIB member.
ARM and FDBR

The FDBR address space will register with ARM unless it has been deliberately disabled via
the ARMRST=N parameter.
When an FDBR system using ARM tracks an IMS system, it notifies ARM that it is doing the
tracking and that ARM should not restart this IMS if it fails. FDBR uses the ASSOCIATE
function of ARM to make this notification. So, even if IMS registers to ARM, IMS will not be
restarted after its failures when FDBR is active and using ARM.
If FDBR is terminated normally, it notifies ARM. This tells ARM to restart the tracked IMS if
this IMS has previously registered to ARM. This is appropriate because FDBR is no longer
available to perform the recovery processes.
424
When ARM is used for IMS, an installation can choose either to use or not to use ARM for
FDBR. If FDBR does not use ARM, it cannot tell ARM not to restart IMS. In this case, a failure
of IMS will cause FDBR to do its processing and ARM to restart IMS. This could be
advantageous. You would expect FDBR processing to complete before the restart of IMS by
ARM completes. If so, locks would be released quickly by FDBR and a restart of IMS by ARM
would occur automatically.
ARM element name for FDBR

The ARM element name for FDR is the FDR IMS ID.
19.7.3 ARM and IMS Connect

IMS Connect does not use ARM.
19.7.4 ARM in this test example

For the system being used for the creation of these examples, their ARM element names can
be found in Table 19-4.
Table 19-4 ARM element names for this example
Address space name
ARM element name
IRLM
I#$#IR#Innn
(In this case, the system name of n is expanded
by z/OS to 3 characters; thus 1 becomes 001.
SCI
CSLCSInSC
OM
CSLOMnOM
RM
CSLRMnRM
CTL
I#$n
CQS
CQSS#SnCQS
FDBR
F#$n
Note that n = 1, 2, or 3.
Because the SCI, OM and RM address spaces are identical on each of the three systems in
our example, there is no point to having any of them registered to ARM because the address
spaces are already active on each system.
19.7.5 Using the ARM policies

If all the IMS ARM policies have been defined to belong to a single recovery group (for
example, such as IMS), then all the ARM policies can be displayed with a single system
command D XCF,ARMS,DETAIL,RG=IMS with the output showing the ARM policies similar to
those in Table 19-4.
425
19.8 IMS operational issues

This section describes some of the common issues that could be experienced while operating
an IMSplex. It also explains what needs to be done and how commands can be issued to
achieve this.
Because each installation is configured and managed differently, which teams have access to
issue these commands, and which team is responsible for doing this will vary. The information
is provided here for reference. Refer to your own installation to determine who should be
issuing these commands.
A number of common operational issues are discussed in this section. For more complete
information regarding all the options and command formats, refer to IMS Command
Reference Manual V9, SC18-7814.
There are now two types of IMS commands: the Type 1 or classic IMS commands have been
available for many years, and the newer Type 2 commands have the newer sysplex integrated
functionality.
Type 1 IMS commands

These are the traditional types of IMS commands that can be entered from:
A 3270 session beginning with a forward slashmark (/)
A WTOR prompt beginning with a forward slashmark (/)
An MCS console, using the IMSID to route the command to IMS, without a forward
slashmark (/)
From within external automation packages (that is, NetView)
From within an application program as a CMD or ICMD call
Type 2 IMS commands

These new type 2 commands are used from the IMS Single Point Of Control (SPOC), which
can only be accessed via:
ISPF SPOC, accessible via the ISPF Command:
EXEC 'imshlq.SDFSEXEC(DFSAPPL)' 'HLQ(imshlq)
Select option 1 for SPOC, and you will get the panel shown in Figure 19-15 on
page 427.
426
Figure 19-15 Example of the ISPF IMS SPOC
IMS SPOC is available with the Control Center for IMS, which is provided as part of DB2 9
for Linux UNIX and Windows Control Center. This is available via the IMS Web page:
http://www.ibm.com/software/data/ims/imscc/
These commands can also be entered from a REXX program. Details about the sample
REXX program can be found in IMS Common Service Layer Guide and Reference V9,
SC18-7816.
Type 2 commands available with IMS Version 9 are:
DELETE - Used to delete Language Environment options previously altered with the
UPDATE command.
INITIATE - Used for online change and online reorg management across the sysplex.
QUERY - querying the status of IMS components (similarly to the /DISPLAY command), as
well as for querying IMSPLEX or Coupling Facility structure status.
TERMINATE - Used for online change and online reorg management across the sysplex.
UPDATE - Used to update the status of various IMS definitions, similarly to the /ASS or /CHA
command.
As each IMS release comes out, there will be more and more functionality added to the Type
2 command set.
19.8.1 IMS commands

The following examples show commands you can use to query the status of active resources.
427
Type 1 commands
These commands would not normally cover sysplex-related functions.
Displaying the active regions.
Displaying connection to other subsystems (that is, DB2, MQ, OTMA, and so on).
Displaying the status of a particular resource.
Type 2 commands
These commands are designed for sysplex-related functions.
Displaying the status of a particular resource.
Display the status of the different components of an IMSplex.
Display the status of the Coupling Facility structures.
19.8.2 CQS commands

A number of CQS commands can be issued from within IMS. These are related to the Shared
Queues list structures for Message Queues and Fast Path Expedited Message Handler
queues.
CQS queries
The CQS structures can be queried by using the /CQQ IMS command. As shown in
Figure 19-16, the command displays:
1 LEALLOC the list entries allocated.
2 LEINUSE the list entries in use.
3 ELMALLOC the elements allocated.
4 ELMINUSE the elements in use.
TRUCTURE NAME
I#$#MSGQ
I#$#MSGQOFLW
I#$#EMHQ
I#$#EMHQOFLW
LEALLOC1 LEINUSE2 ELMALLOC3 ELMINUSE4 LE/EL

7570
6
7613
5 0001/0001
N/A
N/A
N/A
N/A
N/A
7570
6
7613
5 0001/0001
N/A
N/A
N/A
N/A
N/A
Figure 19-16 IMS response to the command /CQQ STATISTICS STRUCTURE ALL
CQS checkpoints
CQS Checkpoints for individual Coupling Facility list structures can be triggered by the IMS
command /CQCHKPT or /CQC. An example is shown in Figure 19-17.
DFS058I 21:15:15 CQCHKPT COMMAND IN PROGRESS I#$1
DFS1972I CQCHKPT SHAREDQ COMMAND COMPLETE FOR STRUCTURE=I#$#MSGQ
DFS1972I CQCHKPT SHAREDQ COMMAND COMPLETE FOR STRUCTURE=I#$#EMHQ
Figure 19-17 IMS response to the command /CQC SHAREDQ STRUCTURE ALL
CQS Set command

Use the /CQSET or /CQS command to tell CQS whether to take a structure checkpoint during
normal shutdown or not, for example /CQSET SHUTDOWN SHAREDQ ON STRUCTURE ALL.
After it is issued, CQS will always shut down along with IMS; otherwise, it will remain active.
428
19.8.3 IRLM commands

Each IMS image will have a unique IMSID, which can be determined from the respective
outstanding reply.
The IMS workload is now on more than one system, where each has access to the same set
of databases. Control of this access is managed by IRLM.
The communication with IRLM is accomplished by using the standard z/OS modify command
with various options.
STATUS option
The command F irlmproc,STATUS displays status, work units in progress, and detailed lock
information for each DBMS identified to this instance of IRLM (irlmproc is the procedure name
for the IRLM address space.) Figure 19-18 is an example showing an IMS control region and
an FDR region on each system.
DXR101I IR#I001 STATUS SCOPE=GLOBAL
DEADLOCK: 0500
SUBSYSTEMS IDENTIFIED
NAME
T/OUT
STATUS
UNITS
FDRI#$3
0300
UP-RO
0
I#$1
0300
UP
1
DXR101I End of display
662
HELD
0
2
WAITING
0
0
RET_LKS
01
02
Figure 19-18 Response from an IRLM STATUS command F I#$#IRLM,STATUS
1 This shows the FDBR region active in READ ONLY mode.

2 This shows the IMS system active, with two held locks.
STATUS,ALLD option
The ALLD option shows all the subsystems connected to all the IRLMs in the data sharing
group that IRLM belongs to. The RET_LKS field is very important. It shows how many
database records are retained by a failing IRLM and is therefore unavailable to any other IMS
subsystem. See Figure 19-19 for an example.
DXR102I IR#I001 STATUS 731
NAME
STATUS
RET_LKS
FDRI#$1 UP-RO
0
FDRI#$2 UP-RO
0
FDRI#$3 UP-RO
0
I#$1
UP
0
I#$2
UP
0
I#$3
UP
0
IRLMID
002
003
001
001
002
003
IRLM_NAME
IR#I
IR#I
IR#I
IR#I
IR#I
IR#I
IRLM_LEVL
1.009 1
1.009
1.009
1.009 2
1.009
1.009
Figure 19-19 Response from an IRLM STATUS command F I#$#IRLM,STATUS,ALLD
1 This shows all the FDBR regions active with READ ONLY mode, and which IRLM they are
connected to.
2 This shows the IMS systems, together with which IRLM they are connected to.
429
STATUS,ALLI option
The ALLI option shows the names and status of all IRLMs in the data sharing group, as
IRLMS PARTICIPATING IN DATA SHARING GROUP FUNCTION LEVEL=2.025
IRLM_NAME IRLMID STATUS LEVEL SERVICE MIN_LEVEL MIN_SERVICE
IR#I
003
UP
2.025 PK05211
1.022
PQ52360
IR#I
002
UP
2.025 PK05211
1.022
PQ52360
IR#I*
001
UP
2.025 PK05211
1.022
PQ52360
Figure 19-20 Response from an IRLM STATUS command F I#$#IRLM,STATUS,ALLI
STATUS,MAINT option
The MAINT option lists the PTF levels of all the modules active in IRLM. The command is:
F irlmproc,STATUS,MAINT
ABEND option
This option causes IRLM to terminate abnormally. IRLM informs all DBMSs linked to it,
through their status exits, that it is about to terminate. The command is:
F irlmproc,ABEND
RECONNECT option
This option causes IMS to reconnect to the IRLM specified in the IRLMNM parameter in the
IMS control region JCL. This is necessary after an IRLM is restarted following an abnormal
termination and IMS was not taken down. The command is:
F irlmproc,RECONNECT
PURGE option
Warning: Use the PURGE option with extreme caution. It is included in this section for
completeness only.
The PURGE option causes IRLM to release any retained locks it holds for IMS. This
command must be used with care in these situations:
The RECON reflects that database backout was done, but IRLM was not up at time of the
backout.
A decision is made not to recover, or to defer recovery, but the data is required to be
available to other IMS subsystems.
The command is:
F irlmproc,PURGE,IMSname
The PURGE ALL option of this command is even more hazardous. It allows you to release all
retained locks of all IMS subsystems held by a specific IRLM.
430
19.9 IMS recovery procedures

The following section includes failure recovery scenarios of a Coupling Facility, IMS, and
IRLM. These scenarios are presented as a guide to the processing that occurs in a sysplex
data sharing environment when major components fail, together with the actions required to
assist with the resolution.
19.9.1 Single IMS abend without ARM and without FDR

In this scenario, assume that IMS1 was running on system1.
If IMS1 were to abend, regardless of the reason, then other IMS systems in the IMSplex will
continue to function normally. However, any database locks held by the failing IMS system will
be retained, thus locking out any other IMS functions requiring those database blocks.
The abending IMS system will then need to be restarted manually. This could be on the same
system or on an alternate system.
If started with the automatic restart parameter enabled (refer to the IMS PARMLIB member
DFSPBxxx option AUTO=Y), then IMS will automatically detect that an emergency restart
is required.
If IMS is started with the automatic restart parameter disabled (refer to the IMS parmlib
member DFSPBxxx option AUTO=N), then when the WTOR message as shown in
Figure 19-21 is displayed, the emergency restart command /ere. must be entered. During
the emergency restart, the in-flight database updates will be backed out and any locks
held will be released.
DFS810A IMS READY
yyyyddd/hhmmsst I#$1CTL
Figure 19-21 DFS810A message indicating a manual IMS restart is required
Attention: When entering the /ere command, the period after ere is required. Otherwise,
the WTOR message DFS972A *IMS AWAITING MORE INPUT will be displayed. If this does
occur, simply reply with a period.
IRLM, SCI, OM, RM. or CQS are unaffected by this.

All IMS Connect instances simply lose connection with the failing IMS system, and will
automatically reconnect after IMS has restarted, as shown in Figure 19-22.
HWSD0282I COMMUNICATION WITH DS=I#$1
HWSD0290I CONNECTED TO DATASTORE=I#$1
CLOSED;
Figure 19-22 IMS Connect messages when IMS fails
19.9.2 Single IMS abend with ARM but without FDR

If IMS1 were to abend, regardless of the reason, other IMS systems in the sysplex will
be retained, thus locking out any other IMS functions requiring those database blocks. The
IMS system will then need to be manually restarted on the same system or on a different one.
431
The IRLM, SCI, OM, RM or CQS are unaffected by this.

IMS Connect behaves the same as described in 19.9.1, Single IMS abend without ARM and
without FDR on page 431.
19.9.3 Single IMS abend with ARM and FDR

If IMS1 were to abend, regardless of the reason, the other IMS systems in the IMSplex will
be retained, thus locking out any other IMS functions requiring those database blocks.
FDR will immediately detect that IMS is abending and will manage any database dynamic
backouts and release all locks before ending, as shown in Figure 19-23.
DFS3257I ONLINE LOG CLOSED ON DFSOLP05 F#$1
DFS3257I ONLINE LOG CLOSED ON DFSOLS05 F#$1
DFS4166I FDR FOR (I#$1) DB RECOVERY PROCESS STARTED. REASON = IMS FAILURE 1
DFS3257I ONLINE LOG NOW OPENED ON DFSOLP99 F#$1
DFS3257I ONLINE LOG NOW OPENED ON DFSOLS99 F#$1
DFS3261I WRITE AHEAD DATA SET NOW ON DFSWADS0 F#$1
148 DFS4167A FDR FOR (I#$1) WAITING FOR ACTIVE SYSTEM TO COMPLETE I/O PREVENTION
PREVENTION COMPLETES
$HASP100 ARCHI#$1 ON INTRDR
CONWAY
FROM STC10411 I#$1FDR
IRR010I USERID I#$1
IS ASSIGNED TO THIS JOB.
DFS2484I JOBNAME=ARCHI#$1 GENERATED BY LOG AUTOMATIC ARCHIVING F#$1
DFS4171I FDR FOR (I#$1) ACTIVE IMS TERMINATION NOTIFIED BY XCF. OPERATION RESUME
DFS4168I FDR FOR (I#$1) DATABASE RECOVERY COMPLETED
2
CONWAY
IRR010I USERID I#$1
DFS092I IMS LOG TERMINATED
F#$1
DFS627I IMS RTM CLEANUP ( EOT ) COMPLETE FOR JS I#$1FDR .I#$1FDR .IEFPROC ,RC=00
---------TIMINGS (MINS.)---------JOBNAME STEPNAME PROCSTEP
RC EXCP
CPU
SRB
VECT VAFF CLOCK
SE
-I#$1FDR STARTING IEFPROC
00 11083
.02
.00
.00
.00
35.6
15
-I#$1FDR ENDED. NAMETOTAL CPU TIME=
.02 TOTAL ELAPSED
$HASP395 I#$1FDR ENDED
Figure 19-23 FDR job following an IMS abend
1 This shows that FDBR has recognized that it needs to do a recovery, and the reason why.
2 This shows the FDBR recovery has completed.
ARM will then restart IMS on the same system, which will automatically perform the
emergency restart.
The IRLM, SCI, OM, RM or CQS are unaffected by this.
IMS Connect behaves the same as described in 19.9.1, Single IMS abend without ARM and
without FDR on page 431.
432
19.9.4 Single system abend without ARM and without FDR

All the address spaces on system1 simply failed along with system1, and they need to be
manually restarted on the same or a different system.
When the IMS is finally restarted on the same system, or a different system at a later time,
then the messages shown in Figure 19-24 will be displayed, assuming that AUTO=Y was
specified. In this case, the previous WTOR as part of the DFS3139I message needs to be
located and used to reply with the /ere override. command (note that the period is
required.) During this emergency restart, IMS will perform any dynamic backout and release
any locks it had.
DFS3139I IMS INITIALIZED, AUTOMATIC RESTART PROCEEDING
. . .
*DFS0618A A RESTART OF A NON-ABNORMALLY TERMINATED SYSTEM MUST SPECIFY EMERGENCY BACKUP
OR OVERRIDE
DFS0618A A RESTART OF A NON-ABNORMALLY TERMINATED SYSTEM... I#$1
DFS000I....MUST SPECIFY EMERGENCY BACKUP OR OVERRIDE.
I#$1
DFS3874I LEAVERSE MODE=IOP WAS ISSUED I#$1
DFS3875I LEAVEAVM MODE=NORMAL WAS ISSUED I#$1
DFS3626I RESTART HAS BEEN ABORTED I#$1
DFS3626I RESTART HAS BEEN ABORTED
I#$1
Figure 19-24 Messages following an system failure without ARM, requiring /ere override
The IMS Connect address space will need to be manually restarted.

If multiple IMS Connect address spaces have been defined to listen on the same TCP/IP port,
then all new requests from that port will continue using the alternate IMS Connect address
space. If there are no other IMS Connect address spaces listening on the same TCP/IP port,
then any TCP/IP client trying to use that port will not get any response until the failed IMS
Connect address space has been restarted.
The IMS Connect instances running on other systems will simply lose connection to the failed
IMS system, as described in 19.9.1, Single IMS abend without ARM and without FDR on
page 431.
19.9.5 Single system abend with ARM but without FDR

All the address spaces on system1 simply fail along with system1, and ARM will automatically
restart IMS on any valid system within the sysplex, based on the ARM policy. It will
automatically attempt an emergency restart, and will internally provide the /ERE OVERRIDE.
that was required manually in 19.9.4, Single system abend without ARM and without FDR
on page 433.
IMS Connect will behave as described in 19.9.5, Single system abend with ARM but without
FDR on page 433.
19.9.6 Single system abend with ARM and FDR

In this scenario, assume that IMS1 was running on system1. IMS1FDR was running on
system2.
433
All the address spaces on system1 simply fail along with system1. FDR running on system2
produces the messages as shown in Figure 19-25, and then ARM will automatically restart
IMS on any valid system within the sysplex, based on the ARM policy.
DFS4165W FDR FOR (I#$1) XCF DETECTED TIMEOUT ON ACTIVE IMS
SYSTEM,REASON=SYSTEM,DIAGINFO=0C030384 F#$1
DFS4164W FDR FOR (I#$1) TIMEOUT DETECTED DURING LOG AND XCF SURVEILLANCE
F#$1
DFS4166I FDR FOR (I#$1) DB RECOVERY PROCESS STARTED. REASON = XCF NOTIFICATION
DFS3257I ONLINE LOG NOW OPENED ON DFSOLP04 F#$1
DFS3257I ONLINE LOG NOW OPENED ON DFSOLS04 F#$1
DFS4168I FDR FOR (I#$1) DATABASE RECOVERY COMPLETED
CONWAY
IRR010I USERID I#$1
CONWAY
IRR010I USERID I#$1
CONWAY
IRR010I USERID I#$1
DFS092I IMS LOG TERMINATED
F#$1
DFS627I IMS RTM CLEANUP ( EOT ) COMPLETE FOR JS I#$1FDR .I#$1FDR .IEFPROC
,RC=00
Figure 19-25 FDR address space recovering IMS following an system failure
IMS Connect will behave as described in 19.9.5, Single system abend with ARM but without
FDR on page 433.
19.9.7 Single Coupling Facility failure

If there are two Coupling Facilities, and one of them fails, then the following messages are to
be expected while the backup Coupling Facility manages the recreation of the failed
structures.
IMS control region

Figure 19-26 on page 435 shows the messages displayed by the control region when a
Coupling Facility fails.
434
when the
DFS3306A
DFS3705I
DFS3705I
DFS2500I
DFS2500I
DFS2823I
DFS2574I
DFS0488I
DFS4450I
DFS4450I
preferred Coupling Facility fails

CTL REGION WAITING FOR RM - I#$1
AREA=DFSIVD3B DD=DFSIVD33 CLOSED I#$1
AREA=DFSIVD3B DD=DFSIVD34 CLOSED I#$1
DATASET DFSIVD33 SUCCESSFULLY DEALLOCATED I#$1
DATASET DFSIVD34 SUCCESSFULLY DEALLOCATED I#$1
AREA DFSIVD3B DISCONNECT FROM STR: I#$#VSO1DB2
SUCCESSFUL I#$1
AREA=DFSIVD3B STOPPED
I#$1
STO COMMAND COMPLETED. AREA= DFSIVD3B RC= 0 I#$1
RESOURCE STRUCTURE REPOPULATION STARTING I#$1
RESOURCE STRUCTURE REPOPULATION COMPLETE I#$1
after the preferred Coupling Facility is

DFS4450I RESOURCE STRUCTURE REPOPULATION
made available again:

STARTING I#$1
COMPLETE I#$1
STARTING I#$1
COMPLETE I#$1
Figure 19-26 Coupling Facility failure messages in the IMS control region
As a result of this the VSO database area DFSIVD3B, which was in use at the time, is now
stopped, and marked with an EEQE status. Because this is a write error, it causes the copy of
the CI in cache structure or structures to be deleted. Other sharing systems no longer have
access to the CI.
If any of the other IMS systems try to access this CI, they will receive messages indicating
that another system has this CI as a retained lock, as shown in Figure 19-27.
DFS3304I IRLM LOCK REQUEST REJECTED. PSB=DFSIVP8 DBD=IVPDB3 JOBNAME=IV3H212J
DFS0535A RGN=
1, HSSP CONN PROCESS ATTEMPTED AREA DFSIVD3A PCB LABEL HSSP
DFS0535I RC=03, AREA LOCK FAILED. I#$3
+STATUS FH, DLI CALL = ISRT
Figure 19-27 Messages indicating IRLM retained locks
To resolve this, issue the /VUN AREA DFSIVD3B command to take the area out of VSO, thus
ensuring that all updates are now reflected on DASD. Next, issue the /STA AREA DFSIVD3B
command, and it will be reloaded into VSO without any errors.
IMS DLISAS
Figure 19-28 on page 436 shows the messages displayed by DLISAS when a Coupling
Facility fails.
435
when
the preferred Coupling Facility fails

:IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#OSAM 437
WAS SUCCESSFUL. JOBNAME: I#$1DLI ASID: 0047
CONNECTOR NAME: IXCLO0180001 CFNAME: FACIL01
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#VSAM 440
after the preferred Coupling Facility is made available again:
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#VSAM 615
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#OSAM 616
Figure 19-28 Coupling Facility failure messages in the DLISAS region
IRLM
Figure 19-29 on page 437 shows the messages displayed by IRLM when a Coupling Facility
fails.
436
when the preferred Coupling Facility fails

*IXL158I PATH 0F IS NOW NOT-OPERATIONAL TO CUID: 030F 392
*IXL158I PATH 10 IS NOW NOT-OPERATIONAL TO CUID: 030F 393
DXR143I IR#I001 REBUILDING LOCK STRUCTURE BECAUSE IT HAS FAILED OR AN IRLM
LOST
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#LOCK1 439
WAS SUCCESSFUL. JOBNAME: I#$#IRLM ASID: 0046
CONNECTOR NAME: I#$#$$$$$IR#I001 CFNAME: FACIL01
IXL030I CONNECTOR STATISTICS FOR LOCK STRUCTURE I#$#LOCK1, 461
CONNECTOR I#$#$$$$$IR#I001:
0001001D 00000000 00000008 001E000F
...
...
IXL031I CONNECTOR CLEANUP FOR LOCK STRUCTURE I#$#LOCK1, 462
CONNECTOR I#$#$$$$$IR#I001, HAS COMPLETED.
INFO: 0001001D 00000000 00000000 00000000 00000000 00000004
DXR146I IR#I001 REBUILD OF LOCK STRUCTURE COMPLETED SUCCESSFULLY WITH
2M
LOC
ENTRIES
DXR145I IR#I001 REBUILDING LOCK STRUCTURE AT OPERATORS REQUEST
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#LOCK1 610
0001001D 00000000 00000008 001E000F
...
...
INFO: 0001001D 00000000 00000000 00000000 00000000 00000004
DXR146I IR#I001 REBUILD OF LOCK STRUCTURE COMPLETED SUCCESSFULLY WITH
LOC
ENTRIES
2M
Figure 19-29 Coupling Facility failure messages in IRLM
CQS
Figure 19-30 on page 438 shows the messages displayed by CQS when a Coupling Facility
fails.
437
when the preferred Coupling Facility fails

CQS0202I STRUCTURE I#$#RM
STATUS CHANGED; STATUS=LOST CONNECTION
S#$1C
QUIESCED FOR STRUCTURE REBUILD S#$1CQS
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#RM 438
WAS SUCCESSFUL. JOBNAME: I#$1CQS ASID: 0024
CONNECTOR NAME: CQSS#$1CQS CFNAME: FACIL01
CQS0242E CQS S#$3CQS FAILED STRUCTURE REBUILD FOR STRUCTURE I#$#RM
RC=nnn
RESUMED AFTER STRUCTURE REBUILD S#$1CQS
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#RM 546
CQS0008W STRUCTURE I#$#RM
IS VOLATILE; S#$1CQS
STATUS CHANGED; STATUS=CONNECTION S#$1CQS
REPOPULATION REQUESTED S#$1CQS
QUIESCED FOR STRUCTURE REBUILD S#$1CQS
IXL014I IXLCONN REBUILD REQUEST FOR STRUCTURE I#$#RM 602
CQS0240I CQS S#$2CQS STARTED STRUCTURE COPY
FOR STRUCTURE I#$#RM
CQS0241I CQS S#$2CQS COMPLETED STRUCTURE COPY
FOR STRUCTURE I#$#RM
RESUMED AFTER STRUCTURE REBUILD S#$1CQS
Figure 19-30 Coupling Facility failure messages in the CQS address space
Resource Manager
Figure 19-31 shows the messages displayed by RM when a Coupling Facility fails.
when the preferred Coupling facility fails:
CSL2040I RM RM1RM
IS QUIESCED; STRUCTURE I#$#RM
RM1RM
IS UNAVAILABLE

CSL2041I RM RM1RM
IS AVAILABLE; STRUCTURE I#$#RM
IS AVAILABLE
RM1RM
CSL2020I STRUCTURE I#$#RM
REPOPULATION SUCCEEDED RM1RM
Figure 19-31 Coupling Facility failure messages in the RM address space
19.9.8 Dual Coupling Facility failure

If there is either a single Coupling Facility and it fails, or there are duplex Coupling Facilities
and both of them fail, then the entire sysplex comes down. In this case, all data in the
Coupling Facility is lost.
Following the IPL, the first IMS system will be restarted, and following sequence identifies
what is experienced, and what needs to be done.
438
CQS abend
IMS itself will not complete initialization as it waits for the CQS address space, which abends
because it is unable to connect to the I#$#EMHQ and I#$#MSGQ structures, as shown in
Figure 19-32. The RM address space will probably also abend as a result. The important
piece is message CQS0350W.
Restart the RM address space, which will then also wait for CQS to restart.
CQS0350W
CQS0350W
CQS0350W
CQS0001E
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
...
CQS0350W
CQS0350W
CQS0350W
CQS0001E
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
BPE0006I
CQS LOG CONNECT POSSIBLE LOSS OF DATA 140

LOG STREAM: #@$#.SQ.EMHQ.LOG
STRUCTURE: I#$#EMHQ
S#$1CQS
CQS INITIALIZATION ERROR IN CQSIST10, CQSLOG10 RC=00000020
CQS STRD TCB ABEND U0014-000000A0, THD=STRD DIAG=1004000184
MODULE ID = CQSIST10+0140+2006 EP = 0D0C90A0
PSW = 077C1000 8D0C9898
OFFSET = 000007F8
R0-3 84000000 8400000E 00040001 00000001
R4-7 0C7712B8 0D0C9EF8 0D0E37A0 0D0E1168
R8-11 0000000D 8D0C9666 0C771708 0C713B80
R12-15 8D0C90A0 0D0E11F0 8D0C9672 000000A0
CQS LOG CONNECT POSSIBLE LOSS OF DATA 153
LOG STREAM: #@$#.SQ.MSGQ.LOG
STRUCTURE: I#$#MSGQ
S#$1CQS
CQS INITIALIZATION ERROR IN CQSIST10, CQSLOG10 RC=00000020
CQS STRD TCB ABEND U0014-000000A0, THD=STRD DIAG=1004000184
MODULE ID = CQSIST10+0140+2006 EP = 0D0C90A0
PSW = 077C1000 8D0C9898
OFFSET = 000007F8
R0-3 84000000 8400000E 00040001 00000001
R4-7 0C7132E0 0D0C9EF8 0D0D6860 0D0D3168
R8-11 0000000D 8D0C9666 0C713730 0C713B80
R12-15 8D0C90A0 0D0D31F0 8D0C9672 000000A0
Figure 19-32 CQS abend following complete loss of Coupling Facilities
Cold start CQS

Based on the messages shown in Figure 19-32, there is no possibility of recovering the
contents of the Coupling Facility. From a shared queues perspective, the only option is to cold
start the CQS structures, knowing that the data in the Coupling Facility will be lost. The
process for doing this is based on the CQS Structure Cold Start section documented in IMS
Common Queue Server Guide and Reference Version 9, SC18-7815, which states:
Ensure that all CQSs are disconnected from the structure.
Delete the primary and overflow structures on the Coupling Facility.
Scratch both structure recovery data sets (SRDS 1 and 2) for the structure.
Our experience with this procedure is described in the following section.
Disconnect CQS from the structures

Trying to disconnect CQS from the structures did not work, as shown in Figure 19-33 on
page 440. However, this was handled as described in Delete the primary and overflow
structures on the Coupling Facility on page 440.
439
-SETXCF FORCE,CONNECTION,STRNAME=I#$#EMHQ,CONNAME=ALL
IXC363I THE SETXCF FORCE FOR ALL CONNECTIONS FOR STRUCTURE
I#$#EMHQ WAS REJECTED:
FORCE CONNECTION NOT ALLOWED FOR PERSISTENT LOCK OR SERIALIZED LIST
Figure 19-33 Attempting to disconnect CQS from the CF structures using the SETXCF command
Delete the primary and overflow structures on the Coupling Facility

In our case, we deleted both the EMHQ and MSGQ structures, as well as their corresponding
overflow structures. The structure names are all defined in the CQSSGxxx IMS proclib
member. The commands used in this example are shown in Figure 19-34.
SETXCF
SETXCF
SETXCF
SETXCF
FORCE,STRUCTURE,STRNAME=I#$#MSGQ
FORCE,STRUCTURE,STRNAME=I#$#EMHQ
FORCE,STRUCTURE,STRNAME=I#$#MSGQOFLW
FORCE,STRUCTURE,STRNAME=I#$#EMHQOFLW
Figure 19-34 Commands used to delete all the CQS structures
The resulting output is shown in Figure 19-35, which also shows the connections were
deleted.
-SETXCF FORCE,STRUCTURE,STRNAME=I#$#EMHQ
IXC353I THE SETXCF FORCE REQUEST FOR STRUCTURE
I#$#EMHQ WAS COMPLETED:
STRUCTURE DELETED BUT ALSO RESULTED IN DELETED CONNECTION(S)
Figure 19-35 Result of deleting a CQS structure using the SETXCF command
Scratch both structure recovery data sets (SRDS 1 and 2) for the structure
Important: SCRATCHING both SRDS datasets means to DELETE/DEFINE the datasets.
The term scratch in this context means to delete and redefine the VSAM ESDS used for
both SRDS1 and SRDS2. The data sets involved are also specified in the CQSSGxxx IMS
proclib members SRDSDSN1 and SRDSDSN2 for each structure.
To achieve this a simple IDCAMS delete/define of all the SRDS data sets is required. If you do
not have the IDCAMS define statements available, refer to CQS Structure Recovery Data
Sets in IMS Common Queue Server Guide and Reference Version 9, SC18-7815.
Scratch the CQS log structure

As with the SRDS, the term scratch in this context also means to delete and redefine the
CQS log structure, using the IXCMIAPU utility as shown in Figure 19-36 on page 441 and
Figure 19-37 on page 441. The log structure names can be found in the CQSSGxxx IMS
proclib member.
440
//DEL
EXEC PGM=IXCMIAPU
//SYSPRINT DD
SYSOUT=*
//SYSABEND DD
SYSOUT=*
//SYSIN
DD
*
DATA TYPE(LOGR) REPORT(YES)
DELETE LOGSTREAM NAME(#@$#.SQ.EMHQ.LOG)
DELETE LOGSTREAM NAME(#@$#.SQ.MSGQ.LOG)
DELETE STRUCTURE NAME(I#$#LOGEMHQ)
DELETE STRUCTURE NAME(I#$#LOGMSGQ)
Figure 19-36 Sample JCL to delete the CQS log streams
Figure 19-37 shows the sample JCL to define the log streams.
//DEF
EXEC PGM=IXCMIAPU
//SYSPRINT DD SYSOUT=*
//SYSABEND DD SYSOUT=*
//SYSIN
DD *
DATA TYPE(LOGR) REPORT(YES)
DEFINE STRUCTURE
NAME(I#$#LOGEMHQ)
LOGSNUM(1)
MAXBUFSIZE(65272)
AVGBUFSIZE(4096)
DEFINE STRUCTURE
NAME(I#$#LOGMSGQ)
LOGSNUM(1)
MAXBUFSIZE(65272)
AVGBUFSIZE(4096)
DEFINE LOGSTREAM
NAME(#@$#.SQ.EMHQ.LOG)
STRUCTNAME(I#$#LOGEMHQ)
LS_DATACLAS(LOGR4K)
HLQ(IMSU#@$#)
MODEL(NO)
LS_SIZE(1000)
LOWOFFLOAD(0)
HIGHOFFLOAD(80)
STG_DUPLEX(NO)
RETPD(0)
AUTODELETE(NO)
DASDONLY(NO)
DEFINE LOGSTREAM
NAME(#@$#.SQ.MSGQ.LOG)
STRUCTNAME(I#$#LOGMSGQ)
LS_DATACLAS(LOGR4K)
HLQ(IMSU#@$#)
MODEL(NO)
LS_SIZE(1000)
LOWOFFLOAD(0)
HIGHOFFLOAD(80)
STG_DUPLEX(NO)
RETPD(0)
AUTODELETE(NO)
DASDONLY(NO)
Figure 19-37 Sample JCL to define the log streams
441
Restart CQS again

When restarting CQS again after manually performing the earlier steps, CQS will start,
reallocate the required structures, and automatically COLD start.
IMS will require /ERE OVERRIDE.

As with a normal non-sysplex IMS and z/OS failure scenario, IMS will require the /ere
override. command to restart.
Restart other IMS systems

All other IMS systems in the IMSplex will restart, each with an /ere override. command.
This will automatically start the CQS address space as well, and these CQS address spaces
will issue a WTOR asking what type of CQS start to perform. These should all be responded
to with COLD, as shown in Figure 19-38.
*094 CQS0032A ENTER CHECKPOINT LOGTOKEN FOR CQS RESTART FOR STRUCTURE I#$#MSGQ
*095 CQS0032A ENTER CHECKPOINT LOGTOKEN FOR CQS RESTART FOR STRUCTURE I#$#EMHQ
R 94,COLD
R 95,COLD
Figure 19-38 CQS requesting the type of start
IMS full function database recovery messages

Any IMS updates to full function IMS databases at the time of the error will be automatically
backed out, as shown in Figure 19-39. In cases like this, it is advisable to ask DBAs to confirm
that the databases are validated for any error messages.
DFS2500I
DFS2500I
DFS682I
DFS968I
DFS980I
DFS2500I
DFS682I
DFS968I
DFS968I
DFS980I
DATABASE IVPDB1I SUCCESSFULLY ALLOCATED

I#$1
DATABASE IVPDB2
SUCCESSFULLY ALLOCATED
I#$1
BATCH-MSG PROGRAM DFSIVP7 JOB IV3H211J MAY BE RESTARTED FROM CHKPT ID
DBD=IVPDB2
WITHIN PSB=DFSIVP7 SUCCESSFULLY BACKED OUT
I#$1
3:08:35 BACKOUT PROCESSING HAS ENDED FOR DFSIVP7
I#$1
DATABASE IVPDB1
SUCCESSFULLY ALLOCATED
I#$1
BATCH-MSG PROGRAM DFSIVP6 JOB IV3H210J MAY BE RESTARTED FROM CHKPT ID
DBD=IVPDB1
WITHIN PSB=DFSIVP6 SUCCESSFULLY BACKED OUT
I#$1
DBD=IVPDB1I WITHIN PSB=DFSIVP6 SUCCESSFULLY BACKED OUT
I#$1
3:08:35 BACKOUT PROCESSING HAS ENDED FOR DFSIVP6
I#$1
Figure 19-39 Messages indicating IMS batch backout has occurred for full function IMS databases
IMS fast path database recovery messages

If IMS has defined any fast path DEDBs using the Virtual Storage Option (VSO), then in an
IMSplex, they will be loaded into the Coupling Facility.
When the IMS system is restarted, the messages found in Figure 19-40 on page 443 are
issued. They indicate that:
The data sets are allocated.
Area has reconnected to the VSO structure.
Preopen/preload processing continues.
Depending upon the status of the system at the time, the VSO DEDBs may require forward
recovery, because committed updates may have been written to the IMS logs, but may not
442
have been updated on the DASD version of the databases. Again, consult your DBA for
validation and advice.
DFS980I 3:08:35 BACKOUT PROCESSING HAS ENDED FOR DFSIVP6
I#$1
DFS2500I DATASET DFSIVD31 SUCCESSFULLY ALLOCATED I#$1
DFS2500I DATASET DFSIVD31 SUCCESSFULLY ALLOCATED
I#$1
DFS2500I DATASET DFSIVD32 SUCCESSFULLY ALLOCATED I#$1
DFS2500I DATASET DFSIVD32 SUCCESSFULLY ALLOCATED
I#$1
DFS2822I AREA DFSIVD3A CONNECT TO STR: I#$#VSO1DB1
SUCCESSFUL I#$1
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#VSO1DB1 981
WAS SUCCESSFUL. JOBNAME: I#$1CTL ASID: 0047
CONNECTOR NAME: I#$1 CFNAME: FACIL01
STRUCTURE I#$#VSO1DB1, CONNECTOR NAME I#$1
CFNAME
-------- --------------------------------FACIL01
FACIL02
SUCCESSFUL I#$1
DFS2823I AREA DFSIVD3A DISCONNECT FROM STR: I#$#VSO1DB1
SUCCESSFUL I#$1
SUCCESSFUL
I#$1
....
DFS3715I DEDB AREA PREOPEN PROCESS STARTED, RSN=00 I#$1

DFS3715I DEDB AREA PREOPEN PROCESS STARTED, RSN=00 I#$1
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#VSO1DB1 010
WAS SUCCESSFUL. JOBNAME: I#$1CTL ASID: 0047
CONNECTOR NAME: I#$1 CFNAME: FACIL01
SUCCESSFUL I#$1
SUCCESSFUL I#$1
DFS3719I DEDB AREA PREOPEN PROCESS COMPLETED, RSN=00 I#$1
DFS2821I PRELOAD COMPLETED FOR ALL SHARED VSO AREAS I#$1
DFS3719I DEDB AREA PREOPEN PROCESS COMPLETED, RSN=00 I#$1
DFS2821I PRELOAD COMPLETED FOR ALL SHARED VSO AREAS I#$1
DFS994I EMERGENCY START COMPLETED. I#$1
Figure 19-40 VSO DEDB messages during IMS restart
19.9.9 Complete processor failures

In this case, the Coupling Facilities remained active, but all systems within the Parallel
Sysplex failed, along with the IMS systems that were running.
Following the restart, the IRLM, SCI, OM, RM address spaces all restarted normally.
CQS address space restarted

CQS was automatically restarted by IMS, and the additional messages shown in Figure 19-41
on page 444 shows how the log structures were used to recreate the environment during
restart.
443
CQS0353I CQS LOG READ STARTED FROM LOGTOKEN 00000000000F9B5D 483

LOG #@$#.SQ.MSGQ.LOG
STRUC I#$#MSGQ
S#$1CQS
CQS0353I CQS LOG READ COMPLETED, LOG RECORD COUNT 138 484
LOG #@$#.SQ.MSGQ.LOG
STRUC I#$#MSGQ
S#$1CQS
CQS0030I SYSTEM CHECKPOINT COMPLETE, STRUCTURE I#$#MSGQ
, LOGTOKEN 000000
CQS0353I CQS LOG READ STARTED FROM LOGTOKEN 0000000000079385 487
LOG #@$#.SQ.EMHQ.LOG
STRUC I#$#EMHQ
S#$1CQS
CQS0353I CQS LOG READ COMPLETED, LOG RECORD COUNT 32 488
LOG #@$#.SQ.EMHQ.LOG
STRUC I#$#EMHQ
S#$1CQS
CQS0030I SYSTEM CHECKPOINT COMPLETE, STRUCTURE I#$#EMHQ
, LOGTOKEN 000000
CQS0020I CQS READY S#$1CQS
, LOGTOKEN 000000
, LOGTOKEN 000000
Figure 19-41 CQS restart messages following a complete processor failure
IMS control region

IMS will require an /ERE OVERRIDE. command to be manually entered following this system
failure. Otherwise, IMS will recover all active tasks and databases normally.
19.9.10 Recovering from an IRLM failure.

If one of the IRLM address spaces abends, then the following sequence would occur.
Abending IRLM
The messages that the abending IRLM will display as it shuts down are shown in
Figure 19-42. The statistics have been suppressed due to their size, but they are basically a
hex dump of the lock structure at the time IRLM abended.
DXR122E IR#I001 ABEND UNDER IRLM TCB/SRB IN MODULE DXRRL020 ABEND CODE=Sxxx
...(statistics suppressed)
INFO: 00010032 00000000 00000000 00000000 00000000 00000004
DXR121I IR#I001 END-OF-TASK CLEANUP SUCCESSFUL - HI-CSA
457K - HI-ACCT-CSA
Figure 19-42 Messages of interest from abending IRLM
Other IRLM
The other IRLM address spaces in the IRLM group will receive similar error messages, as
444

...(statistics suppressed)
IXL020I CLEANUP FOR LOCK STRUCTURE I#$#LOCK1, 962
CONNECTION ID 01, STARTED BY CONNECTOR I#$#$$$$$IR#I003
INFO: 0001 00010032 0000003A
IXL021I GLOBAL CLEANUP FOR LOCK STRUCTURE I#$#LOCK1, 963
CONNECTION ID 01, BY CONNECTOR I#$#$$$$$IR#I003
HAS COMPLETED.
INFO: 00000000 00000000 00000000 00000000 00000000 00000000
IXL022I LOCAL CLEANUP FOR LOCK STRUCTURE I#$#LOCK1, 964
CONNECTION ID 01, BY CONNECTOR I#$#$$$$$IR#I003
HAS COMPLETED.
INFO: 00000000 00000000 00000000 00000000 00000000 00000000
DXR137I IR#I003 GROUP STATUS CHANGED. IR#I 001 HAS BEEN DISCONNECTED FROM THE
DATA SHARING GROUP
Figure 19-43 Messages when another IRLM in the group has failed
IMS with abending IRLM

The IMS system attached to the abending IRLM receives messages like those shown in
Figure 19-44. The figure also shows the DEDB IVP job IV3H212J timing out with the S522
abend. The other jobs abended with U3303 when allowed to process.
DFS3715I DEDB AREA RLM REVR PROCESS STARTED, RSN=00 I#$1
DFS3705I AREA=DFSIVD3B DD=DFSIVD33 CLOSED I#$1
DFS3705I AREA=DFSIVD3B DD=DFSIVD34 CLOSED I#$1
DFS2500I DATASET DFSIVD33 SUCCESSFULLY DEALLOCATED I#$1
DFS2823I AREA DFSIVD3B DISCONNECT FROM STR: I#$#VSO1DB2
SUCCESSFUL I#$1
DFS2574I AREA=DFSIVD3B STOPPED
I#$1
DFS3719I DEDB AREA RLM REVR PROCESS COMPLETED, RSN=00 I#$1
DFS0535A HSSP DISC PROCESS ATTEMPTED AREA DFSIVD3A I#$1
DFS0535I RC=03, AREA LOCK FAILED. I#$1
DFS554A IV3H212J 00003 G
DFSIVP8 (2)
522,0000 PSB
DFS552I BATCH
REGION IV3H212J STOPPED ID=00003 TIME=0228 I#$1
DFS3705I AREA=DFSIVD3A DD=DFSIVD31 CLOSED I#$1
DFS3705I AREA=DFSIVD3A DD=DFSIVD32 CLOSED I#$1
SUCCESSFUL I#$1
DFS2574I AREA=DFSIVD3A STOPPED
I#$1
Figure 19-44 Messages in the IMS control region for an IRLM abend
IRLM status
The status of other IRLM address spaces following this failure shows IMS1 now in an SFAIL
status, as shown in Figure 19-45 on page 446. This means that the IRLM that IMS is
identified to has been disconnected from the data sharing group. Any modify-type locks held
by IMS have been retained by IRLM.
445

NAME
STATUS
RET_LKS
FDRI#$1 UP-RO
0
FDRI#$2 UP-RO
0
I#$1
SFAIL
8
I#$2
UP
0
I#$3
UP
0
IRLMID
002
003
001
002
003
IRLM_NAME
IR#I
IR#I
IR#I
IR#I
IR#I
IRLM_LEVL
1.009
1.009
1.009
1.009
1.009
Figure 19-45 Displaying IRLM status ALLD after the IRLM failure
Restarting IRLM
Restart IRLM normally.
Reconnect IMS to IRLM

Reconnect the IMS by using the command F imsstc,RECONNECT, as shown in Figure 19-46.
F I#$1CTL,RECONNECT
DFS626I - IRLM RECONNECT COMMAND SUCCESSFUL. I#$1
Figure 19-46 Response from IMS IRLM reconnect command
The IRLM address space rejoins the data sharing group, as shown in Figure 19-47.
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#LOCK1 758
WAS SUCCESSFUL. JOBNAME: I#$#IRLM ASID: 004D
DXR141I IR#I001 THE LOCK TABLE I#$#LOCK1
WAS ALLOCATED IN A VOLATILE
DXR132I IR#I001 SUCCESSFULLY JOINED THE DATA SHARING GROUP WITH
2M LOCK
TABLE LIST ENTRIES
Figure 19-47 IRLM reconnection messages
19.10 IMS startup

This section explains the process needed to start all the components of the IMS sysplex. It
also provides examples showing the messages you can expect within each component.
Which address spaces you require on each system within your IMSplex will vary, depending
on your availability and or workload sharing requirements. In our example, we have started all
possible address spaces on every system.
To start the systems on system1, issue the system commands as shown in Figure 19-48 on
page 447.
446
S
S
S
S
S
S
S
I#$#SCI
I#$#OM
I#$#RM
I#$#IRLM
I#$1CTL
I#$1FDR (possibly on a different system)
I#$1CON
Figure 19-48 #@$1 commands to start IMS I#$#
The SCI, RM, OM, and IRLM address spaces are required to be active before IMS will start.
When starting the IMS Control region, it will automatically start the DLISAS, DBRC, and CQS
address spaces.
The following figures display the messages that indicate the various address spaces have
started successfully.
19.10.1 SCI startup

The SCI address space is required to be active before RM, OM, or IMS will complete
initialization. Figure 19-49 shows the message indicating SCI is active.
CSL0020I SCI READY SCI1SC
Figure 19-49 Message indicating the SCI address space has started
19.10.2 RM startup
The RM address space requires CQS to be active before it completes initialization.
Figure 19-50 shows the messages indicating RM is both waiting for CQS and then active.
CSL0003A RM
CSL0020I RM
WAITING FOR CQS

READY
Figure 19-50 Messages indicating the RM address space has started
19.10.3 OM startup
The OM address space requires SCI to be active before it completes initialization.
Figure 19-51 shows the messages indicating OM is both waiting for SCI and then active.
CSL0003A OM
CSL0020I OM
WAITING FOR SCI

READY
Figure 19-51 Messages indicating OM address space has started
19.10.4 IRLM startup

The IRLM address space is required before IMS will complete initialization. Figure 19-52 on
page 448 shows the messages indicating that IRLM has connected to ARM and the LOCK
structure in the Coupling Facility.
447
DXR117I IR#I001 INITIALIZATION COMPLETE

DXR172I IR#I001 I#$#IR#I001 ARM READY COMPLETED. 307
MVS ARM RETURN CODE = 00000000,
MVS ARM REASON CODE = 00000000.
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#LOCK1 654
DXR141I IR#I001 THE LOCK TABLE I#$#LOCK1
WAS ALLOCATED IN A VOLATILE
DXR132I IR#I001 SUCCESSFULLY JOINED THE DATA SHARING GROUP WITH
2M LOCK
TABLE LIST ENTRIES
Figure 19-52 Messages indicating IRLM address space has initialized
19.10.5 IMSCTL startup

The IMS Control region requires DLI, DBRC, SCI, RM, OM, CQS and IRLM to all be active
before IMS initialization will complete. Figure 19-53 shows the messages that indicate that
IMS is waiting on the other address spaces, as well as those that indicate that IMS restart has
completed.
DFS3306A CTL REGION WAITING FOR SCI
DFS227A - CTL REGION WAITING FOR DLS REGION (I#$1DLI ) INIT
DFS3306A CTL REGION WAITING FOR RM
DFS3306A CTL REGION WAITING FOR OM
DFS0226A CTL REGION WAITING FOR CQS (I#$1CQS ), RESPONSE TO CONNECT REQUEST
*002 DFS3139I IMS INITIALIZED, AUTOMATIC RESTART PROCEEDING
*003 DFS039A IR#I NOT ACTIVE. REPLY RETRY, CANCEL, OR DUMP.
DFS994I WARM START COMPLETED
DFS2360I 00:01:22 XCF GROUP JOINED SUCCESSFULLY.
Figure 19-53 Messages indicating the IMS control region address space initialization
The automated start of OM and SCI address spaces is controlled by the parameters in
DFSCGxxx member in IMS.PROCLIB. The parameters are as follows:
448
RMENV=
Y is the default and indicates that an RM address space is not

required. RMENV=Y does not allow the control region to automatically
start the SCI or OM address spaces.
SCIPROC=
This parameter is used to specify the procedure name for the SCI
address space, which IMS will automatically start if not already
started. This will only occur if RMENV=N is also specified.
OMPROC=
This parameter is used to specify the procedure name for the OM

address space, which IMS will automatically start if not already
started. This will only occur if RMENV=N is also specified.
Attention: In Figure 19-53, the region waiting messages for SCI, RM, OM, and CQS are
not highlighted and could be lost with all the other messages produced at IMS startup, but
as soon as these address spaces are active, IMS will continue automatically.
If the DFS039A message waiting on IRLM WTOR appears, then operations or automation
will have to respond RETRY to the message before IMS will continue.
If an IMS automatic (AUTO=Y) restart is done, then the IMS WTOR DFS3139I will appear.
As soon as this is used, then the IMS WTOR DFS996I *IMS READY* message will appear.
If an IMS manual (AUTO=N) restart is done, then only the IMS WTOR DFS996A message
will appear.
19.10.6 DLISAS startup

The DLISAS address space is automatically started by IMS (and is required even without an
IMS sysplex). Figure 19-54 shows the DLISAS initialization messages, as well as the
allocation and connections to the OSAM and VSAM structures in the Coupling Facility.
DFS228I - DLS RECALL TCB INITIALIZATION COMPLETE I#$1
DFS228I - DLS REGION STORAGE COMPRESSION INITIALIZED I#$1
DFS228I - DLS REGION DYNAMIC ALLOCATION INITIALIZED I#$1
DFS3386I OSAM CF CACHING RATIO= 050:001, 2 I#$1
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#OSAM 558
STRUCTURE I#$#OSAM, CONNECTOR NAME IXCLO0180001
CFNAME
---------------------------------------FACIL02
FACIL01
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#VSAM 565
STRUCTURE I#$#VSAM, CONNECTOR NAME IXCLO0170001
CFNAME
---------------------------------------FACIL02
FACIL01
DFS3382I DL/I CF INITIALIZATION COMPLETE I#$1
DFS228I - DLS REGION INITIALIZATION COMPLETE I#$1
DFS2500I DATABASE DI21PART SUCCESSFULLY ALLOCATED
I#$1
Figure 19-54 Messages indicating DLISAS has initialized
19.10.7 DBRC startup

The DBRC address space is automatically started by IMS (and is required even without an
IMS sysplex). Figure 19-55 shows the message indicating that DBRC has completed
initialization.
DFS3613I - DRC TCB INITIALIZATION COMPLETE
Figure 19-55 Messages indicating DBRC has initialized
449
19.10.8 CQS startup

The CQS address space is started automatically by IMS. Figure 19-56 shows the messages
when CQS is started, and the allocation and connections it makes with both the MSGQ and
EMHQ structures.
QS0008W STRUCTURE I#$#MSGQ
IS VOLATILE; CONSIDER STRUCTURE CHECKPOINT
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#MSGQ 548
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#EMHQ 551
CQS0008W STRUCTURE I#$#EMHQ
CQS0008W STRUCTURE I#$#RM
IS VOLATILE; S#$1CQS
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#RM 554
CIXL014I IXLCONN REQUEST FOR STRUCTURE I#$#MSGQOFLW 561
STRUCTURE I#$#MSGQOFLW, CONNECTOR NAME CQSS#$1CQS
CFNAME
-------- --------------------------------FACIL01
FACIL02
CQS0008W STRUCTURE I#$#MSGQOFLW
IXL014I IXLCONN REQUEST FOR STRUCTURE I#$#EMHQOFLW 582
STRUCTURE I#$#EMHQOFLW, CONNECTOR NAME CQSS#$1CQS
CFNAME
-------- --------------------------------FACIL01
FACIL02
CQS0008W STRUCTURE I#$#EMHQOFLW
, LOGTOKEN nnnn...
, LOGTOKEN nnnn...
CQS0020I CQS READY S#$1CQS
, LOGTOKEN nnnn...
, LOGTOKEN nnnn...
CQS0220I CQS S#$1CQS STARTED STRUCTURE CHECKPOINT FOR STRUCTURE I#$#EMHQ
CQS0220I CQS S#$1CQS STARTED STRUCTURE CHECKPOINT FOR STRUCTURE I#$#MSGQ
CQS0200I STRUCTURE I#$#MSGQ
QUIESCED FOR STRUCTURE CHECKPOINT S#$1CQS
CQS0200I STRUCTURE I#$#EMHQ
QUIESCED FOR STRUCTURE CHECKPOINT S#$1CQS
CQS0201I STRUCTURE I#$#MSGQ
RESUMED AFTER STRUCTURE CHECKPOINT S#$1CQS
CQS0201I STRUCTURE I#$#EMHQ
RESUMED AFTER STRUCTURE CHECKPOINT S#$1CQS
, LOGTOKEN nnnn...
, LOGTOKEN nnnn...
CQS0221I CQS S#$1CQS COMPLETED STRUCTURE CHECKPOINT FOR STRUCTURE I#$#MSGQ
CQS0221I CQS S#$1CQS COMPLETED STRUCTURE CHECKPOINT FOR STRUCTURE I#$#EMHQ
Figure 19-56 CQS has initialized and connected
The messages in Figure 19-56 indicate that CQS has initialized and connected with the
Shared Queues structures for MSGQ and EMHQ.
450
19.10.9 FDBR startup

The optional FDBR address space would typically be started on a system different from the
IMS system it is monitoring. Figure 19-57 shows the message indicating that FDBR has
completed initialization.
For FDR to complete initialization, the SCI and OM address spaces needs to be active on the
system. If they are not active, then FDR will wait for them.
DFS3306A CTL REGION WAITING FOR SCI - F#$1
DFS3306A CTL REGION WAITING FOR OM - F#$1
DFS4161I FDR FOR (I#$1) TRACKING STARTED
Figure 19-57 Messages indicating the FDBR address space initialization
If IRLM is not already active on the system where FDR is starting, then the FDR address
space will abend, as shown in Figure 19-58. If this happens, restart IRLM before starting
FDBR.
DFS4179E FDR FOR (I#$1) IRLM IDENT-RO FAILED, RC=08 REASON=4008
DFS629I IMS RST TCB ABEND - IMS 0574
F#$1
F#$1
Figure 19-58 Abend in FDBR if it is started without IRLM
19.10.10 IMS Connect startup

The messages indicating IMS Connect has started are shown in Figure 19-59.
HWSM0590I
HWSD0290I
HWSD0290I
HWSD0290I
HWSS0780I
HWSS0790I
HWSS0790I
HWSS0790I
HWSC0010I
CONNECTED TO IMSPLEX=I#$#
CONNECTED TO DATASTORE=I#$1
; M=DSC1
; M=DSC1
; M=DSC1
TCPIP COMMUNICATION ON HOSTNAME=TCPIP
OPENED; M=
LISTENING ON PORT=7101
STARTED; M=SDOT
STARTED; M=SDOT
STARTED; M=SDOT
WELCOME TO IMS CONNECT!
Figure 19-59 IMS Connect startup messages
If SCI is not active on the system when IMS Connect starts, it will still connect to the IMS
systems, but will also receive the message shown in Figure 19-60.
HWSI1720W REGISTRATION TO SCI FAILED: MEMBER=I#$1CON
Figure 19-60 SCI failure messages at IMS Connect startup
19.11 IMS shutdown

This section describes the tasks needed to shut down the various address spaces used in an
IMS sysplex, and the messages that you can expect to receive.
451
19.11.1 SCI/RM/OM shutdown

The SCI address space can be shut down along with all the other SCI, OM, and RM address
spaces within the IMSplex with the single command F I#$#SCI,SHUTDOWN CSLPLEX.
19.11.2 IRLM shutdown

The IRLM address space can be shut down with the command C I#$#IRLM.
19.11.3 IMSCTL shutdown

The IMS control region can be shut down with the IMS command /CHE DUMPQ or /CHE FREEZE,
which shuts down the DLISAS, DBRC, and FDBR regions automatically.
FDBR shutdown
If the IMS control region is shut down, then FDBR will be shut down. However, if you only
want FDBR to shut down, then use the command F I#$1FDR,TERM.
19.11.4 CQS shutdown

If the CQS command /CQSET SHUTDOWN SHAREDQ ON STRUCTURE ALL has been issued, then it
means that CQS will always shut down as part of IMS shut down, as described in CQS Set
command on page 428.
If the command /CQSET SHUTDOWN SHAREDQ OFF STRUCTURE ALL has been issued, then CQS
will remain active when IMS is shut down. In this case, to shut down CQS, issue the
command P cqsname.
If P cqsname is issued and CQS is not shut down but the CQS0300I message shown in
Figure 19-61 is displayed, it means that there were clients still connected to the CQS. The
/DIS CQS command can be used to identify an IMS connected to CQS. However, currently
there is no way to easily determine if RM is connected to CQS and as such, it is difficult to find
why this message is being received.
CQS0300I MVS STOP COMMAND REJECTED, RC=01000004
Figure 19-61 CQS message indicating it cannot stop while clients are connected
19.11.5 IMS Connect shutdown

IMS Connect is shut down by responding to the outstanding WTOR HWSC0000I *IMS
CONNECT READY* with the command CLOSEHWS. The messages indicating IMS Connect has
shut down are displayed in Figure 19-62 on page 453.
452
*143 HWSC0000I *IMS CONNECT READY* I#$1

R 143,CLOSEHWS
HWSS0770I LISTENING ON PORT=7101
TERMINATED; M=SSCH
TERMINATED; M=SSCH
TERMINATED; M=SSCH
HWSS0781I TCPIP COMMUNICATION FUNCTION CLOSED; M=SOCC
HWSD0260I DS=I#$1
TRANSMIT THREAD TERMINATED; M=DXMT
HWSD0260I DS=I#$1
RECEIVE THREAD TERMINATED; M=DREC
HWSD0260I DS=I#$2
HWSD0260I DS=I#$2
HWSD0260I DS=I#$3
HWSD0260I DS=I#$3
HWSM0560I IMSPLEX=I#$#
TRANSMIT THREAD TERMINATED; M=OXMT
HWSM0560I IMSPLEX=I#$#
RECEIVE THREAD TERMINATED; M=OREC
CLOSED; M=DSCL
CLOSED; M=DSCL
*144 HWSC0000I *IMS CONNECT READY* I#$1
CLOSED; M=DSCL
HWSM0582I COMMUNICATION WITH IMSPLEX=I#$#
CLOSED; M=DSCL
HWSM0580I IMSPLEX COMMUNICATION FUNCTION CLOSED; M=DOC3
HWSC0020I IMS CONNECT IN TERMINATION
BPE0007I HWS BEGINNING PHASE 1 OF SHUTDOWN
BPE0008I HWS BEGINNING PHASE 2 OF SHUTDOWN
BPE0009I HWS SHUTDOWN COMPLETE
Figure 19-62 Messages indicating IMS Connect has shut down
19.12 Additional information

For additional information about IMS, refer to:
IMS home page:
http://www.ibm.com/ims/
IBM Redbooks publication IMS Primer, SG24-5352
IBM Press publication An Introduction to IMS, 2004, ISBN 0131856715
453
454
20
Chapter 20.
WebSphere MQ
This chapter provides an overview of various operational considerations to keep in mind
when WebSphere MQ is implemented in a Parallel Sysplex.
455
20.1 Introduction to WebSphere MQ

WebSphere MQ is a subsystem used for the transport of data between applications. The
applications communicate with each other and can be active on the same system, on
different systems, or on different platforms altogether. This will be totally transparent to the
application.
MQ transports the data between applications in the form of a message, which is a string of
bytes meaningful to the application that uses it.
WebSphere MQ messages have two parts:
Application data
The content and structure of the application data is defined by the application programs
that use them.
Message descriptor
The message descriptor identifies the message and contains additional control
information, such as the type of message and the priority assigned to the message by the
sending application.
When one application wants to send data to another application, it delivers the message to a
part of MQ called a queue. A queue is a data structure used to store messages until they are
retrieved by an application. The messages typically get removed from the queue when the
receiving application asks the queue manager to receive a message from the named queue.
The queue manager owns and manages the set of resources that are used by WebSphere
MQ, which includes:
Page sets that hold the WebSphere MQ object definitions and message data
Logs that are used to recover messages and objects in the event of queue manager
failure
Processor storage
Connections through which different application environments such as CICS, IMS, and
Batch can access the WebSphere MQ API
The WebSphere MQ channel initiator, which allows communication between
WebSphere MQ on your z/OS system and other systems
Figure 20-1 on page 457 shows the channel initiator and the queue manager with
connections to different application environments.
456
Figure 20-1 Relationship between application, channel initiator and queue managers
If the queue to which the message is sent is not on the same system as the sender
application, another part of MQ is used to transport the message from the local system to the
remote system. The channel initiator is responsible for transporting a message from one
queue manager to another using a transmission protocol such as TCP/IP or SNA.
Channel initiator code runs on a z/OS system as a started task named xxxxCHIN.
Queue manager code runs on a z/OS system as a started task named xxxxMSTR, with xxxx
being the respective subsystem ID.
Chapter 20. WebSphere MQ
457
Figure 20-2 Queue managers within a sysplex
Figure 20-2 displays two queue managers within a sysplex. Each queue manager has a
channel initiator and a local queue. Messages sent by queue managers on AIX and Windows
are placed on the local queue, from where they are retrieved by an application. Reply
messages are returned via a similar route.
When MQ is running in a Parallel Sysplex, the need may arise to access a queue from more
than one queue manager due to workload management and availability requirements. This is
where shared queues fit in.
A shared queue is a type of local queue. The messages on that queue can be accessed by
one or more queue managers that are in a sysplex. The queue managers that can access the
same set of shared queues form a group called a queue-sharing group.
Any queue manager in the queue-sharing group can access a shared queue. This means that
you can put a message on to a shared queue on one queue manager, and get the same
message from the queue from a different queue manager. This provides a rapid mechanism
for communication within a queue-sharing group that does not require channels to be active
between queue managers.
458
Figure 20-3 Queue-sharing group
Figure 20-3 displays three queue managers and a Coupling Facility, which form a
queue-sharing group. All three queue managers can access the shared queue in the
Coupling Facility.
An application can connect to any of the queue managers within the queue-sharing group.
Because all the queue managers in the queue-sharing group can access all the shared
queues, the application does not depend on the availability of a specific queue manager; any
queue manager in the queue-sharing group can service the queue.
At least two CF structures are needed for shared queues. One is the administrative structure.
The administrative structure contains no queues or messages. It only contains internal queue
manager information and has a fixed name of queuesharinggroupCSQ_ADMIN.
Subsequent structures are used for queues and messages. Up to 63 structures can be
defined to contain queues or messages for a particular queue-sharing group. The names of
these structures are elective, but the first four characters must be the queue- sharing group
name.
Queue-sharing groups have a name of up to four characters. The name must be unique in
your network, and be different from any queue manager names.
Figure 20-4 on page 460 illustrates a queue-sharing group that contains two queue
managers. Each queue manager has a channel initiator and its own local page sets and log
data sets. Each member of the queue-sharing group must also connect to a DB2 system.
The DB2 systems must all be in the same DB2 data-sharing group so that the queue
managers can access the DB2 shared repository, which contains shared object definitions.
These are any type of WebSphere MQ object (for example, a queue or channel) that is
defined only once so any queue manager in the group can use them.
459
Figure 20-4 Queue-sharing group with two queue managers
After a queue manager joins a queue-sharing group, it will have access to the shared objects
defined for that group. You can use that queue manager to define new shared objects within
the group. If shared queues are defined within the group, you can use this queue manager to
put messages to and get messages from those shared queues.
Any queue manager in the group can retrieve the messages held on a shared queue. You
can enter an MQSC command once, and have it executed on all queue managers within the
queue-sharing group as though it had been entered at each queue manager individually. The
command scope attribute is used for this.
20.2 Sysplex considerations

In a z/OS Parallel Sysplex operating system, images communicate using a Coupling Facility.
WebSphere MQ can use the facilities of the sysplex environment for enhanced availability.
Removing the affinities between a queue manager and a particular z/OS image allows a
queue manager to be restarted on a different z/OS image in the event of an image failure.
The restart mechanism can be manual, ARM, or system automation, if you ensure the
following:
All page sets, logs, bootstrap data sets, code libraries, and queue manager configuration
data sets are defined on shared volumes.
The subsystem definition has sysplex scope and a unique name within the sysplex.
The level of early code installed on every z/OS image at IPL time is at the same level.
Virtual IP addresses (VIPA) are available on each TCP stack in the sysplex, and you have
configured WebSphere MQ TCP listeners and inbound connections to use VIPAs rather
than default host names.
460
You can additionally configure multiple queue managers running on different operating
system images in a sysplex to operate as a queue-sharing group, which can take advantage
of shared queues and shared channels for higher availability and workload balancing.
20.3 WebSphere MQ online monitoring

Monitoring of WebSphere MQ is achieved with an ISPF interface and MQ commands.
20.4 MQ ISPF panels

MQ provides an ISPF interface to allow the display, creation, and manipulation of MQ objects;
see Figure 20-5.
IBM WebSphere MQ for z/OS - Main Menu
Complete fields. Then press Enter.
Action
. . . . . . . . . . 1
0.
1.
2.
3.
List with filter

List or Display
Define like
Alter
+
4.
5.
6.
7.
Manage
Perform
Start
Stop
Object type . . . . . . . . QUEUE

Name . . . . . . . . . . . *
Disposition . . . . . . . . A Q=Qmgr, C=Copy, P=Private, G=Group,
S=Shared, A=All
Connect name . . . . . .
Target queue manager . .
- connected or
Action queue manager . .
Response wait time . . .
. PSM3
. PSM3
remote
. PSM3
. 5
- local queue manager or group

queue manager for command input
- command scope in group
5 - 999 seconds
Figure 20-5 WebSphere MQ ISPF main menu
You can connect to a queue manager and define the queue managers on which your
requests should be executed by filling in the following three fields:
Connect name: This is the queue manager to which you actually connect.
Target queue manager: With this parameter, you specify on which queue manager you
want to input your request.
Action queue manager: On the action queue manager, this is where the commands are
actually executed.
461
List Queues - PSM3

Type action codes, then press Enter.
1=Display
2=Define like
3=Alter
Name
<> *
CICS01.INITQ
GROUP.QUEUE
GROUP.QUEUE
ISF.CLIENT.SDSF._/%3.REQUESTQ
ISF.MODEL.QUEUE
PSM1
PSM1.XMITQ
PSM2
PSM2.XMITQ
PSM3.DEAD.QUEUE
PSM3.DEFXMIT.QUEUE
PSM3.LOCAL.QUEUE
SHARED.QUEUE
SYSTEM.ADMIN.CHANNEL.EVENT
SYSTEM.ADMIN.CONFIG.EVENT
SYSTEM.ADMIN.PERFM.EVENT
Row 1 of 35
Press F11 to display queue status.
4=Manage
Type
QUEUE
QLOCAL
QLOCAL
QLOCAL
QALIAS
QMODEL
QREMOTE
QLOCAL
QREMOTE
QLOCAL
QLOCAL
QLOCAL
QLOCAL
QLOCAL
QLOCAL
QLOCAL
QLOCAL
Disposition
ALL
PSM3
QMGR
PSM3
COPY
PSM3
GROUP
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
SHARED
QMGR
PSM3
QMGR
PSM3
QMGR
PSM3
Figure 20-6 Display of WebSphere MQ objects and their disposition
20.4.1 WebSphere MQ commands

WebSphere MQ includes the following commands for monitoring the status of MQ objects.
Use the appropriate MQ command prefix:
Display the status of all channels
DISPLAY CHSTATUS(*)
The channel status displayed is RETRYING, which may suggest that the connection to the
CF structure for our queues or messages for this particular queue-sharing group has failed.
CSQM293I -PSM3 CSQMDRTC 1 CHSTATUS FOUND MATCHING REQUEST CRITERIA
CSQM201I -PSM3 CSQMDRTC DIS CHSTATUS DETAILS
CHSTATUS(TO.PSM2)
CHLDISP(PRIVATE)
XMITQ(SYSTEM.CLUSTER.TRANSMIT.QUEUE)
CONNAME(PST2)
CURRENT
CHLTYPE(CLUSSDR)
STATUS(RETRYING)
SUBSTATE( )
STOPREQ(NO)
RQMNAME( )
END CHSTATUS DETAILS
CSQ9022I -PSM3 CSQMDRTC ' DIS CHSTATUS' NORMAL COMPLETION
Figure 20-7 Channel status display
462
Display the status of all queues; this will provide information about the queue and, if the CF
structure is filling up unexpectedly, may help to identify when an application is looping; see
Figure 20-8 for an example.
DISPLAY QSTATUS(*)
-PSM3 DIS QSTATUS(*)
CSQM293I -PSM3 CSQMDRTC 25 QSTATUS FOUND MATCHING REQUEST CRITERIA
CSQM201I -PSM3 CSQMDRTC DIS QSTATUS DETAILS
QSTATUS(CICS01.INITQ)
TYPE(QUEUE)
QSGDISP(QMGR)
END QSTATUS DETAILS
QSTATUS(GROUP.QUEUE)
TYPE(QUEUE)
QSGDISP(COPY)
END QSTATUS DETAILS
QSTATUS(PSM1.XMITQ)
TYPE(QUEUE)
QSGDISP(QMGR)
...
CSQ9022I -PSM3 CSQMDRTC ' DIS QSTATUS' NORMAL COMPLETION
Figure 20-8 Display the status of all queues
The CFSTATUS command displays the current status of all structures including the
administrative structure; see Figure 20-9 for an example. You can display three different types
of status information:
SUMMARY: Gives an overview of the status information.
CONNECT: Shows all members connected to the structure and in the case of a
connection failure, failure information.
BACKUP: Shows backup date and time, RBA information, and the queue manager that
did the backup.
DISPLAY CFSTATUS(A*) TYPE(SUMMARY)
CSQM293I -PSM3 CSQMDRTC 1 CFSTATUS FOUND MATCHING REQUEST CRITERIA
CSQM201I -PSM3 CSQMDRTC DISPLAY CFSTATUS DETAILS
CFSTATUS(APPL01)
TYPE(SUMMARY)
CFTYPE(APPL)
STATUS(ACTIVE)
SIZEMAX(10240)
SIZEUSED(1)
ENTSMAX(2217)
ENTSUSED(35)
FAILTIME( )
FAILDATE( )
END CFSTATUS DETAILS
CSQ9022I -PSM3 CSQMDRTC ' DISPLAY CFSTATUS' NORMAL COMPLETION
Figure 20-9 DISPLAY CFSTATUS output
463
20.5 WebSphere MQ structure management and recovery

This section explains structure management and recovery in more detail. It has been divided
into the following four areas:
Changing the size of an MQ structure

Moving a structure from one CF to another
Recovering MQ structures from a CF failure
Recovering from the failure of a connected system
20.5.1 Changing the size of an MQ structure

You may need to modify the structure size and rebuild it. Perform the following steps using
the appropriate z/OS System Commands:
1. Check the current MQ structure's size and location.
D XCF,STR,STRNAME=mq structure name
3. Extend the structure size using the ALTER command.
SETXCF START,ALTER,STRNM=mq structure name,SIZE=new size
20.5.2 Moving a structure from one CF to another

It may become necessary to move a structure from one CF to another due to load
rebalancing or to empty out the CF so that all structures can be removed prior to CF
maintenance.
To move a MQ structure to another CF using REBUILD, perform the following steps:
1. All activity to the old structure must be temporarily stopped.
a. Check the current MQ structure size, location, and connectors.
b. Check the free space in the new location.
2. A new structure must be allocated in the alternate CF.
a. Perform the rebuild.
SETXCF START,RB,STRNM=mq structure nam,LOC=OTHER
3. All structure data is copied from the old structure to the new structure.
4. All connections are moved from the original structure to the new structure.
5. Activity is resumed.
6. The original structure is deleted.
7. Check the current MQ structure size, location, and connectors.
464
MQ structure rebuilds should normally be performed when there is little or no MQ activity. The
rebuild process is fully supported by MQ, but there is a brief period of time where access to
the shared queues in the structure is denied.
20.5.3 Recovering MQ structures from a CF failure

Be prepared to recover in the event of a CF failure. All MQ queue managers will abend with
code S5C6 and need to be restarted after the CF is made available again. This occurs if they
lose connectivity because they are unable to detect if the other queue managers are still
connected to the CF; it also ensures data integrity. Perform the following steps to recover
from a CF failure:
1. Check the status of the MQ structure.
2. Check the status of the queue-sharing group from all the MQ systems using the
appropriate MQ command prefix:
DIS QMGR,QSGNAME
3. Display CF storage and connectivity information.
D CF,CFNM=CF Name
D XCF,CF,CFNM=CF Name
4. Recover the CF.
5. Move all structures that normally reside there back into the CF.
SETXCF START,REALLOC
6. Start MQ again so that the queue managers can recover from the CF failure.
Recommendation: Use system-managed duplexing on all MQ structures and thereby
avoid system abends in the event of a loss of connectivity related to a single CF failure.
20.5.4 Recovering from the failure of a connected system

The following actions are required for MQ to recover from the failure of an image in a Parallel
Sysplex. The CF structures will be unaffected by this outage but the surviving queue
managers on other systems will perform recovery on behalf of the failing queue manager.
Perform the following steps to recover MQ from a system outage:
1. Display the system status to ascertain the failing image.
D XCF,S,ALL
2. Check the status of the queue-sharing group from all the MQ systems using the
appropriate MQ command prefix.
DIS QMGR,QSGNAME
3. Check the status of the MQ structure from any other system in the sysplex.
The number of connectors will have been reduced by 1.
4. IPL the failed image.
5. Start your DB2 subsystem, and then start WebSphere MQ. When MQ shared queues use
DB2 tables, DB2 must be started first.
6. MQ will initiate recovery automatically.
465
20.6 WebSphere MQ and Automatic Restart Manager

The Automatic Restart Manager (ARM) is a z/OS recovery function that can improve the
availability of your WebSphere MQ queue managers. ARM improves the time required to
reinstate a queue manager by automatically restarting the batch job or started task (referred
to as an element) when it unexpectedly terminates. ARM preforms this without operator
intervention.
If a queue manager or a channel initiator fails, ARM can restart it on the same LPAR. If z/OS
fails, ARM can restart WebSphere MQ and any related subsystem automatically on another
LPAR within the sysplex.
20.6.1 Verifying the successful registry at startup

Issue the following MVS command to obtain the current ARM status of an element:
D XCF,ARMS,JOBNAME=MQ STC,DETAIL
The results from this command display current statistics such as the first and last ARM restart
time, and that the job is currently available for ARM restarts; see Figure 20-10.
0
1
0
0
0
1
200
PACING :
0
FREECSA:
0
0
ELEMENT NAME :SYSMQMGRPSM3
JOBNAME :PSM3MSTR STATE
:AVAILABLE
CURR SYS :#@$3
JOBTYPE :STC
ASID
:004C
INIT SYS :#@$3
EVENTEXIT:*NONE*
ELEMTYPE:SYSMQMGR LEVEL
:
2
TOTAL RESTARTS :
0
INITIAL START:07/04/2007 23:43:35
RESTART THRESH :
0 OF 0
RESTART TIMEOUT:
300
LAST RESTART:*NONE*
Figure 20-10 ARMSTATUS DETAIL output
Only the queue manager should be restarted by ARM. The channel initiator should be
restarted from CSQINP2 initialization data set. Set up your WebSphere MQ environment so
that the channel initiator and associated listeners are started automatically when the queue
manager is restarted.
For more detailed information about ARM, refer to Chapter 6, Automatic Restart Manager
on page 83.
466
21
Chapter 21.
Resource Recovery Services

This chapter introduces Resource Recovery Services (RRS) and provides an overview of
operational considerations to keep in mind when it is implemented in a Parallel Sysplex.
467
21.1 Introduction to Resource Recovery Services

Resource Recovery Services (RRS) provides a global syncpoint manager that any resource
manager (RM) on z/OS can exploit. It enables transactions to update protected resources
managed by many resource managers.
21.1.1 Functional overview of RRS

Some of the functions performed by RRS are:
Coordinate the two-phase commit process used by the exploiter
The two-phase commit protocol is a set of actions used to make sure that an application
program either makes all changes to the resources represented by a single unit of
recovery (UR), or makes no changes at all. It verifies that either all changes or no changes
are applied even if one of the elements, application, system, or the resource manager,
fails. The protocol allows for restart and recovery processing to take place after system or
subsystem failure.
Create an association between a unit of recovery and a work context
A work context is a representation of a work request (transaction). It may consist of a
number of units of recovery.
Preserve the UR state across all failures
Exploit the system logger for recovery logs
RRS runs in its own address space, which should be started at IPL time.
Figure 21-1 displays the association between application, resource manager, synchpoint
manager, and RRS logstreams.
Figure 21-1 RRS overview
468
21.2 RRS exploiters

There are many exploiters of RRS, each having its own resource manager (RM). Within the
RM there are three distinct types: data managers, communication managers, and work
managers, as explained here.
21.2.1 Data managers

The data managers are DB2, IMS DB, and VSAM.
Data managers allow an application to read and change data. To process a syncpoint event,
a data resource manager would take actions such as committing or backing out changes to
the data it manages.
21.2.2 Communication managers

The communication managers are APPC, TRPC, and WebSphere MQ.
Communication managers control access to distributed resources and act as an extension to
the synchpoint manager. A communications resource manager provides access to distributed
resources by allowing an application to communicate with other applications and resource
managers, possibly located on different systems. It acts as an extension to the syncpoint
manager by allowing the local syncpoint manager to communicate with other syncpoint
managers as needed to ensure coordination of the distributed resources the application
accesses.
21.2.3 Work managers

The work managers are IMS, CICS, DB2 Stored Procedure, and WebSphere for z/OS.
Work managers are resource managers that control applications access to system
resources. To process a synchpoint event, a work manager might ensure that the application
is in the correct environment to allow the synchpoint processing to continue.
IMS currently provides the ability to disable RRS if desired.
21.3 RRS logstream types

Table 21-1 lists all the logstream types available to RRS. ARCHIVE and METADATA are
optional logstreams.
Table 21-1 RRS logstream types
Log stream type
Log stream name
Description
RRS Archive log
ATR.grpname.ARCHIVE
An optional logstream that contains information about completed

URs. Useful when you are trying to trace the history of a problem.
RRS main UR
state log
ATR.grpname.MAIN.UR
Contains information about active URs, including URs that are

active but have been delayed. Useful when you are trying to find
which transaction caused a problem.
RRS RM data log
ATR.grpname.RM.DATA
Contains information about the resource managers that are

currently using RRS.
Chapter 21. Resource Recovery Services
469
Log stream type
Log stream name
Description
RRS delayed UR
state log
ATR.grpname.DELAYED.
UR
Contains information about the state of active URs when UR

completion is delayed.
RRS restart log
ATR.grpname.RESTART
Contains information about incomplete URs that resource

managers might need after a system or RRS failure. Enables a
functioning RRS instance to take over incomplete work left over
from an RRS instance that failed.
RRS Metadata log
ATR.grpname.METADATA
An optional logstream that contains information created by a

resource manager for its own use.
If the ARCHIVE logstream is defined and write activity to the RRS ARCHIVE log stream is
high, this may impact the performance throughput of all RRS transactions actively in use by
RRS. This log stream is optional and only needed by the installation for any post-transaction
history type of investigation. If you choose not to use the ARCHIVE log stream, a warning
message is issued at RRS startup time about not being able to connect to the log stream,
RRS, however, will continue its initialization process.
21.4 Starting RRS

To invokes the RRS procedure and create the RRS address space, issue the following
system command:
START RRS,SUB=MSTR
If you have created a different procedure for starting RRS, then use that member name.There
is a sample start procedure available in SYS1.SAMPLIB(ATRRRS).
Warm start
The normal mode of operation is a warm start. This occurs when valid data is found in the
RM.DATA log stream. For RRS to access data about incomplete transactions, all defined
RRS log streams should be intact. METADATA and ARCHIVE are optional log streams and
do not need to be defined. Figure 21-2 on page 471 displays the messages you would expect
after an RRS warm start.
470
ATR221I RRS IS JOINING RRS GROUP #@$#PLEX ON SYSTEM #@$3

IXL014I IXLCONN REQUEST FOR STRUCTURE RRS_RMDATA_1 840
WAS SUCCESSFUL. JOBNAME: IXGLOGR ASID: 0016
CONNECTOR NAME: IXGLOGR_#@$3 CFNAME: FACIL01
IXL014I IXLCONN REQUEST FOR STRUCTURE RRS_MAINUR_1 841
IXL014I IXLCONN REQUEST FOR STRUCTURE RRS_DELAYEDUR_1 842
IXL014I IXLCONN REQUEST FOR STRUCTURE RRS_RESTART_1 843
IXG231I IXGCONN REQUEST=CONNECT TO LOG STREAM ATR.#@$#PLEX.ARCHIVE
DID 844
NOT SUCCEED FOR JOB RRS. RETURN CODE: 00000008 REASON CODE: 0000080B
DIAG1: 00000008 DIAG2: 0000F801 DIAG3: 05030004 DIAG4: 05020010
ATR132I RRS LOGSTREAM CONNECT HAS FAILED FOR 845
OPTIONAL LOGSTREAM ATR.#@$#PLEX.ARCHIVE.
RC=00000008, RSN=0000080B
IXG231I IXGCONN REQUEST=CONNECT TO LOG STREAM ATR.#@$#PLEX.RM.METADATA
846
DID NOT SUCCEED FOR JOB RRS. RETURN CODE: 00000008 REASON CODE:
0000080B DIAG1: 00000008 DIAG2: 0000F801 DIAG3: 05030004 DIAG4:
05020010
ATR132I RRS LOGSTREAM CONNECT HAS FAILED FOR 847
OPTIONAL LOGSTREAM ATR.#@$#PLEX.RM.METADATA.
RC=00000008, RSN=0000080B
ASA2011I RRS INITIALIZATION COMPLETE. COMPONENT ID=SCRRS
Figure 21-2 Typical messages produced after a RRS warm start in a Parallel Sysplex
Note the error messages for the 1 ARCHIVE and 2 METADATA logstreams.
Cold start
When RRS finds an empty RM.DATA log stream, it cold starts. RRS flushes any log data
found in the MAIN.UR and DELAYED.UR log streams to the ARCHIVE log, if it exists. An RRS
cold start applies to the entire RRS logging group, which may contain some or all members of
the sysplex. The logstreams are shared across all systems in the sysplex that are in that
logging group. After an RRS cold start, there is no data available to RRS to complete any
work that was in progress. RRS can be cold started by stopping all RRS instances in the
logging group, and deleting and redefining the RM.DATA log stream using the IXCMIAPU
utility.
There is a sample procedure available in SYS1.SAMPLIB(ATRCOLD) which deletes and
then defines the RM.DATA logstream. This forces a cold start when RRS tries to initialize.
RRS should only be deliberately cold-started in very controlled circumstances such as:
The first time that RRS is started
When there is a detected data loss in RM.DATA
For a controlled RRS cold start, all resource managers that require RRS should be stopped
on all systems that are a part of the RRS logging group to be cold started. Use the RRS ISPF
panels to check on resource manager status. Check that no incomplete URs exist for any
resource manager.
471
21.5 Stopping RRS

Use these commands only at the direction of the system programmer. RRS should be
running at all times. Stopping RRS can cause application programs to abend or wait until
RRS is restarted.
SETRRS CANCEL
This will terminate (abend) all incomplete commit and backout requests, then pass the return
codes to the requesting application programs.
SETRRS SHUTDOWN (available only with z/OS V1R8 and above)
This provides a normal shutdown command to bring down RRS without resulting in an X'058'
abend. All the currently active resource managers will be unset. After the unset processing is
completed, the RRS jobstep task and all of its subtasks will be normally terminated to clean
up the address space.
In addition to the RRS infrastructure tasks, there are also timed process tasks and server
tasks running in the RRS address space. These tasks will be shut down normally as well.
Syncpoint processing for the outstanding work will be stopped by unsetting exits of the
resource manager. The resource manager will need to reset its exits and restart with RRS
upon restart.
21.6 Displaying the status of RRS

Using the D RRS,RM,S (available only with z/OS V1R8 and above) command, you can:
Display all the RMs using RRS

Display their current state
Display the system they are operating on
Display the groupname in use
Figure 21-3 shows the output from the D RRS,RM,S command.
ATR602I
23.52.32 RRS RM SUMMARY 148
RM NAME
STATE
CSQ.RRSATF.IBM.PSM2
Reset
DFHRXDM.#@$CWE2A.IBM
Reset
DFHRXDM.#@$C1A2A.IBM
Reset
DFHRXDM.#@$C1T2A.IBM
Reset
DSN.RRSATF.IBM.D#$2
Run
DSN.RRSPAS.IBM.D#$2
Run
SYSTEM
#@$2
#@$2
#@$2
#@$2
#@$2
#@$2
GNAME
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
Figure 21-3 Output from D RRS,RM,S
The following system commands may assist in identifying delays within RRS.
D RRS,UR,S (available only with z/OS V1R8 and above)
472
Figure 21-4 shows output from the D RRS,UR,S command.

ATR601I
04.57.34 RRS UR SUMMARY 114
URID
SYSTEM
C0D8CB4F7E15C0000000000001010000 #@$3
C0D8CB517E15C3740000000001010000 #@$3
C0D8CB607E15C6E80000000001010000 #@$3
C0D8CB617E15CA5C0000000001010000 #@$3
C0D8CB767E15CDD00000000001010000 #@$3
C0D8EA777E15D1440000000001010000 #@$3
GNAME
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
#@$#PLEX
ST
FLT
FLT
FLT
FLT
FLT
FLT
TYPE COMMENTS
Unpr
Unpr
Unpr
Unpr
Unpr
Unpr
Figure 21-4 Output from D RRS,UR,S
D RRS,UR,DETAILED,URID=C0D8CB4F7E15C0000000000001010000
21.7 Display RRS logstream status

Using the D LOGGER,LOGSTREAM,LSN=ATR.* command you can:
Determine the logstream/structure name association

Determine the number of connections
Determine whether staging data sets are being used
Determine the logstream status
Figure 21-5 on page 474 displays the output from the D LOGGER,LOGSTREAM,LSN=ATR.*
command.
473
IXG601I
LOGSTREAM
STRUCTURE
----------------ATR.#@$#PLEX.DELAYED.UR
RRS_DELAYEDUR_1
SYSNAME: #@$2
SYSNAME: #@$1
SYSNAME: #@$3
RRS_MAINUR_1
SYSNAME: #@$2
SYSNAME: #@$1
SYSNAME: #@$3
RRS_RESTART_1
SYSNAME: #@$2
SYSNAME: #@$1
SYSNAME: #@$3
ATR.#@$#PLEX.RM.DATA
RRS_RMDATA_1
SYSNAME: #@$2
SYSNAME: #@$1
SYSNAME: #@$3
NUMBER OF LOGSTREAMS:
#CONN STATUS
------ -----000003 IN USE
1
2
000003 IN USE
000003 IN USE
000003 IN USE
000004
Figure 21-5 Display RRS logstream status
In the figure, 1 displays the association between logstream and CF structure name, the
number of connections (there are three), and the status. As shown in 2, duplexing is taking
place in the IXGLOGR data space.
21.8 Display RRS structure name summary

Using the D XCF,STR command you can:
Display the RRS Structure names
Display when they were allocated
Display their current status
Figure 21-6 on page 475 displays the output of the D XCF,STR command.
474

STRNAME
ALLOCATION TIME
RRS_ARCHIVE_1
--RRS_DELAYEDUR_1 06/27/2007 01:06:59
RRS_MAINUR_1
06/27/2007 01:06:58
RRS_RESTART_1
06/27/2007 01:07:00
RRS_RMDATA_1
06/27/2007 01:06:57
STATUS
NOT ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
ALLOCATED
Figure 21-6 Display RRS structure status summary
21.9 Display RRS structure name detail

The D XCF,STR,STRNM=RRS_RMDATA_1 command provides you with more detailed information
about an individual CF structure, in this case RRS_RMDATA_1.
The command displays:
Whether it is duplexed 1
Alternative CFs 2
In which CF the structure is allocated 3
Its connections 4
Figure 21-7 on page 476 displays the output of the D XCF,STR,STRNM=RRS_RMDATA_1

command.
475

STRNAME: RRS_RMDATA_1
STATUS: ALLOCATED
TYPE: LIST
POLICY INFORMATION:
POLICY SIZE
: 16000 K
FULLTHRESHOLD : 80
ALLOWAUTOALT : NO
REBUILD PERCENT: 5
DUPLEX
: DISABLED
ENFORCEORDER : NO
1
2
ACTIVE STRUCTURE
---------------ALLOCATION TIME: 06/27/2007 01:06:57
CFNAME
: FACIL01
3
ACTUAL SIZE
: 9216 K
ENTRIES: IN-USE:
12 TOTAL:
14021,
ELEMENTS: IN-USE:
39 TOTAL:
14069,
PHYSICAL VERSION: C0CEC7FD 73B897CC
LOGICAL VERSION: C0CEC7FD 73B897CC
DISPOSITION
: DELETE
ACCESS TIME
: 0
MAX CONNECTIONS: 32
# CONNECTIONS : 3
CONNECTION NAME
---------------IXGLOGR_#@$1
IXGLOGR_#@$2
IXGLOGR_#@$3
ID
-03
01
02
VERSION
-------00030046
0001010C
00020053
SYSNAME
-------#@$1
#@$2
#@$3
JOBNAME
-------IXGLOGR
IXGLOGR
IXGLOGR
ASID
---0016
0016
0016
0% FULL
0% FULL
STATE
---------------ACTIVE
ACTIVE
ACTIVE
Figure 21-7 Display RRS structure status in more detail
21.10 RRS ISPF panels

ISPF panels are shipped to allow an installation to work with RRS. They can be used to
troubleshoot RRS problems. Figure 21-8 on page 477 shows the RRS Primary ISPF panel.
In this example, the RRS/ISPF interface is invoked via a REXX EXEC with the assumption
that all the required libraries have already been allocated:
ADDRESS 'ISPEXEC'
"SELECT PANEL(ATRFPCMN) NEWAPPL(RRSP) PASSLIB"
476
Option ===> 1
Select an option and press ENTER:
1
2
3
4
5
6
Browse an RRS log stream

Display/Update RRS related Resource Manager information
Display/Update RRS Unit of Recovery information
Display/Update RRS related Work Manager information
Display/Update RRS UR selection criteria profiles
Display RRS-related system information
Figure 21-8 RRS Primary ISPF panel
After invoking the RRS ISPF primary panel, you are able to display or update the various
logstream types; see Figure 21-9.
RRS Log Stream Browse Selection
Command ===>
Provide selection criteria and press Enter:
Select a log stream to view:

4 1. RRS Archive log
2. RRS Unit of Recovery State logs
3. RRS Restart log
4. RRS Resource Manager Data log
5. RRS Resource Manager MetaData log
Level of report detail:

1
1. Summary
2. Detailed
RRS Group Name . . .

Default Group Name: : #@$#PLEX
Output data set . . ATR.REPORT
Optional filtering:
Entries from . . . .
through
. . . . . .
UR identifier . . .
RM name . . . . . .
SURID
local date in yyyy/mm/dd format

local time in hh:mm:ss format
local date in yyyy/mm/dd format
local time in hh:mm:ss format
(Options 1,2,3)
(Option 4,5)
(Options 1,2,3)
Figure 21-9 RRS logstream browse
The RRS Resource Manager Data log gives details about the RM such as:
The RM name, which will identify the component

When it was last active and on what system
On which systems it may be restarted
The logstream name
Figure 21-10 on page 478 displays more detail about the RRS structure status.
477
Menu Utilities Compilers Help

ssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss
BROWSE
MACNIV.ATR.REPORT
Line 00000000 Col 001 080
Command ===>
Scroll ===> PAGE
********************************* Top of Data *********************************
RRS/MVS LOG STREAM BROWSE SUMMARY REPORT
READING ATR.#@$#PLEX.RM.DATA
LOG STREAM
#@$2
2007/06/27 02:58:49.450178 BLOCKID=00000000000155B9
RESOURCE MANAGER=CSQ.RRSATF.IBM.PSM2
LOGGING SYSTEM=#@$2
RESOURCE MANAGER MAY RESTART ON ANY SYSTEM
RESOURCE MANAGER WAS LAST ACTIVE WITH RRS ON SYSTEM #@$2
LOG NAME IS CSQ3.MQ.RRS.IBM.PSM2
RESTART ANYTIME SUPPORTED
LOG INSTANCE NUMBER: 2007/06/27 06:58:04.069718
#@$2
2007/06/27 02:58:49.451368 BLOCKID=0000000000015691
RESOURCE MANAGER=CSQ.RRSATF.IBM.PSM1
LOGGING SYSTEM=#@$2
RESOURCE MANAGER MAY RESTART ON ANY SYSTEM
RESOURCE MANAGER WAS LAST ACTIVE WITH RRS ON SYSTEM #@$1
LOG NAME IS CSQ3.MQ.RRS.IBM.PSM1
RESTART ANYTIME SUPPORTED
LOG INSTANCE NUMBER: 2007/06/27 06:58:03.066048
Figure 21-10 Output from RRS resource manager data log
21.11 Staging data sets, duplexing, and volatility

Staging data sets are required for DASD only log streams. They are optional for CF
structure-based logstreams and staging data sets.
Whether or not the logstreams are duplexed with staging data sets on disk or in a IXGLOGR
data space will depend on the following:
When STG_DUPLEX(YES) with DUPLEXMODE(UNCOND) is specified in the
LOGSTREAM definition in the LOGR policy, LOGGER will write each transaction to a
staging data set on DASD every time a transaction is written to the CF.
If DUPLEXMODE is CONDITIONAL, LOGGER will check for either condition of
VOLATILITY or FAILURE DEPENDENCE. If either is true, then LOGGER will DUPLEX to
a staging data set as described previously.
If STG_DUPLEX(NO) is specified on the LOGSTREAM definition then LOGGER will write
each transaction to interim storage located in the IXGLOGR data space every time a
transaction is written to the CF.
Your CF is VOLATILE (as opposed to NON-VOLATILE) if it does not have a backup battery
or alternate power source.
Your system is FAILURE DEPENDENT if it shares a power source with the CF. Essentially, a
power failure that affects one system or CF would affect the other. The system is FAILURE
INDEPENDENT when there is no shared power source between the two.
You could potentially be exposed to data loss if:
The system is FAILURE DEPENDENT, the CF is VOLATILE, and the LOGSTREAMs are
not DUPLEXed using staging data sets.
478
If the system is FAILURE DEPENDENT and the CF is NON-VOLATILE or the system is

FAILURE INDEPENDENT and the CF is VOLATILE, you could be exposed to data loss if
you lose or shut down both the system and CF while not DUPLEXing using staging data
sets. The other factor in this scenario that causes the data loss is if any exploiters of
LOGGER still have connections to their LOGSTREAMs.
If the system is FAILURE INDEPENDENT and the CF is NON-VOLATILE, then you do not
need to DUPLEX to limit exposure to the data loss condition.
When RRS suffers a loss of data:
If a loss of data was detected against the RM.DATA logstream a cold start is required for
RRS to successfully initialize. It is strongly recommended that you use unconditional
duplexing for the RM.DATA log because any loss of data, unresolved gap, or permanent
error will force an RRS cold start.
If a loss of data was detected against the RESTART logstream, RRS has already
successfully initialized but any resource manager attempting to RESTART with RRS will
fail with messages ATR212I ATR209I. To allow these resource managers to restart with
RRS, an RRS cold start is necessary.
System Logger normally keeps a second copy of the data written to the CF in a data
space. This provides two copies of the data, so that one copy will always be available in
the event of the failure of either z/OS or the CF. This is satisfactory as long as the CF is
failure-independent (in a separate CPC and non-volatile) from z/OS. If the CF is in the
same CPC as a connected z/OS image, or uses volatile storage, the System Logger uses
DASD data sets, known as staging data sets, to maintain copies of the log stream data
that would otherwise be vulnerable to a failure that impacts both the CF and the z/OS
system.
Although the use of staging data sets is useful from the point of view of guaranteeing the
availability of the data, there is a performance overhead when using them. The response time
from a CF is generally in the order of 100 times faster than the response time from DASD. If
Logger is using staging data sets, which must be on DASD, the requesting task is not told that
the write request is complete until both the write to the CF and the I/O to DASD is finished.
This has an adverse effect on every transaction that is causing data to be written to the
associated logstream.
One method of determining whether staging data sets are in use for a logstream is by using
the command:
D LOGGER,LOGSTREAM,STRNAME=xxxxxxx
Logging functions that use DASD for staging data sets to duplicate data in structures benefit
greatly from CF Duplexing. Using CF Duplexing instead of staging data sets makes
duplication more practical and thus improves availability. Eliminating Staging Data Sets for
the System Logger as a requirement for availability if a CF or Logger structure fails. This is
expected to provide both CPU and response time benefit.
21.12 RRS Health Checker definitions

There is a sample Health Checker procedure which is shipped with z/OS and is available in
SYS1.SAMPLIB member name ATRHZS00. It contains override policies for RRS checks.
For more information about Health Checker, refer to Chapter 12, IBM z/OS Health Checker
on page 257.
479
21.13 RRS troubleshooting using batch jobs

There are sample procedures available in SYS1.SAMPLIB to assist in diagnosing problems
within RRS. Follow the modification instructions in the samples to run it in your own
environment.
ATRBATCH
Produce a readable version of a RRS logstream; Figure 21-11shows an ATRBATCH
sample report.
RRS/MVS LOG STREAM BROWSE DETAIL REPORT
READING ATR.#@$#PLEX.MAIN.UR
LOG STREAM
#@$3
2007/02/25 17:04:21.098521 BLOCKID=0000000000037681
URID=C03647C67E5DBF140000000001020000 LOGSTREAM=ATR.#@$#PLEX.MAIN.UR
PARENT URID=00000000000000000000000000000000
SURID=N/A
WORK MANAGER NAME=#@$3.Q6G4BRK.00BD
STATE=InCommit
EXITFLAGS=00840000 FLAGS=20000000
LUWID=
TID=
GTID=
FORMATID=
GTRID=
(decimal)
(hexadecimal)
BQUAL=
RMNAME=DSN.RRSATF.IBM.D81D
ROLE=Participant
CMITCODE=00000FFF BACKCODE=00000FFF PROTOCOL=PresumeAbort
READING ATR.#@$#PLEX.DELAYED.UR
LOG STREAM
#@$3
2005/06/30 12:12:05.907016 BLOCKID=0000000000000001
URID=BD3D78047E62CBA00000001601020000 LOGSTREAM=ATR.#@$#PLEX.DELAYED.UR
PARENT URID=00000000000000000000000000000000
SURID=N/A
WORK MANAGER NAME=#@$3.PMQ2BRK1.015F
STATE=InPrepare
EXITFLAGS=00040000 FLAGS=20000000
LUWID=
TID=
GTID=
FORMATID=
GTRID=
(decimal)
(hexadecimal)
BQUAL=
RMNAME=DSN.RRSATF.IBM.D61B
ROLE=Participant
CMITCODE=00000FFF BACKCODE=00000FFF PROTOCOL=PresumeNothing
Figure 21-11 ATRBATCH sample report
ATRBDISP
Produce detailed information about every UR known to RRS.
480
21.14 Defining RRS to Automatic Restart Manager

If RRS fails, it can use Automatic Restart Manager (ARM) to restart itself in a different
address space on the same system. RRS, however, will not restart itself following a SETRRS
CANCEL command. To stop RRS and cause it to restart automatically, use the FORCE
command with ARM and ARMRESTART.
To make automatic restart possible, your installation must:
Provide an ARM couple data set that contains, either explicitly or through defaults, an
ARM policy for RRS. When setting up your ARM policy, use the element name
SYS_RRS_sysname for RRS.
Activate the ARM couple data set through a COUPLExx parmlib member or a SETXCF
operator command. The data set must be available when RRS starts and when it restarts.
Ensure that no element-restart denies the restart of an RRS element or changes its
restart. An exception is an exit routine that vetoes RRS restart but then itself starts the
RRS address space. This technique, however, might delay other elements in the restart
group that have to wait for RRS services to become available.
As with other ARM elements, an ENF signal for event 38 occurs when RRS registers with
automatic restart management or is automatically restarted. For information about ARM
parameters, refer to Chapter 6, Automatic Restart Manager on page 83.
481
482
22
Chapter 22.
z/OS UNIX
This chapter discusses the UNIX System Services environment, which is now called z/OS
UNIX. It describes a shared zFS/HFS environment and examines some zFS commands.
For more information about these topics, refer to z/OS UNIX System Services Planning,
GA22-7800 or z/OS System Services Command Reference, SA22-7802.
For more information about ZFS, refer to z/OS Distributed File Service zFS Administration,
SC24-5989.
483
22.1 Introduction
z/OS UNIX is a component of z/OS that is a certified UNIX implementation, XPG4 UNIX 95. It
was the first UNIX 95 system not derived from the AT&T source code. It includes a shell
environment, OMVS, which can be accessed from TSO.
z/OS UNIX allows UNIX applications from other platforms to run on IBM z/OS mainframes. In
many cases a recompile is all that is needed. Additional effort may be advisable for enhanced
z/OS integration. Programs using hardcoded ASCII numerical values may need adjustment to
support the EBCDIC character set.
Database access (DB2 using Call Attach) is one example of how z/OS UNIX can access
services found elsewhere in z/OS. Such programs cannot be ported to z/OS platforms without
rewriting. Conversely, a program that adheres to standards such as POSIX and ANSI C is
easier to port to the z/OS UNIX environment.
Numerous core System z subsystems (such as TCP/IP) and applications rely on z/OS UNIX.
z/OS 1.9 introduced several new z/OS UNIX features and included improved Single UNIX
Specification Version 3 (UNIX 03) alignment.
22.2 z/OS UNIX file system structure

To be POSIX-compliant, z/OS UNIX needed to support the slash file management system,
as shown in Figure 22-1, which consists of directories and files. This style of file
management, which is used by UNIX, Linux, and various other operating systems, is known
as hierarchical file management.
USS File System Structure

/ Directory
Directory
Directory
Directory
File
File
Directory
File
File
File
Directory
File
File
Directory
File
File
File
File
File
File
File
File
Figure 22-1 z/OS UNIX file system structure
484
Directory
Directory
There are four different types of UNIX file systems supported by z/OS:
Hierarchical File System (HFS)

Temporary File System (TFS)
Network File System (NFS)
System z File System (zFS)
22.2.1 Hierarchical File System

The original UNIX System Services supported a single file system type, Hierarchical File
System (HFS). HFS provided the slash file system support needed.
To z/OS, an HFS is simply one of a number of different data set types (Sequential,
Partitioned, VSAM, and so on). When mounted and made available to the z/OS UNIX
environment, an HFS becomes a container that stores many files and directories. An HFS
can be allocated either by a batch job or by using ISPF 3.2 with DSTYPE=HFS, as seen in
Figure 22-2.
//HFSALLOC
//HFS
//SYSPRINT
//ALLOC0
//
JOB (SWR,1-1),PETER,CLASS=A,MSGCLASS=T,NOTIFY=&SYSUID
EXEC PGM=IEFBR14
DD SYSOUT=*
DD DISP=(,CATLG),DSN=SYSU.LOCAL.BIN,
SPACE=(CYL,(5,4,1)),DSNTYPE=HFS
Figure 22-2 Allocating an HFS data set
1 When an HFS is allocated, it needs to have some directory blocks included. If directory
blocks=0 is specified, or if blocks is omitted, the data set allocates but it is unusable. An error
such as Errno=80x No such device exists; Reason=EF096056 occurs when the system tries
to use it.
22.2.2 Temporary File System

A Temporary File System (TFS) is stored in memory and delivers high-speed I/O. It is a
similar concept to the VIO data set. The /tmp and /dev directories are good candidates for
TFS usage. If /tmp is backed by a TFS, keep in mind that all data in it will lost after a system
restart. The default TFS construct will store the data written to the TFS in the OMVS address
space. It is possible to use a separate address space to support the TFS environment. This is
done by using a colony address space.
22.2.3 Network File System

A Network File System (NFS) is a distributed file system that enables users to access z/OS
UNIX files and directories that are located on remote computers as though they were local.
NFS is independent of machine types, operating systems, and network architectures. Most
UNIX installations have NFS support.
Chapter 22. z/OS UNIX
485
22.2.4 System z File System

A System z File System (zFS) is a UNIX file system that is in a VSAM linear data set. The
newer filesystem, zFS, is more complicated initially to define and support. The advantage of a
zFS is it performs better than an HFS and is easier to manage. The zFS requires started task
(STC) to be running to enable support. It also needs to be defined in the BPXPRMxx parmlib.
The JCL for a typical zFS supporting task is shown in Figure 22-3.
SYS1.PROCLIB(ZFS)
===>
Scroll ===> CSR
***************************** Top of Data *****************************
//ZFS
PROC REGSIZE=0M
//ZFZGO
EXEC PGM=BPXVCLNY,REGION=&REGSIZE,TIME=1440
//IOEZPRM DD DISP=SHR,DSN=SYS1.PARMLIB(IOEFSPRM)
<--ZFS PARM FILE
//*
Figure 22-3 zFS JCL for STC
The JCL to define a zFS can be seen in Figure 22-4. The process has two steps. Step 1
creates a linear VSAM data set. Step 2, which requires the zFS STC to be active, formats the
data set so it can be used by z/OS UNIX. In this case, the data set is a multi-volume
SMS-managed data set.
//ZFSDEFN JOB
PETER4,MSGCLASS=S,NOTIFY=&SYSUID,CLASS=A
//*
USER=SWTEST
//DEFINE
EXEC
PGM=IDCAMS
//SYSPRINT DD
SYSOUT=*
//SYSUDUMP DD
SYSOUT=*
//AMSDUMP DD
SYSOUT=*
//SYSIN
DD
*
DEFINE CLUSTER (NAME(SYSU.LOCAL.BIN)
VOLUME(* * * * * * * * * *)
LINEAR CYL(500 100) SHAREOPTIONS(2))
/*
//CREATE
EXEC
PGM=IOEAGFMT,REGION=0M,
// PARM=('-aggregate SYSU.LOCAL.BIN -compat')
//SYSPRINT DD
SYSOUT=*
//STDOUT
DD
SYSOUT=*
//STDERR
DD
SYSOUT=*
//SYSUDUMP DD
SYSOUT=*
//CEEDUMP DD
SYSOUT=*
//*
2
3
Figure 22-4 Define a zFS data set
The steps are explained in more detail here:

1 The first step is to define a linear VSAM data set. The data set name can be any name that
matches your naming standards. Note there is no need, nor is there any advantage, in using
a special names, such using a HLQ of OMVS or a LLQ of ZFS. By specifying 10 asterisks (*)
in the volume parameter, the data set is SMS-managed and can grow to 10 volumes in size.
2 The second step creates an aggregate, or a filesystem name that is usable by z/OS UNIX.
486
22.3 z/OS UNIX files

This section discusses z/OS UNIX files and the different types of file systems.
22.3.1 Root file system

The root file system is the starting point for the overall file system structure. It consists of the
root (/) directory, system directories, and files. A system programmer defines the root file
system. The system programmer must have an OMVS UID of 0 to allocate, mount, and
customize the root directories.
Note: The system programmer can either be assigned an OMVS UID of 0 or be given
access to issue the SU command. It is simpler to assign a UID of 0 but some sites may
require the system programmer to issue the SU command.
There are two z/OS UNIX configurations available, shared and non-shared. In a non-shared
environment a file system, whether it is HFS or ZFS, can be mounted read-only on multiple
z/OS systems within a sysplex or it can be mounted read-write on a single z/OS system within
a sysplex. As a consequence if the root filesystem, which is associated with an IPL volume, is
shared by multiple systems, it can only be mounted read-only. The default configuration is
non-shared. In a shared environment, file systems can be mounted read-write on multiple
systems within the sysplex. The scope of a shared environment is often called an HFSplex.
The scope of an HFSplex does not need to match the scope of the sysplex, but there can only
be a single HFSplex within a sysplex. For example, in Figure 22-5 the sysplex contains
systems A, B, C, Y, and Z. The HFSplex consists of systems A,B,C, Systems Y and Z are
stand-alone, from a z/OS UNIX perspective
The file system SYSU.LOCAL.BIN can be mounted read-write in systems A, B and C, or it
can be mounted read on all systems.
H F S P le x - s y s te m s A ,B ,C
S ta n d a lo n e S y s te m s
Figure 22-5 HFSPLEX can be smaller than a sysplex
487
22.3.2 Shared environment

The IBM Redbooks publication ABCs of z/OS System Programming Volume 9, SG24-6989,
explains how to set up the shared file system. One way to determine whether you are running
a shared file system environment is to use the D OMVS,F command. In Figure 22-6, notice 2
and 3, which are the lines that define an owner or owning system. This indicates this was
issued in a file sharing system.
D OMVS,F
BPXO045I 19.28.59 DISPLAY OMVS 659
OMVS
0010 ACTIVE
OMVS=(00,FS)
TYPENAME
DEVICE ----------STATUS----------- MODE
ZFS
97 ACTIVE
RDWR
NAME=OMVS.#@$3.SYSTEM.ZFS
PATH=/#@$3
AGGREGATE NAME=OMVS.#@$3.SYSTEM.ZFS
OWNER=#@$3
AUTOMOVE=U CLIENT=N
ZFS
3 ACTIVE
READ
NAME=OMVS.ZOSR18.#@$#R3.ROOT
PATH=/#@$#R3
AGGREGATE NAME=OMVS.ZOSR18.#@$#R3.ROOT
OWNER=#@$2
AUTOMOVE=Y CLIENT=N
. . .
MOUNTED
LATCHES
07/16/2007 L=33
21.30.01
Q=0
2
07/16/2007
21.30.00
L=15
Q=0
Figure 22-6 D OMVS,F in a file sharing environment
1 The D OMVS,F command displays all the file systems in the HFSplex.
2 A file system that is owned by system #@$3 with AUTOMOVE=U. In this case, when
system #@$3 is shut down, the file system is unmounted and becomes unusable.
3 A file system that is owned by system #@$2 with AUTOMOVE=Y. In this case, when
system #@$2 is shut down, the file systems ownership will be taken up by another system in
the sysplex and it will still be available.
In a shared file system environment, instead of each system having read-only its own root file
system, there is a single HFSplex-wide root file system. If this root file system is damaged or
needs to be moved, then the entire HFSplex needs to be restarted. This file system should be
very small and consist of directories and links only.
When a file system is mounted, there is a new attribute, called automove, that is assigned.
The automove attribute indicates what is to happen to the file system when the system that
owns the file system is shut down. There are a number of options available, but the result
effectively is that the file system is unmounted and made unavailable or it is moved to another
system.
When a file system is mounted read-only, the owning system has no impact because every
system directly reads the file system. When the file system is mounted read-write, then the
owning system is important. All updates are performed by the owning system. When a
different system wants to update a file or directory in the file system, the update is
communicated to the owning system, using XCF, which then does the actual update.
The I/O on the requesting system is not completed until the owning system indicates it has
completed the I/O. As a consequence, it is possible to have significant extra XCF traffic
caused by a file system being inappropriately owned. For example, consider a HTTP server
running on system MVSA that logs all the HTTP traffic. When the log file is in a file system
488
owned by MVSB, then all the HTTP logging writes will be transferred, using XCF, from MVSA
to MVSB.
As systems are IPLed, the ownership of file systems, especially those that are
automove-enabled, can change. This not only can cause significant extra XCF traffic, but can
also impact the response time. Your z/OS system programmer is responsible for maintaining
the mount configuration, and therefore should know the optimal configuration.
22.4 zFS administration

Figure 22-3 on page 486 shows a sample proc for the zFS support task. This STC is started
automatically when the z/OS UNIX environment is started. The STC name is defined in
Figure 22-7; in this case it is called ZFS.
. . .
FILESYSTYPE TYPE(ZFS)
ENTRYPOINT(IOEFSCM)
ASNAME(ZFS)
. . .
/* Type of file system to start

/* Entry Point of load module
/* Procedure name
*/
*/
*/ 1
Figure 22-7 BPXPRMxx with for ZFS
1 The ZFS supporting task STC name; the JCL needs to be in a system proclib, such as
SYS1.PROCLIB.
With z/OS 1.7, the zFS supporting address space was terminated with P ZFS. With z/OS 1.8
and later, it is stopped with the F OMVS,STOPFS=ZFS command; this can be seen in
Figure 22-8.
F OMVS,STOPPFS=ZFS
014 BPXI078D STOP OF ZFS REQUESTED. REPLY 'Y' TO
PROCEED. ANY OTHER REPLY WILL CANCEL THIS STOP.
14y
IOEZ00050I zFS kernel: Stop command received.
IOEZ00048I Detaching aggregate OMVS.ZOSR18.#@$#R3.ROOT
IOEZ00387E System #@$3 has left group IOEZFS, aggregate recovery in progress.
IOEZ00387E System #@$3 has left group IOEZFS, aggregate recovery in progress.
IOEZ00357I Successfully left the sysplex group.
IOEZ00057I zFS kernel program IOEFSCM is ending
IEF352I ADDRESS SPACE UNAVAILABLE
$HASP395 ZFS
ENDED
015 BPXF032D FILESYSTYPE ZFS TERMINATED. REPLY
'R' WHEN READY TO RESTART. REPLY 'I' TO IGNORE.
1
2
4
5
Figure 22-8 Stopping the ZFS address space
1 Command to stop the zFS support address space.

2 Message requesting confirmation that the support address space is to be stopped.
3 Message indicating that a zFS data set is being stopped.
4 Message indicating this system is no longer part of the HFSplex.
5 Message indicating the address space is stopping.
6 If the stoppage was not for an IPL, then the zFS support address space could be restarted.
by replying R to this message when the change is complete.
489
The zFS address space cannot be started manually; that is, issuing S ZFS does not work. If
message BPX0232D is given a reply of I, then the only way to restart the ZFS address space
is with the SETOMVS RESET= command (or the SET OMVS= command). We recommend using the
SETOMVS RESET= command, because the alternative can significantly alter the OMVS
configuration.
Unlike an HFS, which will allocate a secondary extent automatically, a zFS needs to be
explicitly grown. This can be done using the z/OS UNIX command zfsadm. If you prefer to do
this in a batch job, then Figure 22-9 shows an example.
//ZFSGRWX JOB 'GROWS ZFS',CLASS=A,MSGCLASS=S,NOTIFY=&SYSUID
//STEP0
EXEC PGM=IKJEFT01
//SYSPROC DD DSN=SYS1.SBPXEXEC,DISP=SHR
//SYSTSPRT DD SYSOUT=*
//SYSTSIN DD *
oshell rm /tmp/zfsgrw_*
//STEP1
EXEC PGM=IEBGENER
//SYSIN
DD DUMMY
//SYSUT2
DD PATH='/tmp/zfsgrw_in',
//
PATHDISP=(KEEP),FILEDATA=TEXT,
//
PATHOPTS=(OWRONLY,OCREAT,OEXCL),
//
PATHMODE=(SIRWXG,SIRWXU,SIRWXO)
//SYSUT1
DD *
zfsadm grow -aggregate SYSU.LOCAL.BIN -size 0
//CONFIG EXEC PGM=BPXBATCH,REGION=0M,PARM='SH /tmp/zfsgrw_in'
//STDERR
DD SYSOUT=*
//STDOUT
DD SYSOUT=*
//SYSUDUMP DD SYSOUT=*
Figure 22-9 ZFSADM - batch
1 Delete some files in /tmp.

2 Copy the command you want to issue into a z/OS UNIX file.
3 Execute the command by running a BPXBATCH job.
490
2
3
Appendix A.
Operator commands
This appendix lists and describes operator commands that can help you to manage your
Parallel Sysplex environment.
491
A.1 Operator commands table

Table A-1 lists operator commands you can use to manage your Parallel Sysplex
environment. For additional information about these commands, refer to:
z/OS MVS System Commands, SA22-7627
z/OS JES2 Commands, SA22-7526
Table A-1 Useful operator commands
Command
Description
ATS STAR commands

D U,,AS
Display auto switchable tape devices.
V xxxx,AS,ON
Turn on the auto switchable attribute for a specific device.
V xxxx,AS,OFF
Turn off the auto switchable attribute for a specific device.
Configuration commands
D IOS,CONFIG
Display IOS config information.
D IOS,GROUP
Display the systems that belong to the same IOS group.
D M=CPU
Display CPU information.
D M=CHP(nn)
Display channel path information for a specific channel.
D M=DEV(nnnn)
Display channel path information for a specific device.
D U,IPLVOL
Display information about the IPL volume.
D IPLINFO
Display IPL information for this system.
D OPDATA
Display operator information.
D PARMLIB
Display the PARMLIB data sets and volumes.
D SYMBOLS
Display the static system symbols.
D SSI
Display information about all the subsystems.
ACTIVATE IODF=xx
Activate a specific IODF data set.
Console commands
D C
Display console characteristics.
D C,A,CA
Display console associations for active consoles.
D C,B
Display consoles with messages queuing for output.
D CNGRP
Display members of the active console group.
D EMCS,S
Display a list of EMCS consoles.
D EMCS,F,CN=consname
Display detailed information about a specific EMCS

console.
V CN(*),ACTIVATE
Activate the HMC console.
V CN(*),DEACTIVATE
Deactivate the HMC console.
V CN(console),MSCOPE=(*)
Modify this console to receive messages from the system

it is defined on.
492
Command
Description
V CN(console),MSCOPE=(*ALL)
Modify this console to receive messages from all systems

in the sysplex.
V CN(console),MSCOPE=(sys1,sys2,...)
Modify this console to receive messages from specific

systems in the sysplex.
V CN(console),ROUT=(ALL)
Modify this console to receive messages with all routing

codes.
V CN(console),ROUT=(rcode1,rcode2,...)
Modify this console to receive messages with specific

routing codes.
RO sysname,command
Route command to a specific system.
RO *ALL,command
Route command to all systems.
RO *OTHER,commands
Route command to all systems except the system where

this command was entered.
DEVSERV commands
DS P,nnnn
Display the status of a specific device.
DS QP,nnnn
Display the PAV configuration of a specific device.
DS SMS,nnnn
Display the SMS information of a specific device.
DQ QD,nnnn
Display diagnostic information about the status of a

specific device and its control unit.
ETR commands
D ETR
Display the status of STP or the Sysplex Timer.
SETETR,PORT=n
Enable ETR port 0 or 1.
GRS commands
D GRS,A
Display GRS configuration information.
D GRS,ANALYZE
Display an analysis of system contention.
D GRS,C
Display GRS contention information.
D GRS,RES=(*,dsname)
Display enqueue contention for a single data set.
D GRS,DEV=nnnn
Display RESERVE requests for a specific device.
D GRS,DELAY
Display jobs which are delaying a T GRSRNL command.
D GRS,SUSPEND
Display jobs which are suspended, pending the

completion of a T GRSRNL command.
T GRSRNL=(xx)
Implement a new GRS RNL member dynamically.
JES2 commands
$D MEMBER(*)
Display every member of the JES2 MAS.
$E MEMBER(sysname)
Perform cleanup processing after a member of a MAS

fails.
$D JOBQ,SPOOL=(%>n)
Display jobs using more than a specified percentage of

the spool.
Appendix A. Operator commands
493
Command
Description
$D MASDEF
Display the MAS environment.
$D CKPTDEF
Display the checkpoint configuration.
$E CKPTLOCK,HELDBY=sysname
Reset the checkpoint lock.
$T CKPTDEF,RECON=Y
Initiate JES2 checkpoint reconfiguration dialog.
$T SPOOLDEF,...
Change spooldef parameters dynamically.
$S XEQ
Allow initiators to select new work.
$P XEQ
Stop initiators selecting new work.
$T JOBCLASS(*),QHELD=Y|N
Hold or release specified job queues.
Logger commands
D LOGGER
Display Logger status.
D LOGGER,CONN
Display all logstreams with connections to the system

where you issued this command.
D LOGGER,L
Display logstream sysplex information.
D LOGGER,STR
Display logstreams defined to any structure.
SETLOGR FORCE,DISC,LSN=logstreamname
Disconnect all connections to a specific logstream from

the system where the command was issued.
SETLOGR FORCE,DEL,LSN=logstreamname
Delete a specific logstream from the LOGR CDS.
LOGREC commands
D LOGREC
Display the status of logrec recording.
SETLOGRC LOGSTREAM
Activate logstream logrec recording.
SETLOGRC DATASET
Activate data set logrec recording.
Operlog commands
D C,HC
Display the status of the hardcopy log.
V OPERLOG,HARDCOPY
Activate Operlog.
V OPERLOG,HARDCOPY,OFF
Deactivate Operlog.
PDSE commands
V SMS,PDSE,MONITOR
Display the status of the PDSE monitor.
V SMS,PDSE,MONITOR,ON|OFF
Turn PDSE monitor processing on or off.
V SMS,PDSE,ANALYSIS
Analyze the state of the PDSE subsystem.
V SMS,PDSE,FREELATCH
Release a latch that the ANALYSIS command has

identified is frozen.
SMF commands
D SMF
Display SMF information.
T SMF=xx
Activate a new SMF parmlib member.
SETSMF parameter
Add or replace an SMF parameter dynamically.
494
Command
Description
SMSVSAM commands
D SMS,SMSVSAM,ALL
Display SMSVSAM server address space.
D SMS,CFLS
Display information about the lock structure in the CF.
D SMS,CFCACHE(structurename|*)
Display information about cache structures in the CF.
D SMS,CFVOL(volid)
Display a list of CF cache structures that contain data for

the specified volume.
D SMS,CICSVR(ALL)
Display information about the CICSVR address space.
D SMS,LOG(ALL)
Display information about the logstreams DFSMStvs is

using.
D SMS,DSNAME(dsname)
Display information about jobs that have a data set open

for DFSMStvs access.
D SMS,JOB(job)
Display information about a specific job using DFSMStvs

services.
D SMS,TRANVSAM
Display status of DFSMStvs.
V SMSVSAM,ACTIVE
Start SMSVSAM.
V SMSVSAM,TERMINATESERVER
Stop SMSVSAM.
V SMS,CFCACHE(cachename),E|Q
Change the state of a cache structure.
UNIX System Services commands

D OMVS,O
Display the current configuration options.
D OMVS,F
Display a list of all HFS and zFS file systems.
D OMVS,A=ALL
Display process information for all UNIX System Services

address spaces.
T OMVS=xx
Activate a new BPXPRMxx parmlib member.
SETOMVS parameter
Add or replace an OMVS parameter dynamically.
VTAM commands
D NET,STATS,TYPE=CFS
Display VTAM connection to ISTGENERIC structure.
WLM commands
D WLM
Display WLM information
V WLM,POLICY=policyname
Activate a WLM service policy
V WLM,APPLENV=applenv,RESUME
Start an application environment.
V WLM,APPLENV=applenv,QUIESCE
Stop an application environment
F WLM,RESOURCE=resource,ON|OFF|RESET
Modify WLM resource state
E task,SRVCLASS=srvclass
Move a task to a different service class.
XCF commands
D XCF
Display systems in the sysplex.
495
Command
Description
D XCF,S,ALL
Display systems in the sysplex and their status

time stamp.
D XCF,COUPLE
Display couple data set information.
D XCF,PI
Display pathin devices and structures.
D XCF,PI,DEV=ALL
Display status of pathin devices and structures.
D XCF,PO
Display pathout devices and structures.
D XCF,PO,DEV=ALL
Display status of pathout devices and structures.
D XCF,POL
Display information about active policies.
D XCF,POL,TYPE=type
Display information about a specific policy.
D XCF,STR
Display a list of all the structures defined in the CFRM

policy.
D XCF,STR,STAT=ALLOC
Display the allocated structures.
D XCF,STR,STRNAME=strname
Display detailed information for a specific structure.
D CF
Display detailed information about the CFs.
D XCF,CF
Display information about the CFs.
D XCF,ARMSTATUS
Display information about ARM.
D XCF,ARMSTATUS,DETAIL
Display detailed information about ARM.
V XCF,sysname,OFFLINE
Vary a system out of the sysplex.
SETXCF COUPLE,ACOUPLE=dsn,TYPE=type
Add an alternate couple data set for a specific

component.
SETXCF COUPLE,PSWITCH,TYPE=type
Remove the primary couple data set and replace it with

the alternate couple data set.
SETXCF START,POL,TYPE=type,POLNAME=polname
Start a policy.
SETXCF START,REALLOC
Reallocates CF structures according to the preflist in the

active CFRM policy.
SETXCF START,RB,POPCF=cfname
Reallocates CF structures into a specific CF according to

the preflist in the active CFRM policy.
SETXCF START,RB,CFNM=cfname,LOC=OTHER
Reallocates CF structures into another CF according to

their preflist.
SETXCF START,RB,STRNAME=strname
Rebuild a CF structure.
SETXCF START,RB,DUPLEX,STRNAME=strname
Start duplexing for a CF structure.
SETXCF START,RB,DUPLEX,CFNAME=cfname
Start duplexing for all the CF structures in a specific CF.
SETXCF START,ALTER,STRNAME=strname,SIZE=nnnn
Start CF structure alter processing to change the

structure size.
SETXCF FORCE,STR,STRNAME=strname
Delete a persistent CF structure.
SETXCF FORCE,CON,STRNAME=strname,CONNAME=conname
Delete a failed-persistent connection.
SETXCF FORCE,STRDUMP,STRNAM=strname
Delete a structure dump for a specific structure.
496
Command
Description
SETXCF START,CLASSDEF,CLASS=class
Start a transport class.
SETXCF START,MAINTMODE,CFNM=cfname
Place a CF into maintenance mode.
SETXCF START,PI,DEV=nnnn
Start an inbound signalling path via a CTC.
SETXCF START,PI,STRNM=(strname)
Start an inbound signalling path via a CF structure.
SETXCF START,PO,DEV=nnnn
Start an outbound signalling path via a CTC.
SETXCF START,PO,STRNM=(strname)
Start an outbound signalling path via a CF structure.
SETXCF START,CLASSDEF,CLASS=class
Start a transport class.
SETXCF STOP,MAINTMODE,CFNM=cfname
Make a CF available for allocations again after

maintenance is complete.
SETXCF STOP,PI,DEV=nnnn
Stop an inbound signalling path via a CTC.
SETXCF STOP,PI,STRNM=(strname)
Stop an inbound signalling path via a CF structure.
SETXCF STOP,PO,DEV=nnnn
Stop an outbound signalling path via a CTC.
SETXCF STOP,PO,STRNM=(strname)
Stop an outbound signalling path via a CF structure.
SETXCF MODIFY,...
Modify XCF parameters:

pathin
pathout
transport classes
497
498
Appendix B.
List of structures
Table B-1 on page 500 in this appendix lists information about the exploiters of the Coupling
Facility including structure name, structure type, structure disposition, connection disposition,
and whether the structure supports rebuild.
499
B.1 Structures table

Table B-1 Coupling Facility structure information
Exploiter
Structure name
Structure
type
Structure
disposition
Connection
disposition
Support
rebuild?
CICS DFHLOG
user defined
List
Delete
Delete
Yes
CICS DFHSHUNT
user defined
List
Delete
CICS Log of Logs
user defined
List
CICS Data Tables
DFHCFLS_...
List
Keep
Keep
No
CICS Named Counter Server
DFHNCLS_...
List
Keep
Keep
No
CICS Temporary Storage
DFHXQLS_...
List
Delete
Delete
No
CICS/VSAM RLS Cache
user defined
Cache
Delete
Delete
Yes
CICS/VSAM RLS Lock
IGWLOCK00
Lock
Keep
Keep
Yes
DB2 SCA
grpname_SCA
List
Keep
Delete
Yes
DB2 V8 GBP
grpname_GBP...
Cache
Delete
Keep
Yes
Enhanced Catalog Sharing
SYSIGGCAS_ECS
Cache
Delete
Delete
Yes
GRS Star
ISGLOCK
Lock
Delete
Delete
Yes
DFSMShsm Common Recall

Queue
SYSARC_..._RCL
List
IMS Lock
user defined
Lock
Keep
Keep
IMS OSAM
user defined
Cache
Delete
Delete
IMS VSAM
user defined
Cache
Delete
Delete
IMS Fast Path
user defined
IMS Fast Path Shared Message

Queue (EMH)
user defined
List
Keep
Keep
IMS Fast Path Overflow Shared

Message Queue
user defined
IMS FF Shared Message Queue
user defined
List
Keep
Keep

Overflow
user defined

Log stream
user defined
List
Keep
Keep
IMS Fast Path Shared Message

Queue Log stream
user defined
List
Delete
IMS Resource Manager
user defined
List
Keep
IMS VSO
user defined
Cache
Delete
Intelligent Resource Director
SYSZWLM_cpuidcputype
Cache
Delete
IRLM (DB2)
grpname_LOCK1
Lock
Keep
CICS/VR
500
Yes
Keep
No
Keep
Yes
Exploiter
Structure name
Structure
type
Structure
disposition
Connection
disposition
Support
rebuild?
IRLM (IMS)
user defined
Lock
Keep
Keep
Yes
JES2 Checkpoint
user defined
List
Keep
Delete
No
LOGREC Log stream
user defined
List
Delete
Yes
MQ Shared Queues Admin
mqgrpname
List
Keep
No
MQ Shared Queues
Applications
mqgrpname
List
Keep
No
OPERLOG Log stream
user defined
List
Delete
Yes
RACF Backup DB Cache
IRRXCF00_B00n
Cache
Delete
Delete
Yes
RACF Primary DB Cache
IRRXCF00_P00n
Cache
Delete
Delete
Yes
RRS Archive Log stream
user defined
List
Delete
RRS DELAYED_UR Log stream
user defined
List
Delete
RRS MAIN_UR Log stream
user defined
List
Delete
RRS RESTART Log stream
user defined
List
Delete
RRS RMDATA Log stream
user defined
List
Delete
SmartBatch
SYSASFPnnnn
List
Delete
Delete
Yes
System Logger
user defined
List
Delete
Keep
Yes
TCP/IP Sysplex Ports
EZBEPORT
List
TCP/IP Sysplex-wide security

associations
EZDVIPA
List
Tivoli System Automation for

z/OS Automation Manager
HSA_LOG
List
Tivoli System Automation for

z/OS Health Checker
ING_HEALTHCHKLOG
List
VSAM/RLS Cache
IGWCACHEn
Cache
VSAM/RLS Lock
IGWLOCK00
Lock
Keep
VTAM GR
ISTGENERIC or user
defined
List
Delete
Keep
Yes
VTAM MNPS
ISTMNPS
List
Keep
Keep
Yes
XCF
IXC...
List
Delete
Delete
Yes
z/OS Health Checker
HZS...
List
Delete
Yes
WebSphere
Appendix B. List of structures
501
502
Appendix C.
Stand-alone dump on a Parallel

Sysplex example
This appendix provides an example of a stand-alone dump (SAD) that was taken on z/OS
system AAIS in the P04AAIBM Parallel Sysplex. The P04AAIBM sysplex environment
consists of two z/OS images named AAIS and AAIL running z/OS 1.8.
Note: The Parallel Sysplex environment used for the stand-alone dump is from the IBM
Australia test bed.
We IPLed the SADMP program from DASD using device address 4038.
Our SAD output was written to a DASD data set, SYS1.SADMP. It was allocated across two
3390-3 disk volumes, SP413A (413A) and SP413B (413B), using the AMDSADDD REXX
exec in SYS1.SBLSCLI0.
Note: The input and output devices you use for a stand-alone dump in your installation will
vary. Consult your system programmer for this information.
We used the HMC as the console for the SADMP program. You can use any console that is
defined to the SADMP program.
503
C.1 Reducing SADUMP capture time

Note: The tasks described in C.1 and C.2 should be carried out by your system
programmer, but they are included here for completeness.
The best stand-alone dump performance is achieved when the dump is taken to a striped
DASD stand-alone output data set, also called a multi-volume dump group. The stand-alone
dump output data set needs to be placed on a volume behind a modern control unit, like the
IBM Enterprise Storage subsystem, as opposed to non-ESS DASD or dumping to tape.
C.2 Allocating the SADUMP output data set

The stand-alone output data set is allocated by invoking REXX exec AMDSADDD in
SYS1.SBLSCLI0. Refer to z/OS V1R8.0 MVS Diagnosis Tool and Service Aids, GA22-7589,
for detailed information about this topic.
When the REXX exec is invoked, various prompts are issued. One of the prompts will be for
the output data set; it asks whether it should be allocated multivolume.
Recommendation: Allocate the output data set as multivolume, because this will reduce
the amount of time needed to capture the stand-alone dump.
When dumping to a DASD dump data set, there are significant performance improvements
observed with striping the data to a multivolume stand-alone dump data set. Striping refers
to the use of spanned volumes for the stand-alone dump output data set. To implement
striping, specify a volume list (VOLLIST) in the AMDSADDD REXX exec to designate a list of
volumes to use for the data set. In our example, we used volumes SP413A and SP413B.
C.3 Identifying a DASD output device for SAD

For this example, we allocated the stand-alone output data set with a name of SYS1.SADMP
on volumes SP413A and SP413B on device addresses 413A and 413B.
C.4 Identifying a tape output device for SAD

If you are using tapes as the output device for a SAD and the tape device is not part of an
Automated Tape Library (ATL), mount a scratch in a tape drive that is online to the z/OS
system being dumped. However, if you will be using a tape device that is part of an ATL, then
additional steps are required before performing the SAD to the ATL device, as explained
here:
1. Take note of the volser for a scratch tape to be used for the SAD and enter it into the ATL.
2. Using the console for the ATL, ensure that the tape device being used is configured as a
stand-alone device.
3. Enter the tape device address on the ATL console.
4. Enter the volser of the scratch tape to be used through the ATL console.
5. The ATL console should issue a message indicating that the mount is complete.
504
Restriction: At this time, there is no support for the 3584/3592 tape library to be used as a
SAD output medium.
C.5 Performing a hardware stop on the z/OS image

On the HMC, in the Groups Work Area, select the CPC Images icon and double-click it; see
Figure C-1.
Figure C-1 HMC Groups Work Area
Note: Your installation may have enabled the Lock out disruptive tasks: radio button
on the image icon. If that is the case, you must select the No radio button before
proceeding; see Figure C-2.
Figure C-2 Task Information - Lock out disruptive tasks
After double-clicking the CPC Images icon, a list of images is displayed as shown in
Figure C-3 on page 506.
Appendix C. Stand-alone dump on a Parallel Sysplex example
505
Figure C-3 AAIS (CPC Images Work Area)
On the HMC, in the CPC Images Work Area, we selected system AAIS by single-clicking it to
highlight it. Then we double-clicked the STOP All icon in the CPC Recovery window, as
shown in Figure C-4.
Figure C-4 Selecting STOP All on HMC
A confirmation panel was displayed, as shown in Figure C-5 on page 507.
506
Figure C-5 Confirmation panel for STOP All
For this example, we selected Yes to confirm that the STOP All action should continue. The
panel shown in Figure C-6 displays the progress of the STOP All request for system AAIS.
Figure C-6 Progress panel for STOP All
507
C.6 IPLing the SAD program

We selected the LOAD icon on the CPC Recovery window of the HMC and performed these
steps:
1. We selected Load Normal.
Attention: Do not use the LOAD CLEAR option. Using the LOAD CLEAR option
erases main storage, which means that you will not be able to diagnose the failure
properly.
2. We selected Store Status.
3. We entered the device address of the SAD program: 4038.
4. We left the Load Parameter field blank.
5. We clicked OK.
Figure C-7 displays the Load panel with all of the required fields.
Figure C-7 Load panel for SAD
The Load Task Confirmation panel was then displayed, as shown in Figure C-8 on page 509.
508
Figure C-8 Load Task Confirmation panel
We clicked Yes on the Load Task Confirmation panel. This was followed by the Load
Progress panel, as shown in Figure C-9.
Figure C-9 Load Progress panel
C.7 Sysplex partitioning

While the SAD IPL was in progress, we received messages on system AAIL, as shown in
Figure C-10 on page 510. These messages indicated that sysplex partitioning had started. In
response to message IXC402D, we replied DOWN.
Note: We did not need to perform a System Reset of AAIS, because the IPL of the SAD
program does this.
509
40 IXC402D AAIS LAST OPERATIVE AT 15:42:29. REPLY DOWN AFTER SYSTEM

RESET, OR INTERVAL=SSSSS TO SET A REPROMPT TIME.
R 40,DOWN
IXC101I SYSPLEX PARTITIONING IN PROGRESS FOR AAIS REQUESTED BY XCFAS.
REASON: SYSTEM STATUS UPDATE MISSING
IEA257I CONSOLE PARTITION CLEANUP IN PROGRESS FOR SYSTEM AAIS.
ISG011I SYSTEM AAIS - BEING PURGED FROM GRS COMPLEX
ISG013I SYSTEM AAIS - PURGED FROM GRS COMPLEX
IEA258I CONSOLE PARTITION CLEANUP COMPLETE FOR SYSTEM AAIS.
Figure C-10 Sysplex partitioning during SAD
C.8 Sending a null line on Operating System Messages task

When the SAD IPL completed, the Operating System Messages window for AAIS displayed,
as shown in Figure C-11.
Tip: Issue the V CN(*),ACTIVATE command prior to entering any other commands.
Figure C-11 Identifying the HMC console to SAD
By issuing V CN(*),ACTIVATE, we identified the HMC console to the SAD program.
C.9 Specifying the SAD output address

The SAD program then requested the output device address, as shown in Figure C-12 on
page 511.
Note: Depending on how your installation has configured the SAD program, you may not
receive prompting.
510
Figure C-12 Specifying an output address for SAD
We replied with the address of our DASD device (413A) that had SYS1.SADMP on it.
Note: If the SAD output data set has been allocated across multiple DASD volumes, you
specify the first device address to the AMD001A message.
C.10 Confirming the output data set

If the SAD program recognizes that the output device specified has been used before,
message AMD096A is issued. The message will ask whether the output data set should be
used or another one specified.
Attention: Consult your system programmer before replying to the AMD096A message.
In our SAD, the output data set was created new and we did not receive the AMD096A
message, as seen in Figure C-13.
Figure C-13 Specifying the output address
Note: In our SAD example, we received messages AMD091I and AMD092I as seen in
Figure 18-38. These messages were issued because the SAD output data set was
originally created using a data set name of SYS1.AAIS.SADMP. When the SAD program
was generated, we allowed the output SAD data set name to default to SYS1.SADMP.
511
C.11 Entering the SAD title

As seen in Figure C-13 on page 511, we were then prompted for the SAD title with message
AMD011A. In your case, enter some meaningful text at the prompt.
C.12 Dumping real storage

The SAD program then started to dump real storage, as shown in Figure C-14.
Figure C-14 Dumping real storage
C.13 Entering additional parameters (if prompted)

Because we were running an unmodified SAD program, when the dump of real storage
completed, we received message AMD056I as shown in Figure C-15 on page 513. In your
case, you may receive additional prompts, depending on how your systems programmer has
set up the SAD program.
512
Figure C-15 SAD complete
C.14 Dump complete

Message AMD056I was issued when the dump completed, as shown in Figure C-15.
C.15 Information APAR for SAD in a sysplex environment

Note: The following information is derived from an IBM Informational APAR II08659
describing a procedure for taking a stand-alone dump in a sysplex environment. You can
use it to validate the procedures that you may have in place at your installation.
This Information APAR provides recommendations for taking a standalone dump of z/OS
when the z/OS system resides in a sysplex.
Scenario: A z/OS system is not responding

Examples of when a stand-alone dump may be needed
Consoles do not respond.
z/OS is in a WAIT state.
A stand-alone dump has been requested by IBM Level 2 support.
A z/OS system is in a status update missing condition and has been or is waiting to be
removed from the sysplex.
513
Here are the HIGH LEVEL steps to perform when taking a stand-alone dump of a z/OS
system that resides in a sysplex. Assume that the z/OS system to be dumped is SYSA.
Procedure A
Important: Follow each step in order.
1. Perform the STOP function to place the SYSA CPUs into the stopped state.
2. IPL the stand-alone dump program
3. Issue VARY XCF,SYSA,OFFLINE from another active z/OS system in the sysplex if message
IXC402D or IXC102A is not already present.
4. Reply DOWN to message IXC402D, IXC102A
Notes on Procedure A
You do not have to wait for the stand-alone dump to complete before issuing the VARY
XCF,SYSA,OFFLINE command.
Performing Procedure A steps 3 and 4 immediately after IPLing the stand-alone dump will
expedite sysplex recovery actions for SYSA. This will allow resources held by SYSA to be
cleaned up quickly, and enable other systems in the sysplex to continue processing.
After the stand-alone dump is IPLed, z/OS will be unable to automatically ISOLATE
system SYSA via SFM, so message IXC402D or IXC102A will be issued after the VARY
XCF,SYSA,OFFLINE command or after the XCF failure detection interval expires. You must
reply DOWN to IXC402D/IXC102A before sysplex partitioning can complete.
Do not perform a SYSTEM RESET in response to IXC402D or IXC102A after IPLing the
stand-alone dump. The SYSTEM RESET is not needed in this case because the IPL of
stand-alone dump causes a SYSTEM RESET to occur. After the stand-alone dump is
IPLed, it is safe to reply DOWN to IXC402D or IXC102A.
If there is a time delay between Procedure A steps 1 and 2, then use Procedure B.
Executing Procedure B will help to expedite the release of resources held by system
SYSA while you are preparing to IPL the stand-alone dump program.
Procedure B
Important: Follow each step in order unless otherwise stated.
1. Execute the STOP function to place the SYSA CPUs into the stopped state.
2. Perform the SYSTEM RESET-NORMAL function on SYSA.
3. Issue VARY XCF,SYSA,OFFLINE from another active z/OS system in the sysplex if message
IXC402D or IXC102A is not already present.
4. Reply DOWN to message IXC402D or IXC102A.
5. IPL the stand-alone dump program. This step can take place any time after step 2.
Notes on Procedure B
Performing Procedure B steps 3 and 4 immediately after doing the SYSTEM RESET will
expedite sysplex recovery actions for SYSA. This will allow resources held by SYSA to be
cleaned up quickly, and enable other systems in the sysplex to continue processing.
After a SYSTEM RESET is performed, z/OS will be unable to automatically ISOLATE
system SYSA via SFM, so message IXC402D or IXC102A will be issued after the VARY
514
XCF,SYSA,OFFLINE command or after the XCF failure detection interval expires. You must
reply DOWN to IXC402D/IXC102A before sysplex partitioning can complete.
Both of these procedures emphasize the expeditious removal of the failing z/OS system
from the sysplex. If the failed z/OS is not partitioned out of the sysplex promptly, some
processing on the surviving z/OS systems might be delayed.
Attention: Do not IPL standalone dump more than once. Doing so will invalidate the dump
of z/OS. To restart stand-alone dump processing, perform the CPU RESTART function on
the CPU where the stand-alone dump program was IPLed.
For additional information about stand-alone dump procedures, refer to z/OS V1R8.0 MVS
Diagnosis Tool and Service Aids, GA22-7589.
515
516
Related publications
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.
IBM Redbooks
For information about ordering these publications, see How to get Redbooks on page 519.
Note that some of the documents referenced here may be available in softcopy only.
CICS Workload Management Using CICSPlex SM and the z/OS/ESA Workload Manager,
GG24-4286
Getting the Most Out of a Parallel Sysplex, SG24-2073
DB2 in the z/OS Platform Data Sharing Recovery, SG24-2218
IMS/ESA Version 6 Guide, SG24-2228
IMS/ESA Data Sharing in a Parallel Sysplex, SG24-4303
Automating CICS/ESA Operations with CICSPlex SM and NetView, SG24-4424
OS/390 z/OS Multisystem Consoles Implementing z/OS Sysplex Operations, SG24-4626
OS/390 z/OS Parallel Sysplex Configuration Cookbook, SG24-4706
CICS and VSAM Record Level Sharing: Recovery Considerations, SG24-4768
JES3 in a Parallel Sysplex, SG24-4776
IMS/ESA Parallel Sysplex Implementation: A Case Study, SG24-4831
IMS/ESA Version 6 Shared Queues, SG24-5088
IMS Primer, SG24-5352
Merging Systems into a Sysplex, SG24-6818
Systems Programmers Guide to: z/OS System Logger, SG24-6898
IMS in the Parallel Sysplex Volume I: Reviewing the IMSplex Technology, SG24-6908
IMS in the Parallel Sysplex Volume II: Planning the IMSplex, SG24-6928
IMS in the Parallel Sysplex Volume III: IMSplex Implementation and Operations,
SG24-6929
ABCs of z/OS System Programming Volume 9, SG24-6989
Server Time Protocol Planning Guide, SG24-7280
Implementing REXX Support in SDSF, SG24-7419
Other publications
These publications are also relevant as further information sources:
OS/390 Parallel Sysplex Recovery, GA22-7286
z/OS V1R8.0 MVS Diagnosis Tool and Service Aids, GA22-7589
517
z/OS UNIX System Services Planning, GA22-7800

OS/390 z/OS Initialization and Tuning Reference, GC28-1752
OS/390 z/OS Planning: Global Resource Serialization, GC28-1759
OS/390 Planning Operations, GC28-1760
OS/390 z/OS Setting Up a Sysplex, GC28-1779
OS/390 z/OS System Commands, GC28-1781
OS/390 z/OS System Messages, Vol 5 (IGD-IZP), GC28-1788
OS/390 JES2 Commands, GC28-1790
OS/390 JES2 Initialization and Tuning Guide, GC28-1791
OS/390 JES3 Commands, GC28-1798
SmartBatch for OS/390 Customization, GC28-1633
OS/390 Parallel Sysplex Test Report, GC28-1963
IBM OMEGAMON z/OS Management Console Users Guide, GC32-1955
S/390 9672 Parallel Transaction Server Operations Guide, GC38-3104
JES2 Initialization and Tuning Guide, SA22-7532
z/OS JES2 Initialization and Tuning Reference, SA22-7533

z/OS V1R10.0 JES3 Commands, SA22-7540
MVS Planning: Operations, SA22-7601
z/OS V1R10.0 MVS Programming: Sysplex Services Reference, SA22-7618
z/OS V1R10.0 MVS Setting up a Sysplex, SA22-7625
z/OS JES2 Commands, SA22-7526
z/OS MVS System Commands, SA22-7627
z/OS MVS System Messages Volume 10 (IXC - IZP), SA22-7640
SDSF Operation and Customization, SA22-7670
z/OS System Services Command Reference, SA22-7802
IBM Health Checker for z/OS Users Guide, SA22-7994
IMS Command Reference Manual V9, SC18-7814
IMS Common Queue Server Guide and Reference Version 9, SC18-7815
IMS Common Service Layer Guide and Reference V9, SC18-7816
IMS Database Recovery Control (DBRC) Guide and Reference Version 9, SC18-7818
IMS Connect Guide and Reference, Version 9, SC18-9287
RACF System Programmers Guide, SC23-3725
z/OS Distributed File Service zFS Administration, SC24-5989
IMS/ESA Operations Guide, SC26-8741
OS/390 SDSF Guide and Reference, SC28-1622
OS/390 Security Server (RACF) Command Language Reference, SC28-1919
Hardware Management Console Operations Guide, SC28-6837
z/OS V1R8.0 Communications Server: SNA Resource Definition Reference, SC31-8778
z/OS Communications Server: IP System Administration Commands, SC31-8781
518
IBM Press publication An Introduction to IMS, 2004, ISBN 0131856715

Parallel Sysplex Performance: XCF Performance Considerations white paper
http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP100743/
Online resources
These Web sites are also relevant as further information sources:
IBM homepage for Parallel Sysplex
http://www.ibm.com/systems/z/advantages/pso/index.html/
How to get Redbooks

You can search for, view, or download Redbooks, Redpapers, Technotes, draft publications
and Additional materials, as well as order hardcopy Redbooks publications, at this Web site:
ibm.com/redbooks
Help from IBM

IBM Support and downloads
ibm.com/support
IBM Global Services
ibm.com/services
Related publications
519
520
Index
Numerics
0A2 Wait State 66

9037 32
cache structure 8
Cache structure (GBPs) 369
capacity
unused 4
capping 124
CDS 23
CF 368
CICS 351
data tables
CICS 355
DB2 369
failure 22
CICS recovery 352
IMS 413
locking 390
log stream 309
outages 9
physical information 15
receiver (CFR) 16
role in locking 5
role in maintaining buffer coherency 5
sender (CFS) 16
structure 184
structures 9, 22, 105
VTAM 326
CFRM
CDS allocation during IPL 45
Couple Data Set 9
initialization 46
policies 85, 115
policy 9, 21
Channel-to-Channel adapter (CTC) 6, 184
checkpoint data set (CQS) 415
CICS
APPLID 346
ARM 362
CF 351
CF data tables 355
CICSPlex 364
co-existing with IMS 400
commands for generic resources 333
deregister from a generic resource group 333
interregion communication (IRC) 347
introduction 346
journal 348
log 348
managing generic resources 333
named counter server 359
remove from a generic resource group 333
shared temporary storage 352
transaction routing 348
VTAM 347
with IMS DB 398
A
abnormal stop 68
action message retention facility (AMRF) 297
adding
a system image to a Parallel Sysplex 50
CF 120
alerts 304
applications
critical 84
ARM 252
automation 84
cancelling with ARMRESTART parameter 98
changing ARM policy 90
CICS 362
cross-system restarts 100
defining RRS 481
description 84
operating with ARM 98
policies 90, 425
policy management 90
restrictions 99
same system restarts 98
starting ARM policy 90
ARM (see Automatic Restart Manager) 394
ARM Couple Data Set 7
ARMRESTART 94
ARMWRAP 96
ATRHZS00 479
AUTOEMEM 26
automated monitoring 14
Automatic Restart Manager (ARM) 7, 84, 394
ARM 25
IMS 421
WebSphere MQ 466
automation
ARM 99
automove 488
B
base sysplex
definition 2
Batch Message Processing (BMP) region 403
BronzePlex 9
buffer
coherency 5
Business Application Services (BAS) 365
Index
521
XCF 346
CICS Multi Region Option 8
CICSPlex 364
Coordinating Address Space (CAS) 365
Environment Services System Services (ESSS) 365
System Manager (CMAS) 365
CICSPlex System Manager 8
CICSPlex System Manager (CPSM) 363
CLEANUP 23
CLEANUP interval 64
clone 3
CMDSYS 298
commands
cancelling with ARMRESTART parameter 98
CF 14
CFCC 137
changing ARM policy 90
display 14
GRS 28
IMS 429
IRLM 429
JES2 25
miscellaneous 32
POPULATECF parameter of REBUILD 154
REBUILD POPULATECF 154
ROUTE 34
starting ARM policy 90
table of 492
V 4020,ONLINE,UNCOND 194
XCF 14
Common Queue Server (CQS) 410
common time 4
need for, in a sysplex 4
connection disposition 109
connection state 109
CONNFAIL 75
console 27
activation 47
buffer shortage 295
CMDSYS parameter 298
EMCS 299
extended MCS 286
group 290
initialization 40
IPL process 292
master 286
message flood automation (MFA) 301
message scope (MSCOPE) 289
messages 28
remove 304
removing 291
ROUTCDE 293
SYSCONS 291
z/OS management 304
console groups 288
consoles 283
sysplex
naming 288
consoles in a sysplex
managing 283
522
MSCOPE implications 289

control information 166
Coordinated Timing Network (CTN) 32
Couple Data Set (CDS) 23, 45, 165
alternate 168
concurrent failure 180
configuration 167
failure 177
primary 168
replacement 172
spare 168
Coupling Facility 102
activate 124, 137
adding 120
CFRM policy 136
Control Code (CFCC). 122
displaying structures 105
fencing services 75
moving structures 129
processor capping 124
processor weight 124
removing 125, 131
restoring 136
structures 408
system logger considerations 307
volatility 144
coupling facility
alternate 129
CFRM policy 132
commands 14
description 8
displaying logical view 103
dynamic dispatching 144
failures
IMS 417
processor storage 124, 127
Coupling Facility (CF)
connection to sysplex 40
link failure 75
loss of connectivity 75
structure duplexing 113
Coupling Facility Control Code 9
Coupling Facility Control Code (CFCC)
commands 137
Coupling Infiniband (CIB) 17
CPC
activate 124
Power-on Reset (POR) 124
CQS 410
Cross-Invalidation
introduction 5
Cross-System Coupling Facility (XCF) 183
connectivity 184
Cross-System Extended Service (XES) 74
Customer Information Control System (see CICS) 346
D
D XCF,S,ALL 67
DA command
display active panel example 238
Data Entry Data Base Virtual Storage Option (DEDB

VSO) 414
data sharing 5, 367, 406
DB2 131
IMS 131
RACF 133
database manager
multiple instances 5
database sharing
IMS 401
DB2 367368
CF 368369
data sharing 131, 367
GBP user-managed duplexing 371
intersystem resource lock manager 368
introduction 368
lock structures 391
restart light 393
structure 369
DBCTL 398
DBRC (IMSDBRC) 403
DCCTL
IMS TM with DB2 399
determining
Automatic Restart Manager status 25
CF names and node descriptor 15
names of all structures defined in the CFRM policy 19
number and names of CFs 15
number and names of systems in the sysplex 14
number and types of Couple Data Sets 23
physical information about the CF 15
system status 14
which structures are in each CF 15
DEVSERV 29
DFHLOG 349
DFHSHUNT 349
DISPLAY command
D NET,STATS,CFS 331
displaying system logger status 310
display command 14
DLI Separate Address Space (DLISAS) 403
DUPLEX keyword 115
duplexing 371, 474, 478
IMS 416
system-managed 114
user-managed 114
dynamic system interchange (DSI) 276
dynamic workload balancing 7
E
element 466
EMCS console 299
ETR
mode 31
Event Notification Facility (ENF) 310
Expedited Message Handler Queue (EMHQ) structure
415
exploiting installed capacity 4
extended MCS (EMCS) consoles 287
extended MCS consoles 286
External Timer References 31
F
failed-persistent state 326
failure
CF 22
IPL failure scenarios 52
processor 443
Failure Detection Interval 6
failure management
restart management 3
Fast Path databases 406
fencing 64, 75
G
Generic Resources 324, 402
CICS 333
managing 330
TSO 334
Global Resource Serialization (GRS) 8, 28
initialization 40
Global Resource Sharing (GRS)
initialization 47
GoldPlex 10
Group Buffer Pool (GBP) 369
GRS
commands 28
description 8
ring initialization in first IPL 47
H
Hardware Management Console (HMC) 102, 122, 291
image profile 122
reset profile 122
health check 304
Health Checker 258
RRS 479
HFS 485
HFSplex 487
Hierarchical File System (HFS) 485
HMC
SAD 503
HZSPRINT 266
HZSPROC 258
I
IBM Tivoli OMEGAMON z/OS Management Console 304
IEARELCN 304
IEARELEC 304
IEECMDPF program to create command prefixes 35
IEEGSYS 300
IMS
ARM 421
CF failures 417
CF structures 408, 413
Connect 323
data sharing with shared queues 409
Index
523
database sharing 401

duplexing 416
IMS TM with DB2 399
in a sysplex 406
introduction 398
IRLM commands 429
operating in a sysplex 426
recovery procedures 431
structures 413
IMS Connect
introduction 342
IMS control region (IMSCTL) 402
IMS DB/DC 400
IMSplex 404
operating 426
Information Management System (IMS) 397
Integrated Cluster Bus (ICB) 102
Integrated Fast Path (IFP) regions 403
Internal Coupling Channel (IC) 102
Internal Resource Lock Manager (IRLM) 407
interregion communication (IRC) 347
Inter-System Channel (ISC) 102
Inter-system Resource Lock Manager (IRLM) 368
IODF
data set 34
IPL
after shutdown 50
ARM cross-system restarts 100
first system 40
IXC207A 52, 55
IXC211A 54
messages on SYSCONS 292
of additional system 50
of first system after abnormal shutdown 48
of first system after normal shutdown 41
overview 40
z/OS system image 39
IPL problems
COUPLExx parmlib member syntax errors 54
IXC207A 52
IXC211A 54
maximum number of systems reached 52
no CDS specified 54
unable to establish connectivity 56
wrong CDS names specified 55
IRLM
commands 429
IRLM lock structures 414
IXC202I 52
IXC207A 52, 55
IXC211A 54
IXC256A 181
IXCMIAPU 313
IXGLOGR 309
J
Java 305
Java Batch Processing (JBP) regions 403
Java Message Processing (JMP) region 403
524
JES2
checkpoint 204
checkpoint definitions 25
clean shutdown on any JES2 in a MAS 220
cold start 214
on an additional JES2 in a MAS 216
commands 25
hot start 219
in a Parallel Sysplex 201
loss of CF checkpoint reconfiguration 213
monitor 227
Multi-Access Spool (MAS) 26, 202
reconfiguration dialog 132
remove checkpoint structure 132
restart 213
SDSF JC command 248
SDSF MAS panel 247
shutdown 220
thresholds 204
warm start 216
JES2 CF checkpoint reconfiguration 208
JES2AUX 203
JES2MON 203
JES3 271
in a sysplex 273
networking with TCP/IP 277
operator commands 281
JESPLEX 202
JESXCF 203, 273
L
list structure 8, 388
list structure (SCA) 369
Load Balancing Advisor (LBA) 323
introduction 341
lock structure 8, 369
lock structures 391
lock table entry (LTE) 391
locking
DB2 and IRLM 390
locking in a sysplex 5
LOG command 233
log records 4
log token (key) 308
LOGREC 7, 320
disable 133
logstream
management 320
M
master console 286
MCS
extended 286
Message Flood Automation (MFA) 301
Message Processing Regions (MPR) 403
Message Queue (MSGQ) structure 415
message queues 404
message scope (MSCOPE) 289
messages
console 28
IXC207A 52
IXC211A 54
monitoring
automation product 14
monitoring JES2 227
MSCOPE implications 289
Multi-Access Spool (MAS) 26, 36, 202
multiple console support (MCS) 284
multiregion operation (MRO)
CICS
multiregion operation (MRO) 347
multisystem console support 284
N
named counter server 359
NCS structure 360
Network File System (NFS) 485
networking in a Parallel Sysplex
CICS generic resources 333
deregister CICS from a generic resource group 333
deregister TSO from a generic resource group 334
determine status of generic resources 330
managing CICS generic resources 333
remove CICS from a generic resource group 333
NJE 277
Nucleus Initialization Program (NIP) 40
O
OLDS data sets 404
OMEGAMON 304
OMVS 484
Open Systems Adapter (OSA) 324
operator
commands table 492
OPERLOG 7, 133, 233
SDSF OPERLOG panel 235
outage
masking 3
P
Parallel Sysplex
abnormal shutdown 68
activation 45
CFRM initialization 46
checking if SFM is active 61
checking that system is removed 68
CICS 346
consoles 283
Coupling Facility 102
definition 2
description 2
description of CF 8
description of GRS 8
IMS 406, 426
IPL 3940
of additional system 50
of first system after abnormal shutdown 48
of first system after normal shutdown 41
overview 40
problem scenarios 41
problems in a Parallel Sysplex 52
IPL of additional system 50
IPLing scenarios 41
managing 32
managing JES2 in a Parallel Sysplex 194
MVS closure 63
normal shutdown 63
partitioning 63
remove z/OS systems 5960
removing system 62
running a standalone dump on a Parallel Sysplex 71
SAD 71
SFM settings 62
shutdown overview 5960
shutdown with SFM active 66, 69
shutdown with SFM inactive 64, 69
stand-alone dump example 503
sysplex cleanup 67, 70
Test Parallel Sysplex 10
TOD clock setting in Parallel Sysplex 43
wait state
hex.0A2 66
Parallel Sysplex over Infiniband (PSIFB) 17
partitioning 63, 69
PATHIN 44, 51, 185
displaying devices 187188
displaying structures 188
recommended addressing 185
PATHOUT 44, 51, 185
recommended addressing 185
peer monitoring 6
pending state 129
persistent structures 134
planned shutdown 63
PlatinumPlex 10
policies
ARM 425
ARM policy management 90
policy
CFRM 21
change 129
pending state 129
information 166
POPULATECF parameter of REBUILD command 154
POSIX compliant, 484
power outage 48
preface xv
processor
failure 443
R
RACF
initialization 47
Index
525
remove structure 133

Rapid Network Reconnect (RNR) 402
Real Time Analysis (RTA) 365
rebuild
system-managed 114
user-managed 114
REBUILD POPULATECF command 154
rebuilding structures 420
REBUILDPERCENT 76
recovering IMS 431
Recovery 404
recovery
record 180
Recovery Control (RECON) data sets 404
recovery location 26
Redbooks author xvii
Redbooks Web site 519
Contact us xvii
remove
CF 125
consoles 304
removing
CICS from a generic resource group 333
JES2 checkpoint structure 132
RACF data sharing 133
signalling structure 133
RESERVE 29, 204
resetting a system 65
resource
contention 30
serialization 5
RESOURCE (RES) command 251
Resource Measurement Facility (RMF) 237
resource monitor (RM) command 246
Resource Recovery Services (RRS) 468
resource structure 415
restart light
DB2 393
REXX 255
RMF
SDSF DA command 237
ROUTCDE 293
ROUTE 299
ROUTE command 299
routing commands around the sysplex 34
RRS
defining to Automatic Restart Manager (ARM) 481
logstream 473
status display 472
troubleshooting 480
two-phase commit 468
S
SAD (see also stand-alone dump) 64
SCA 388
SCHEDULING ENVIRONMENT (SE) command 248
SDSF
DA command 237
JC command 248
MAS panel 247
526
OPERLOG panel example 235

printing output 239
saving output 239
SYSLOG panel example 233
serialization 5
SETXCF 90, 109, 420
SFM
settings 62
shared file system
z/OS UNIX 488
shared queues
IMS 409
shared temporary storage 352
shared-everything 4
shutdown
abnormal shutdown 68
checking that system is removed 68
MVS closure 63
normal shutdown 63
overview 5960
planned 63
removing system 62
running a standalone dump on a Parallel Sysplex 71
SAD 71
SFM settings 62
sysplex cleanup 67, 70
sysplex partitioning 63
wait state
hex.0A2 66
shutting down
z/OS system image 59
signalling
insufficent signalling paths during IPL 56
remove structure 133
signalling path activation in first IPL 44
signalling path
starting and stopping 190
signalling paths 24
signalling problems 193
SIMETRID 55
single points of failure, avoiding
two of everything 3
single system image 7
SNA
Generic Resources 324
SNA MCS (SMCS) consoles 287
staging data sets 478
stalled member 197
standalone dump (SAD) 5960, 71
Star complex 28
Status Update Missing (SUM) 68, 74
STP 5
mode 32
structure 19
allocation 45
cache 369
DB2 368369
duplexing
system-managed 113
list 369
list (SCA) 388
lock 369
NCS 360
rebuilding 420
resource 415
WebSphere MQ 464
structure full monitoring 119
Structure Recovery Data Sets (SRDS) 415
structures 9
as possible cause of IPL problem 56
CF 22, 105
IMS 408
connection state 109
disposition 109
during CFRM initialization 46
generic resources 330
IMS 413
list of 499
managing CF 147
move to alternate CF 129
rebuild failure 161
rebuilding 147
rebuilding in another CF 152
rebuilding in either CF 148
remove JES2 checkpoint structure 132
remove LOGREC 133
remove OPERLOG 133
remove RACF data sharing 133
remove signalling 133
signalling structures during first IPL 44
stopping rebuild 161
that support rebuild, rebuilding of 147
symbol 36
sympathy sickness 76
SYSCONS 291
SYSLOG 133, 233
sysplex 27, 286
console 27
DB2 394
definition 2
environment 2
GoldPlex 10
IMS 406
JES3 273
master console 286
partitioning 63
PlatinumPlex 10
sympathy sickness 76
three-way 11
timer 31
Sysplex Couple Data Set 6
sysplex couple dataset (CDS) 74
Sysplex Distributor 323
introduction 339
Sysplex Failure Management 6

Sysplex Failure Management (SFM) 61, 74
sysplex time management
TOD clock setting in Parallel Sysplex 43
system
affinity 226
group name 300
logger 307
recovery 3
System Display and Search Facility (SDSF) 232
system failure
JES3 actions 276
System Health Checker 252
System Logger 7
system logger 348
address space 309
directory extents 317
displaying system logger status 310
ENQ serialization 317
offload monitoring 316
remove 133
removing structures 133
structure rebuilds 319
System Network Architecture (SNA) 324
SYSTEM RESET 65
system symbol 36
System z file system (zFS) 486
System-managed Coupling Facility (CF)
T
TCP Sysplex Distributor 7
TCP/IP 277, 323
commands 338
introduction 336
test environment
z/VM 11
three-way sysplex 11
time
common 4
consistency 5
time zone
offset setting 43
Time-Of-Day (TOD)
clock setting 43
Tivoli Enterprise Portal 305
transaction processing (TP) 346
transaction routing 348
transport class 192
troubleshooting
RRS 480
TSO
deregister from a generic resource group 334
monitoring a sysplex 36
two-phase commit 468
U
ULOG command 235
Index
527
UNIX shell environment (OMVS) 484

user-managed duplexing 371
USRJRNL 350
V
value-for-money 4
VARY command
V 4020,ONLINE,UNCOND 194
VARY XCF 74
Virtual IP Address (VIPA), dynamic
dynamic VIPA 339
Virtual Telecommunications Access Method (VTAM) 323
volatility 478
VSAM
cache structures 414
VTAM 347
CF 326
Generic Resources 324
VTAM Generic Resources 7
VTAM in a Parallel Sysplex
CICS generic resources 333
commands for generic resources 330
deregister CICS from a generic resource group 333
deregister TSO from a generic resource group 334
determine status of generic resources 330
managing CICS generic resources 333
remove CICS from a generic resource group 333
W
WADS data sets 404
WebSphere MQ
Automatic Restart Manager (ARM) 466
commands 462
introduction 456
ISPF panels 461
monitoring 461
structure 464
WEIGHT 75
WLM
policies 85
Work Context 468
Workload Manager (WLM) 7, 366
X
XCF 6, 489
CICS 346
commands 14
connectivity, unable to establish 56
initialization 40
signalling 184
signalling paths 24
signalling services 6, 194
stalled member detection 197
starting 46
XCF initialization in IPL 44
XCF initialization restarted in failed IPL 52
XCF/MRO 347
528
XES 102, 113, 416
Z
z/OS
system log 233
z/OS Health Checker 258
z/OS Management Console 304
z/OS UNIX
files 487
introduction 484
z/VM
test environment 11
zFS 486
administration 489


(1.0 spine)
0.875<->1.498
460 <-> 788 pages


Back cover

Understanding
Parallel Sysplex
This IBM Redbooks publication is a major update to the

Parallel Sysplex Operational Scenarios book, originally
published in 1997.
Handbook for sysplex

management
The book is intended for operators and system

programmers, and is intended to provide an understanding
of Parallel Sysplex operations. This understanding,
together with the examples provided in this book, will help
you effectively manage a Parallel Sysplex and maximize its
availability and effectiveness.
Operations best
practices
The book has been updated to reflect the latest sysplex

technologies and current recommendations, based on the
experiences of many sysplex customers over the last 10
years.
It is our hope that readers will find this to be a useful
handbook for day-to-day sysplex operation, providing you
with the understanding and confidence to expand your
exploitation of the many capabilities of a Parallel Sysplex.
Knowledge of single-system z/OS operations is assumed.
This book does not go into detailed recovery scenarios for
IBM subsystem components, such as CICS Transaction
Server, DB2 or IMS. These are covered in great depth in
other Redbooks publications.
INTERNATIONAL
TECHNICAL
SUPPORT
ORGANIZATION
BUILDING TECHNICAL
INFORMATION BASED ON
PRACTICAL EXPERIENCE
IBM Redbooks are developed
by the IBM International
Technical Support
Organization. Experts from
IBM, Customers and Partners
from around the world create
timely technical information
based on realistic scenarios.
Specific recommendations
are provided to help you
implement IT solutions more
effectively in your
environment.
For more information:

ibm.com/redbooks
SG24-2079-01
ISBN 0738432687

Z Os

Uploaded by

Copyright:

Available Formats

Z Os

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Z Os

Uploaded by

Copyright:

Available Formats

What is the purpose of this document?

What is the purpose of this document?

What topics are covered in the document?

What topics are covered in the document?

Front cover

IBM z/OS Parallel Sysplex

Handbook for sysplex

International Technical Support Organization

Second Edition (May 2009)

Copyright International Business Machines Corporation 2009. All rights reserved.

Chapter 3. IPLing systems in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 4. Shutting down z/OS systems in a Parallel Sysplex . . . . . . . . . . . . . . . . . . .

Chapter 5. Sysplex Failure Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 6. Automatic Restart Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

IBM z/OS Parallel Sysplex Operational Scenarios

6.4 ARM policy management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Chapter 8. Couple Data Set management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Starting and stopping a policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 9. XCF management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 10. Managing JES2 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 11. System Display and Search Facility and OPERLOG . . . . . . . . . . . . . . . .

IBM z/OS Parallel Sysplex Operational Scenarios

11.4.1 Example of the DISPLAY ACTIVE panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 12. IBM z/OS Health Checker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 13. Managing JES3 in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 14. Managing consoles in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . . . . . .

14.6.2 Using the ROUTE command. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 15. z/OS system logger considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 16. Network considerations in a Parallel Sysplex . . . . . . . . . . . . . . . . . . . . .

Chapter 17. CICS operational considerations in a Parallel Sysplex. . . . . . . . . . . . . .

IBM z/OS Parallel Sysplex Operational Scenarios

17.5.1 Initiating use of a shared TS structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 18. DB2 operational considerations in a Parallel Sysplex . . . . . . . . . . . . . .

18.9 Using DB2 lock structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 19. IMS operational considerations in a Parallel Sysplex . . . . . . . . . . . . . . .

IBM z/OS Parallel Sysplex Operational Scenarios

19.9.10 Recovering from an IRLM failure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 20. WebSphere MQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 21. Resource Recovery Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Chapter 22. z/OS UNIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Appendix A. Operator commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491

IBM z/OS Parallel Sysplex Operational Scenarios

Copyright IBM Corp. 2009. All rights reserved.

The following terms are trademarks of other companies:

IBM z/OS Parallel Sysplex Operational Scenarios

The team that wrote this book

Copyright IBM Corp. 2009. All rights reserved.

IBM z/OS Parallel Sysplex Operational Scenarios

Become a published author

IBM z/OS Parallel Sysplex Operational Scenarios

Copyright IBM Corp. 2009. All rights reserved.

1.1 Introduction to the sysplex environment

1.2 What is a sysplex

IBM z/OS Parallel Sysplex Operational Scenarios

Figure 1-1 Components of a Parallel Sysplex

1.2.1 Functions needed for a shared-everything environment

IBM z/OS Parallel Sysplex Operational Scenarios

Communication within the sysplex

IBM z/OS Parallel Sysplex Operational Scenarios