HBA Troubleshooting Guide

53-1000885-02 January 30, 2009
Fibre Channel HBA

Troubleshooting Guide
Supporting Models 415, 425, 815, 825
Copyright 2006-2009 Brocade Communications Systems, Inc. All Rights Reserved. Brocade, Fabric OS, File Lifecycle Manager, MyView, and StorageX are registered trademarks and the Brocade B-wing symbol, DCX, and SAN Health are trademarks of Brocade Communications Systems, Inc., in the United States and/or in other countries. All other brands, products, or service names are or may be trademarks or service marks of, and are used to identify, products or services of their respective owners. Notice: This document is for informational purposes only and does not set forth any warranty, expressed or implied, concerning any equipment, equipment feature, or service offered or to be offered by Brocade. Brocade reserves the right to make changes to this document at any time, without notice, and assumes no responsibility for its use. This informational document describes features that may not be currently available. Contact a Brocade sales office for information on feature and product availability. Export of technical data contained in this document may require an export license from the United States government. The authors and Brocade Communications Systems, Inc. shall have no liability or responsibility to any person or entity with respect to any loss, cost, liability, or damages arising from the information contained in this book or the computer programs that accompany it. The product described by this document may contain open source software covered by the GNU General Public License or other open source license agreements. To find-out which open source software is included in Brocade products, view the licensing terms applicable to the open source software, and obtain a copy of the programming source code, please visit http://www.brocade.com/support/oscd.
Brocade Communications Systems, Incorporated

Corporate and Latin American Headquarters Brocade Communications Systems, Inc. 1745 Technology Drive San Jose, CA 95110 Tel: 1-408-333-8000 Fax: 1-408-333-8101 E-mail: [email protected] Asia-Pacific Headquarters Brocade Communications Singapore Pte. Ltd. 30 Cecil Street #19-01 Prudential Tower Singapore 049712 Singapore Tel: +65-6538-4700 Fax: +65-6538-0302 E-mail: [email protected]
European Headquarters Brocade Communications Switzerland Srl Centre Swissair Tour B - 4me tage 29, Route de l'Aroport Case Postale 105 CH-1215 Genve 15 Switzerland Tel: +41 22 799 5640 Fax: +41 22 799 5641 E-mail: [email protected]
Document History
Title
Fibre Channel HBA Troubleshooting Guide Fibre Channel HBA Troubleshooting Guide
Publication number
53-1000885-01 53-1000885-02
Summary of changes
New document Revised with corrections.
Date
December 2008 January 2009
Contents
About this Document

In this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii How this document is organized . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Supported hardware and software . . . . . . . . . . . . . . . . . . . . . . . . . . viii Whats new in this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Document conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Notice to the reader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Additional information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Getting technical help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Document feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Chapter 1
Introduction to troubleshooting
In this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 How to use this manual for troubleshooting . . . . . . . . . . . . . . . . . . . . 1 Gathering problem information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Chapter 2
Isolating Problems
In this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 General problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Resolving installation problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Verifying installation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Errors when installing driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Installer program does not autorun from CD (Windows only) . . 13 Files needed for bfad.sys message appears when removing driver 13 Cannot roll back driver on all HBA instances using Device Manager 14 Host not booting from remote LUN . . . . . . . . . . . . . . . . . . . . . . . 14 Confirming driver package installation. . . . . . . . . . . . . . . . . . . . 14 Host system freezes or crashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 HCM GUI fails to connect with HCM agent . . . . . . . . . . . . . . . . . . . . 17 Verifying Fibre Channel links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Chapter 3
Tools for Collecting Data

In this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02
iii
For detailed information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Data to provide support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Collecting data using host system commands . . . . . . . . . . . . . . . . . 24 Collecting data using BCU commands and HCM . . . . . . . . . . . . . . . 25 Using Support Save . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 HBA data collection using HCM . . . . . . . . . . . . . . . . . . . . . . . . . 28 Collecting data using BCU commands . . . . . . . . . . . . . . . . . . . . 28 Collecting data using Fabric OS commands . . . . . . . . . . . . . . . . . . . 28 Event logs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Host system logs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 HCM logs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Syslog support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Windows Event Log support . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Port statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 IOC statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Fabric statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Remote port statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 FCIP initiator mode statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Logical port statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Virtual port statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Quality of service (QoS) statistics . . . . . . . . . . . . . . . . . . . . . . . . 39 Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Beaconing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Loopback tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 PCI loopback test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Memory test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 HBA temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Ping end points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Trace route . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Echo test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 SCSI test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Collecting SFP data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 SFP diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Port power on management (POM) . . . . . . . . . . . . . . . . . . . . . . 46 Collecting port data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Base port properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Remote port properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Logical port properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Virtual port properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Port log . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Port list . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Port query . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Port speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Authentication settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Displaying settings through HCM . . . . . . . . . . . . . . . . . . . . . . . . 48 Displaying settings through BCU. . . . . . . . . . . . . . . . . . . . . . . . . 48
iv
QoS and target rate limiting settings . . . . . . . . . . . . . . . . . . . . . . . . . 48 BCU commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 HCM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Persistent binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Chapter 4
Performance optimization
In this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Linux tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Solaris tuning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Windows tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Driver tunable parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 OS tunable parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 VMware tuning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Index
vi
About this Document
In this chapter
How this document is organized . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Supported hardware and software. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Whats new in this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Document conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Notice to the reader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Additional information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Getting technical help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Document feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
How this document is organized

This document is organized to help you find the information that you want as quickly and easily as possible. The document contains the following components:
Chapter 1, Introduction to troubleshooting provides an introduction and approach to

troubleshooting HBA problems, as well as tips for gathering problem information. A checklist is also provided to verify that required procedures have been followed during installation.
Chapter 2, Isolating Problems provides information on common HBA problems and

procedures to diagnose and recover from these problems
Chapter 3, Tools for Collecting Data provides a summary of diagnostic and monitoring tools
available through the HCM, Brocade Command Line Utility (BCU), fabric switch operating system and host system to help you isolate and resolve HBA-related problems.
Chapter 4, Performance optimization contains guidelines for optimizing HBA performance on

your host system. This publication is a companion guide to be used with the Brocade Fibre Channel HBA Administrators Guide. That publication provides detailed information on HBA monitoring and diagnostic tools in Host Connectivity Manager (HCM) and the BCU.
NOTE
vii
Supported hardware and software

This section describes HBA hardware and software support.
HBA support
The following Fibre Channel host bus adapters (HBAs) are supported in this release.
Brocade 815. Single-port HBA with a per-port maximum of 8 Gbps using an 8 Gbps SFP+. Brocade 825. Dual-port HBA with a per-port maximum of 8 Gbps using an 8 Gbps SFP+. Brocade 415. Single-port HBA with a per-port maximum of 4 Gbps using a 4 Gbps SFP. Brocade 425 Dual-port HBA with a per-port maximum of 4 Gbps using a 4 Gbps SFP.
Notes:
This publication only supports the HBA models listed above and does not provide information
about the Brocade 410 and 420 Fibre Channel HBAs, also known as the Brocade 400 Fibre Channel HBAs.
Although you can install an 8 Gbps SFP+ into a Brocade 415 or 425 HBA, only 4 Gbps
maximum port speed is possible.
Install only Brocade-branded SFPs in these HBAs.
Fabric OS and switch support

For a current list of compatible servers, switches, storage, and applications, refer to the Brocade HBA web site at www.brocade.com/hba.
Host operating system support

The following operating systems support Brocade Host Connectivity Manager (HCM), Brocade Command Line Utility (BCU), and HBA drivers:
Windows Server 2003, version R2 with SP2 Windows Server 2008 Windows NT (HCM support only) Windows 2000 (HCM support only) Linux RHEL4, RHEL5, SLES9, and SLES10 Solaris 10 (x86 and SPARC) VMware ESX Server 3.5
NOTE
Drivers, BCU, and HCM Agent are supported only on the VMware console Operating System. HCM is supported only on the guest operating system on VMware.
Windows Vista (HCM only) Windows XP (HCM only)
viii
Specific operating system service pack levels and other patch requirements are detailed in the current HBA release notes.
NOTE
Whats new in this document

This is a new document. For further information about new features not covered in this document and documentation updates for this release, refer to the HBA release notes.
Document conventions
This section describes text formatting conventions and important notice formats used in this document.
Text formatting
The narrative-text formatting conventions that are used are as follows: bold text Identifies command names Identifies the names of user-manipulated GUI elements Identifies keywords and operands Identifies text to enter at the GUI or CLI Provides emphasis Identifies variables Identifies paths and Internet addresses Identifies document titles Identifies CLI output Identifies command syntax examples
italic text
code text
For readability, command names in the narrative portions of this guide are presented in mixed lettercase: for example, switchShow. In actual examples, command lettercase is often all lowercase. Otherwise, this manual specifically notes those cases in which a command is case sensitive.
ix
Command syntax conventions

Command syntax in this manual follows these conventions: command
--option, option -argument, arg
Commands are printed in bold. Command options are printed in bold. Arguments. Optional element. Variables are printed in italics. In the help pages, values are underlined or enclosed in angled brackets < >. Repeat the previous element, for example member[;member...] Fixed values following arguments are printed in plain font. For example, --show WWN Boolean. Elements are exclusive. Example: --show -mode egress | ingress
[] variable ... value |
Notes, cautions, and warnings

The following notices and statements are used in this manual. They are listed below in order of increasing severity of potential hazards. A note provides a tip, guidance or advice, emphasizes important information, or provides a reference to related information.
NOTE
ATTENTION
An Attention statement indicates potential damage to hardware or data.
CAUTION A Caution statement alerts you to situations that can be potentially hazardous to you or cause damage to hardware, firmware, software, or data.
DANGER A Danger statement indicates conditions or situations that can be potentially lethal or extremely hazardous to you. Safety labels are also attached directly to products to warn of these conditions or situations.
Key terms
For definitions specific to Brocade and Fibre Channel, see the technical glossaries on Brocade Connect. See Brocade resources on page xi for instructions on accessing Brocade Connect. For definitions specific to this document, see the Brocade Fibre Channel HBA Installation and Reference Manual.
For definitions of SAN-specific terms, visit the Storage Networking Industry Association online dictionary at: http://www.snia.org/education/dictionary
Notice to the reader

This document may contain references to the trademarks of the following corporations. These trademarks are the properties of their respective companies and corporations. These references are made for informational purposes only.
Corporation
Microsoft Corporation
Referenced Trademarks and Products

Windows, Windows Server 2003, Windows Server 2008, Vista, XP, PE for Windows, Hyper V for Windows, Windows Automated Installation Kit (WAIK) Solaris Red Hat Enterprise Linux (RHEL) SUSE Linux Enterprise Server (SLES) ESX Server SPARC
Sun Microsystems, Inc. Red Hat Inc. Novell, Inc VMware Inc. SPARC International, Inc
Additional information
This section lists additional Brocade and industry-specific documentation that you might find helpful.
Brocade resources
For HBA resources, such as product information, software, firmware, and documentation, visit the Brocade HBA web site at www.brocade.com/hba. To get up-to-the-minute product information, join Brocade Connect. Go to http://www.brocadeconnect.com to register at no cost for a user ID and password. For practical discussions about SAN design, implementation, and maintenance, you can obtain Building SANs with Brocade Fabric Switches through: http://www.amazon.com For additional Brocade documentation, visit the Brocade Web site: http://www.brocade.com
xi
Other industry resources

For additional resource information, visit the Technical Committee T11 Web site. This Web site provides interface standards for high-performance and mass storage applications for Fibre Channel, storage management, and other applications: http://www.t11.org For information about the Fibre Channel industry, visit the Fibre Channel Industry Association Web site: http://www.fibrechannel.org
Getting technical help

Contact your Brocade HBA support supplier for hardware, firmware, and software support, including product repairs and part ordering. Provide the following information: 1. General information:
Brocade HBA model number Host operating system version Software name and software version, if applicable syslog message logs bfa_supportsave output. To expedite your support call, use the bfa_supportsave feature to collect debug information from the driver, internal libraries, and firmware. You can save valuable information to your local file system and send it to support personnel for further investigation. For details on using this feature, refer to Using Support Save on page 25.
Detailed description of the problem, including the switch or fabric behavior immediately
following the problem, and specific questions.
Description of any troubleshooting steps already performed and the results.

2. HBA serial number: The HBA serial number and corresponding bar code are provided on the serial number label illustrated below. The serial number label is located on the end of the HBA opposite from the SFP receivers. *FT00X0054E9* FT00X0054E9 You can also display the serial number through the following HCM dialog boxes and BCU commands:
HBA Properties tab in HCM.

Select an HBA in the device tree, then click the Properties tab in the right pane.
BCU adapter --list command.
xii
NOTE
For details on using HCM and BCU commands, refer to the Brocade Fibre Channel HBA Administrators Guide. 3. Port World-Wide Port Name (PWWN). Determine this through the following resources:
Label located on the end of the HBA opposite the SFP receiver slots. This label provides
the WWPN for each port.
Brocade BIOS Configuration Utility.

Select the appropriate HBA port from the initial configuration utility screen, then select Adapter Settings to display the WWNN and PWWN for the port. For details, refer to the Brocade Fibre Channel HBA Installation and Reference Manual.
HCM Port Properties dialog box.

This dialog box displays the PWWN for each HBA port. To display this dialog box, click a port in the device tree, then click the Properties tab in the right pane.
The following BCU commands:

adapter --query <ad_id>
This command displays HBA information. The <ad_id> parameter is the HBAs serial number.
port --list <ad_id>
This command lists all the physical ports on the HBA along with their basic attributes. The <ad_id> parameter is the HBAs serial number.
Document feedback
Quality is our first concern at Brocade and we have made every effort to ensure the accuracy and completeness of this document. However, if you find an error or an omission, or you think that a topic needs further development, we want to hear from you. Forward your feedback to: [email protected] Provide the title and version number of the document and as much detail as possible about your comment, including the topic heading and page number and your suggestions for improvement.
xiii
xiv
Chapter
Introduction to troubleshooting
In this chapter
How to use this manual for troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Gathering problem information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
How to use this manual for troubleshooting

A Fibre Channel HBA is one component in a SAN consisting of Fibre Channel switches, storage devices, and the fiber optic cabling and connections for these components. Although there may be a problem in the HBA or an HBA component, such as a small form factor pluggable (SFP) optical transceiver, the problem could also originate in another SAN component or connections between these components. Before removing and replacing HBAs, launching HBA diagnostics, or even gathering statistics on HBA operation, it is important that you perform the following tasks: 1. Fully describe the problem and gather complete information about the symptoms that suggest a problem exists. Refer to Gathering problem information on page 2. 2. Isolate or resolve the problem by first using information in Chapter 2, Isolating Problems.
First study information in the general HBA problem symptoms, causes, and fixes or actions
in Table 2 on page 6. This table should provide help on many general problems that you may encounter with HBA operation. Fixes and actions often reference the BCU commands, HCM features, and host operating system commands described in Chapter 3, Tools for Collecting Data to gather data for problem isolation or resolution.
If you still require more information to isolate problems, use the following sections in
Chapter 3. Note that these sections are referenced from Table 2 (Troubleshooting General Problems) when appropriate to further isolate problems. Resolving installation problems on page 12. Host system freezes or crashes on page 16. Verifying Fibre Channel links on page 20. 3. Use the BCU commands, HCM features, and host operating system commands described in Chapter 2, Isolating Problems to gather data to help you isolate problems. Although many of these tools are specifically referenced as actions for problems described in Table 2 (Troubleshooting General Problems) in Chapter 2, many more are included that can provide helpful data, such as event logs, operating statistics, and diagnostics. Note that Table 5 on page 24 in Chapter 3 provides a list of useful host system commands for each supported operating system that you can use to gather data. 4. Consider these factors when isolating and resolving the problem:
Can the issue be resolved by using the latest supported combination of host system BIOS,
operating system, operating system updates, or HBA drivers?
Gathering problem information
Does the issue persist when the HBA is installed in a different platform or is connected
using a different switch port, SFP, and cable?
Can this problem be reproduced on one or more HBAs, port, or host system? Can you
identify specific steps that consistently reproduce this problem on one or more hosts?
Is the problem documented in release notes for the HBA, operating system, or host system
BIOS?
Is the problem documented in release notes for the switch and target storage system? Is unexpected behavior intermittent or always present?
If the problem is in a Fibre Channel switch, cabling, storage device, or in connectivity between these components, refer to documentation, help systems, or service providers of that equipment. 5. If you cannot resolve the problem, gather and provide problem information to your HBA support provider for resolution.

Perform the following tasks to obtain as much information as possible before contacting technical support. Be sure to take careful notes for use as a record and reference.
Describe the symptoms that you are observing. Be specific. Here are some examples: - User experiences, such as slow performance or file access. - LEDs not functioning on an HBA port that is connected to the fabric. - All LEDs on HBA port flashing amber. - Expected storage devices not visible from the HCM or host systems storage management
application.
HBA not recognized by host system BIOS. HBA not recognized as PCI device by host system operating system.
What happened prior to the observed symptoms? Describe all observed behavior that is unexpected and compare against expected behavior. Gather information for support. - Use appropriate tools on storage targets to gather information such as disk, tape, and
controller model and firmware levels.
Run the bfa_supportsave BCU command on the host system and save output to a file on your system. This command captures all driver, internal libraries, firmware, and other information needed to diagnose suspected system issues.You can save captured information to the local file system and send it to support personnel for further investigation.
Run the Fabric OS supportSave command on any Brocade switch and save output. This command collects RASLOG, TRACE, supportShow, core file, FFDC data and other support information.
For details on using the Support Save feature, refer to Using Support Save on page 25.
Draw a topology map of the SAN from the HBAs to the storage targets. Include the following:
TABLE 1
Component
HBA
Topology map details

How to identify
Model, World-Wide Name (WWN), and driver release level. Model, WWN, and Fabric OS version. Port WWNs connected to all links. Model and hardware revision.
Fibre Channel switches. Fiber optic links between HBA, switches, and storage ports. Host hardware
The bfa_supportSave and FOS supportsave commands can provide current information for the topology map. Also, consider using the Brocade SAN Health products to provide information on your SAN environment, including an inventory of devices, switches, firmware versions, and SAN fabrics, historical performance data, zoning and switch configurations, and other data. Click the Support tab on www.brocade.com for more information on these products.
Run appropriate diagnostic tools for storage targets. Use additional HCM, BCU, host system, and Fabric OS commands summarized in Chapter 3,
Tools for Collecting Data to gather statistics and problem data on the HBA, host, Fibre Channel links, and connected devices.
Determine what has changed in the SAN. For example, if the SAN functioned without problems
before installing the HBA, then the problem is most likely in the HBA installation or configuration, HBA hardware, or HBA driver package. Other examples to investigate could be changes in the switch or storage system firmware, an offline switch, or a disconnected or faulty cable between the HBA, switch, or storage controller fiber optic ports.
Record the time and frequency of symptoms and the period of time symptoms have been
observed.
Determine if unexpected behavior is intermittent or always present. List steps that have been taken to troubleshoot the problem, including changes attempted to
isolate the problem.
Chapter
Isolating Problems
In this chapter
General problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Resolving installation problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Host system freezes or crashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 HCM GUI fails to connect with HCM agent. . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Verifying Fibre Channel links. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
General problems
Table 2 on page 6 describes general problems related to HBA operation, possible causes, and recommended actions that may fix the problem. Recommended actions may refer you to information in the following locations as appropriate to gather information to further isolate and resolve the problem.
Chapter 3, Tools for Collecting Data in this manual.

Contains procedures to perform HBA diagnostics, display HBA statistics and event logs, and collect data for troubleshooting using BCU commands, HCM options, Fabric OS commands, and your host system commands.
Fabric OS Administrators Guide

Provides detailed information on features available on Brocade storage area network (SAN) products, and how to configure and administer these products
Fabric OS Command Reference Manual.

Provides detailed descriptions of command line interface commands to help system administrators and technicians operate, maintain, and troubleshoot Brocade SAN products.
Fabric OS Troubleshooting and Diagnostic Guide.

Provides help with isolating problems in other Brocade SAN components.
Your hosts operating system documentation and help system.

Provides details on commands for gathering information and isolating problems.
2
TABLE 2
Symptom
General problems
Troubleshooting general problems

Possible Cause
1 2 3 Card not seated properly. Server slot issues. HBA not compatible with host operating system or connected storage systems.
Fix or Action
1 Execute host operating system command to list PCI devices. Refer to the List PCI Devices row in Table 5 on page 24. If the HBA is not listed, perform the following steps. Reseat the HBA. Replace the HBA with an HBA in known working condition to determine whether there is a slot malfunction. Verify compatibility by reviewing the Brocade Server Connectivity Compatibility Matrix. To find this document, log into Brocade Connect on www.brocade.com, then select the Compatibility Information quick link under Documentation Library. Execute your hosts operating system command to list PCI devices. Refer to the List PCI Devices row in Table 5 on page 24. If the HBA is not listed in the output from this command, go on to the next step. Refer to HBA not reported under servers PCI sub-system under the Symptom column in this table. The HBA driver may not be loaded. Refer to Confirming driver package installation on page 14 for methods to verify driver installation.
HBA not reported under servers PCI subsystem.
2 3
No HBAs reported when using the adapter --list BCU command:
1 2
HBA not reported under servers PCI sub-system. HBA driver not loaded.
General problems
TABLE 2
Symptom

Possible Cause
1 2 3 4 5 SFP or cable problems. Link between HBA and switch is down. Switch port is disabled or switch is disabled. HBA port speed or topology mis-match with the switch port. Non-Brocade branded SFP installed.
Fix or Action
1 Ensure that the SFPs and cables are connected properly on both HBA and switch side. Check for any cable damage. Verify HBA side link status by executing the BCU port --list command. Check the FC Addr field for an address and the State field for Linkup. For details on using this command, refer to Port list on page 47. Execute either the Fabric OS switchShow or portShow commands on the attached switch to ensure that the switch or individual port is not disabled or offline. Check the port topology setting on the switch using the Fabric OS portCfgShow command to ensure that Locked L_Port is OFF. Use the portCfgLport command to change the setting to OFF if required. Check the switch port speed using the Fabric OS portCfgShow command to verify that Speed is either AUTO or matches the speed of the attached HBA port (for example, the speed setting for both ports is 4 Gbps). Check port speed on the HBA with the BCU port --list or port --query commands to display the current and configured speed. Refer to Port speed on page 48 and Port query on page 47 for details on using these commands. If non-brocade branded SFPs are inserted on the HBA side or 8 Gbps switch or director, the port link will not come up. On the switch, execute the Fabric OS switchShow command to verify that Mod_Inv (invalid module) does not display for the port state.
Port link does not come up.
2
TABLE 2
Symptom
General problems

Possible Cause Fix or Action
On the HBA, execute the port --list and port --query BCU commands. Refer to Port list on page 47 and Port query on page 47. If an unsupported SFP is detected, the Sfp field displays us (unsupported SFP) for port --list and the Media field displays Unsupported SFP for port --query. For additional actions and fixes for the port link not coming up, refer to Verifying Fibre Channel links on page 20.
Port link does not come up (continued)
Loss of sync and loss of signal errors in port statistics (refer to Port statistics on page 34).
Possible physical link problem.
Ensure that the SFPs and cables are connected properly on both HBA and switch side. Check for any cable damage. Refer to Verifying Fibre Channel links on page 20
Fabric authentication failures
Authenticating configuration is incorrect.
Check authentication settings on the switch and HBA. For the switch, execute the authutil --show Fabric OS command. For the HBA, execute the BCU auth --show command (refer to Authentication settings on page 48). Use the BCU auth --show <port> command on the HBA and Fabric OS authutil --show command on switch. Check the shared secret configuration on the attached switch and HBA. For the switch, execute the secAuthSecret Fabric OS command. For the HBA, execute the auth -secret BCU command (refer to Authentication settings on page 48) for details on using the auth-secret command.
General problems
TABLE 2
Symptom

Possible Cause
1 2 3 4 No fabric connectivity between HBA and target or target is not online. The target and the HBA are not on the same zone. The HBA driver is not loaded. There is a problem with remote port.
Fix or Action
1 Execute the Fabric OS nsAllShow command on the attached switch to verify that the target and the host are online in the fabric and registered in the name server. Execute the Fabric OS cfgActvShow command on the attached switch and verify that the host and target are in the same zone (either using domain area members, port area members, or port or node WWNs). The HBA driver may not be loaded. Refer to Confirming driver package installation on page 14 for methods to verify driver installation. Verify that the remote target port (rport) is reporting itself online by comparing rport online and rport offline statistics (refer to Remote port statistics on page 36). The rport online counter should be one greater than the rport offline counter. If not, clear the counters and try connecting to the remote port again. Verify the rport online and rport offline statistics again. Check LUN mapping and masking using storage array configuration tools. The HBA driver may not be loaded. Refer to Confirming driver package installation on page 14 for methods to verify driver installation.
Target not visible.
LUN not visible
1 2
Missing or improper storage array LUN masking setting. HBA driver not loaded.
I/Os are not failing over immediately on a path failure in MPIO setup.
Improper driver mpiomode setting.
Execute the port --query <port_id> BCU command and ensure fcpim MPIO mode is enabled (which implies zero Path TOV values) or that fcpim MPIO mode is disabled with the expected Path TOV settings (default is 10 seconds). Execute the Fabric OS configure command on the attached switch and change the Maximum logins per port parameter under the F_Port login parameters menu to increase the maximum NPIV I/Ds allowed per port.
Unable to create more than 126 Virtual (NPIV) ports.
The switchs maximum NPIV limit has been reached.
On Linux, the maximum IOPS numbers are very low. On VMware, the maximum IOPS numbers are very low.
The amount of disk I/O requests are Refer to Linux tuning on page 51 for causing low throughput and high latency. suggestions to optimize HBA performance in Linux systems. The amount of disk I/O requests are Refer to VMware tuning on page 53 for causing low throughput and high latency. suggestions to optimize HBA performance in Vmware systems.
2
TABLE 2
Symptom
General problems

Possible Cause
1 2 QoS not enabled on both switch and HBA port. QoS zones not created properly on switch for high, medium, and low priority targets.
Fix or Action
1 Verify if QoS is enabled for an HBA port using the qos -query <port_id> BCU command. Verify if it is enabled on the switch using the islShow command. Verify zones on the switch using cfgActvShow command. Verify that QoS is configured on switch using instructions in the Brocade Fabric OS Administrators Guide.
QoS performance issues
2 3
The HBA is not showing in the fabric.
There is a problem in the fabric or a protocol issue between the HBA and fabric.
Check fabric statistics. Refer to Fabric statistics on page 36 for methods to display fabric statistics for the HBA. If counts for FLOGI sent and FLOLGI accept fabric statistics do not match, suspect fabric problem or protocol issue between HBA and fabric. If fabric offline counts increase and fabric maintenance is not occurring, this may indicate a serious fabric problem. Refer to your switch troubleshooting guide. Refer to Errors when installing driver on page 13 for more information to isolate this problem. Refer to Installer program does not autorun from CD (Windows only) on page 13 for more information to isolate this problem. Refer to Host not booting from remote LUN on page 14 for more information to isolate this problem. Refer to Verifying Fibre Channel links on page 20 for more information to isolate this problem.
Errors when installing bfa_driver_linux--<version>.noarch.rpm driver package. Installer program does not autorun (Windows only).
Appropriate distribution kernel development packages are not installed on your host system for the currently running kernel. Autorun is not enabled on your system.
Host not booting from remote LUN.
Boot from LUN not configured correctly.
Host system freezes or crashes.
1 2 3 4
Fault fiber optic cabling and connections. Faulty or unseated SFPs or unsupported SFPs. Conflicts with port operating speed or topology of attached devices. HBA not compatible with host system. Problem in the fabric or a protocol issue between the HBA and fabric. NPIV is not supported or is disabled on the switch
Virtual devices not listed in name server.
Check virtual port statistics, such as FDISK sent, FDISK accept, and No NPIV support statistics. Refer to Virtual port statistics on page 38 for methods to display virtual port statistics. Refer to Confirming driver package installation on page 14 for methods to verify driver installation.
Operating system errors (blue screen).
HBA driver not loaded.
10
General problems
TABLE 2
Symptom

Possible Cause Fix or Action
The client application cannot connect to Refer to HCM GUI fails to connect with HCM agent on page 17 to isolate and the HCM Agent listening on the configured port for the following reasons: resolve the problem. The agent is not running. The agent not accepting connections on the expected port. The agent is not listening on the expected port. Communication between the client and agent is blocked by a firewall preventing access to the port (usually only a consideration for remote HCM management). 1 2 3 HBA cannot log in to the name server. HBA has a problem registering with the name server. HBA has a problem querying the name server for available storage. Display logical port statistics (refer to Logical port statistics on page 37. Check for the following errors: 1 Increasing name server port login (NS PLOGI) error rejects and unknown name server port login response (NS login unknown rsp) errors means that HBA most likely cannot log in to the name server. 2 Increasing errors of the following types indicates that the HBA has a problem registering with the name server. Name server register symbolic port name identifier (NS RSPN_ID) errors. Name server register symbolic port name identifier response (NS RFT_ID rsp) errors Name server register symbolic port name identifier response rejects (NS RFT_ID rejects). 3 Increasing name server get all port ID response (NS GID_FT rsp), rejects (NS_GID FT rejects), or unknown responses (NS_GID FT unknown rsp) means that the HBA has a problem querying the name server for available storage.
Failed to connect to agent on host... error when using HCM.
The HBA not registering with the name server or cannot access storage.
Driver event messages appearing in host system log files.
Various causes and severity levels.
Follow recommended action in message. Resolve critical-level messages and multiple major or minor-level messages relating to the same issue as soon as possible. For details on event messages, refer to Event logs on page 30.
11
2
TABLE 2
Symptom
Resolving installation problems

Possible Cause
Installing the driver using the driver installer program (bfa_installer.exe) or Brocade Software Installer (GUI or command-based application), then subsequently removing the driver using the Device Manager.
Fix or Action
Do not uninstall the driver using the Device Manager if you have used the Brocade installer programs to install driver instances. Always use the Brocade installer programs to remove the driver. Refer to Files needed for bfad.sys message appears when removing driver on page 13 for more information.
Files needed for bfad.sys message appears when removing driver
Cannot roll back driver on all HBA instances using Device Manager
Installing the driver using the Brocade driver installer program (bfa_installer.exe) or Software Installer (GUI or command-based application), then rolling back driver HBA instances using the Device Manager.
Install the driver for each HBA instances using the Device Manager, then roll back the driver using Device Manager. Use the driver installer program (bfa_installer.exe) or Brocade Software Installer (GUI or command-based application) to install or upgrade the driver, then use the Brocade Software Uninstaller to roll back drivers on all HBA instances in one-step. Refer to Cannot roll back driver on all HBA instances using Device Manager on page 14 for more information.
If troubleshooting actions in Table 2 do not resolve problems, check the installed version of the HBA (chip revision) and driver (fw version) using the adapter --query BCU command. To use this command, refer to Collecting data using BCU commands on page 28. Refer to release notes posted on the Brocade HBA web site (www.brocade.com/hba) for known problems relating to the HBA and driver versions.
NOTE

This section provides a summary of common HBA problems that may occur during or immediately after installing HBAs and describes methods to resolve these problems.
Verifying installation
Problems with HBA operation may be due to improper hardware or software installation, incompatibility between the HBA and your host system, unsupported SFPs installed on the HBA, improper fiber optic cable connected to the fabric, or the HBA not operating within specifications. Determine if problems may exist because of these factors by reviewing your installation with information in the Brocade Fibre Channel HBA Installation and Reference Manual listed in Table 3.
12
TABLE 3
Information
Installation and Reference Manual references

Chapter
Product Overview Product Overview Installation Specifications
Hardware and software compatibility information. Software installation packages supported by host operating system and platforms. Hardware and software installation instructions. Product specifications.
Errors when installing driver

If you are installing the bfa_driver_linux--<version>.noarch.rpm package, the driver module compiles on the system during installation. If driver build errors result when you install the package, verify that the appropriate distribution kernel development packages are installed on your host system for the currently running kernel. These should include the gcc compiler and the kernel sources. If these are not installed, you may need to reinstall the operating system before continuing installation. Be sure to install everything including the developer library options.
Installer program does not autorun from CD (Windows only)

When installing software from a CD created from the ISO image downloaded from www.brocade.com/hba, and installer program does not autorun, make sure that autorun is enabled on your system. Enable autorun if possible and reinsert the CD. Try manually executing the LaunchWinSetup.cmd located on the installation the CD.
Files needed for bfad.sys message appears when removing driver

When removing the driver with Device Manager or using the Brocade Software Uninstaller, program a message appears asking for Files needed for bfad.sys to complete the installation. This occurs if you perform the following sequence of steps: 1. Install the driver using the driver installer program (bfa_installer.exe) or Brocade Software Installer (GUI or command-based application). 2. Uninstall the Brocade HBA using Device Manager. 3. Re-install the driver using the driver installer program (bfa_installer.exe) or Brocade Software Installer (GUI or command-based application). 4. Uninstall the driver using the driver installer program (bfa_installer.exe) program. To avoid this problem, do not uninstall the driver using the Device Manager if you have used the Brocade installer programs to install driver instances. Always use the Brocade installer programs. If only one driver is present in the system, then the Brocade programs also remove the Fibre Channel devices from the Device Manager.
13
Cannot roll back driver on all HBA instances using Device Manager
When using the Windows Device Manager, you can only roll back the driver for the first HBA instance. This occurs if you perform the following sequence of steps: 1. Install the driver using the Brocade driver installer program (bfa_installer.exe) or Software Installer (GUI or command-based application). 2. Roll back driver HBA instances using Device Manager. To avoid this problem, use one of the following methods:
Install the driver for each HBA instances using the Device Manager, then roll back the driver
using Device Manager.
Use the driver installer program (bfa_installer.exe) or Brocade Software Installer (GUI or
command-based application) to install or upgrade the driver, then use the Brocade Software Uninstaller to roll back drivers on all HBA instances in one-step.
Host not booting from remote LUN

If booting the host from a remote boot device, verify whether boot over SAN configuration is complete and correct. For example, verify the following:
A zone is created on the attached switch that contains only the PWWN of the storage system
port where the boot LUN is located and the PWWN of the HBA port.
BIOS or EFI is enabled to support boot over SAN from a specific HBA port. BIOS or EFI is configured to boot from a specific LUN. The hosts operating system, HBA driver, and other necessary files are installed on the boot
LUN.
Confirming driver package installation

HBA driver packages from Brocade contain the current driver, firmware, and HCM agent for specific operating systems. Make sure that the correct package is installed for your operating system. Refer to the installation chapter in the Brocade HBA Installation and Reference Manual. An out-of-date driver may cause the following problems:
Storage devices and targets not being discovered by the device manager or appearing
incorrectly in the hosts device manager.
Improper or erratic behavior of HCM (installed driver package may not support HCM version). Host operating system not recognizing HBA installation. Operating system errors (blue screen).
If driver is not installed, try re-installing the driver or re-installing the HBA hardware and then the driver. You can use HCM and tools available through your hosts operating system to obtain information such as driver name, driver version, and HBA port WWNs.
NOTE
14
Host Connectivity Manager (HCM)

Use the following steps to display the HBA PWWN, driver name and version, firmware name and version, and the BIOS version currently in operation 1. Launch HCM. 2. Select the HBA in the device tree. 3. Select the Properties tab in the right pane to display the Properties dialog box. The dialog box displays HBA properties.
Windows
Use the Device Manager to determine driver installation. Verify if the driver is installed and Windows is recognizing the HBA using the following steps. 1. Open the Device Manager. 2. Expand the list of SCSI and RAID controllers. 3. Right-click the Brocade FC HBA model where you are installing the driver. If you do not see this entry or Fibre Channel Controller displays with a yellow question mark under Other Devices, the driver is not installed. 4. Select Properties to display the Properties dialog box. 5. Click the Driver tab to display the driver date and version. Click Driver Details for more information.
NOTE
If driver is not installed, try re-installing the driver or re-installing the HBA hardware and then the driver.
Linux
Verify if the HBA driver installed successfully using the following commands:
# rpm -qa|grep -i bfa

This command prints the names of the Brocade HBA driver package (bfa) if installed.
# lspci
This is a utility that displays information about all PCI buses in the system and all devices connected to them.
# lsmod
This command displays information about all loaded modules. If bfa appears in the list, the HBA driver is loaded to the system.
# dmesg
This command prints kernal boot messages. For the bfa entry, HBA model and driver version should display if the hardware and driver are installed successfully.
15
Host system freezes or crashes
# modprobe -l bfa
This verifies that the module has loaded. If bfa displays, the module has been loaded to the system.
Solaris
Verify if the HBA driver installed successfully using the following commands.
pkgchk -nv bfa

This checks for and lists the installed HBA driver package files.
pkginfo -l bfa
This displays details about installed Brocade HBA (bfa) drivers. Look for information as in the following example. Note that the VERSION may be different, depending on the driver version you installed. The ARCH and DESC information may also be different, depending on your host system platform. If the HBA driver package is installed, bfa_pkg should display with a completely installed. status.
PKGINST: NAME: CATEGORY: ARCH: VERSION: BASEDIR: VENDOR: DESC: PSTAMP: INSTDATE: HOTLINE: STATUS: bfa Brocade Fibre Channel Adapter Driver system sparc&i386 alpha_bld31_20080502_1205 / Brocade 32 bit & 64 bit Device driver for Brocade Fibre Channel adapters 20080115150824 May 02 2008 18:22 Please contact your local service provider completely installed
VMware
Verify if the HBA driver installed successfully using the following commands:
vmkload_mod -l
This lists installed driver names, R/O and R/W addresses, and whether the ID is loaded. Verify that an entry for bfa exists and that the ID loaded.
cat /proc/vmware/version
This displays the latest versions of installed drivers. Look for a bfa entry and related build number.
Host system freezes or crashes

If the BIOS and the operating system recognize the HBA, but the host system freezes or crashes during startup and does not boot, disconnect all devices from the HBA, then reboot the host system.
If the system does not freeze when rebooted and operates correctly, use the following
information to resolve the problem:
Check for faulty fiber optic cable and cable connections.
16
HCM GUI fails to connect with HCM agent
Try rebooting the system without any connectivity to the switch. This will help isolate any hang caused by switch and device interactions. Reseat SFPs in the HBA. Determine whether the installed SFPs are faulty by observing LED operation by HBA ports. If all LEDs are flashing amber, the SFP is invalid and may not be a required Brocade model. You can also verify SFP operation by replacing them with SFPs in known operating condition. If the problem is resolved after replacement, original SFP is faulty. Check for conflicts with attached devices. Verify that data speed (1-8 Gbps) and connection topology (for example, point-to-point) for devices attached to the HBA are compatible with settings on the HBA port. Although auto may be set, configuring settings manually on the HBA port and devices may allow connection. Also, note that the HBA only supports point-to-point connection topology. Refer to the Brocade Fibre Channel HBA Administrators Guide for procedures to configure HBA ports.
NOTE
Observe the LEDs by HBA ports. Illuminated LEDs indicate connection, link activity, and connection speed negotiated with the attached device. Refer to LED Operation in the Specifications chapter of the Brocade Fibre Channel HBA Installation and Reference Manual.
If the system freezes perform the following tasks: - Verify if the host system firmware supports PCIe specifications listed in the Hardware and
Software Compatibility section, Introduction chapter, of the Brocade Fibre Channel HBA Installation and Reference Manual. If not, download a firmware update to support the HBA.
On Windows systems, determine when the system freezes during the boot process. If it freezes as the driver loads, uninstall and reinstall the driver. If it freezes during hardware recognition, uninstall both the driver and HBA, then reinstall both. Remove the HBA and reboot the system. If the system boots, reinstall the HBA. Reseat the HBA. Uninstall and reinstall the driver. Try installing the HBA into another host system. If the problem does not occur, the HBA may not be compatible with the original host system. If the problem occurs in the new system, replace the HBA.

The HCM Agent installs with the HBA driver package on the host system. It is independent of the HCM user interface (client application), which is installed on the local or remote host. If you receive an error such as Failed to connect to agent on host... when using HCM, this indicates that the client application cannot connect to the HCM Agent listening on the configured port - normally TCP port 34568. This occurs for one of the following reasons:
The agent is not running. The agent not accepting connections on the expected port. The agent is not listening on the expected port.
17
Communications between the client and agent is blocked by a firewall preventing access to the
port (usually only a consideration for remote HCM management).
Linux, Solaris, and VMware systems

Perform the following tasks to isolate and resolve the problem. 1. Verify that the agent is running by executing the appropriate status command for your operating system as described in the Installation chapter of the Brocade Fibre Channel HBA Installation and Reference Manual. Refer to the section on modifying HCM agent operation. 2. If you receive a message that the hcmagent is stopped, restarting the agent should resolve the problem. To restart, use the appropriate start command for your operating system which is also described in the Brocade Fibre Channel HBA Installation and Reference Manual. Note that one command described in the manual restarts the agent, but the agent will not restart if the system reboots or the agent stops unexpectedly. Another command restarts the agent, but the agent will restart if the system reboots. 3. Confirm the HCM agent is responding to requests using the expected user password. Execute the following command to connect to the HCM agent and force it to collect the HBA driver supportsave data.
NOTE
This command is a single line. The localhost can be replaced with a different IP address.
wget --no-check-certificate https://admin:password@localhost:34568/JSONRPCServiceApp/ SupportSaveController.do
If successful, the file SupportSaveController.do (actually a zip format file) will contain the data from the HCM agent. 4. If you are managing a VMware host system through HCM from a remote system, the hosts firewall may be blocking TCP/IP port 34568, which allows agent communication with HCM. Use the following command to open port 34568:
/usr/sbin/esxcfg-firewall-o 34568,tcp,out,https
Use Windows Firewall and Advanced Service (WFAS) to open port 34568.
NOTE
You can change the default communication port (34568) for the agent using procedures in the Installation chapter of the Brocade Fibre Channel HBA Installation and Reference Manual. Refer to the section on modifying HCM agent operation. 5. If HCM is still unable to connect to the HCM agent after using the preceding steps, collect the following data and send to your Support representative for analysis:
Data collected from the previous step in SupportSaveController.do. Data from the HCM application SupportSave feature. Select Tools > SupportSave to
generate a supportsave file. The data file name and location displays when the SupportSave feature runs.
18
HBA agent files on the HBA host (where the HCM agent is installed). Collect these files
using the following command:
tar cvfz hbaagentfiles.tgz /opt/hbaagent
Output collects to hbaagentfiles.tgz.
Data collected on the HBA host from the bfa_supportsave feature using the following
command:
bfa_supportsave
Output collects to a file and location specified when the SupportSave feature runs.
Windows systems
Perform the following tasks to isolate and resolve the problem. 1. Verify that the agent is running by executing the appropriate status command for your operating system described in the Installation chapter of the Brocade Fibre Channel HBA Installation and Reference Manual. Refer to the section on modifying HCM agent operation. 2. If you receive a message that the hcmagent is stopped, restarting the agent should resolve the problem. To restart, use the appropriate start command for your operating system which is also described in the Brocade Fibre Channel HBA Installation and Reference Manual. Note that one command described in the manual restarts the agent, but the agent will not restart if the system reboots or the agent stops unexpectedly. Another command restarts the agent, but the agent will restart if the system reboots. 3. If the HCM agent starts, verify which TCP port the agent is listening on by executing the following command at the Windows command prompt:
netstat -nao | findstr 34568
Output similar to the following should display.

TCP 0.0.0.0:34568 0.0.0.0:0 LISTENING 1960
1960 in the last column is the process identifier for the Windows process listening on the TCP port. Note that this identifier may be different on your system. 4. Enter the following command to confirm that the process identifier bound to TCP port 34568 is for the hcmagent.exe process:
tasklist /svc | findstr 1960
The following should display if the identifier from step 3 is bound to TCP port 34568:
hcmagent.exe 1960 hcmagent
5. If you are managing a Windows 2008 host system through HCM from a remote system, the hosts firewall may be blocking TCP/IP port 34568. Use Windows Firewall and Advanced Service (WFAS) to open port 34568.
NOTE
You can change the default communication port (34568) for the agent using procedures in the Installation chapter of the Brocade Fibre Channel HBA Installation and Reference Manual. Refer to the section on modifying HCM agent operation.
19
Verifying Fibre Channel links
6. If the hcmagent is running and listening on port 34568 and there are no firewall issues (as explained in step 5), but you get the same Failed to connect to agent on host..." error when using HCM, collect the following data. Send this data to your Support representative for analysis:
Copies of output from the commands in step 3 and step 4. Files from the output directory created after you execute the bfa_supportsave feature.
To collect these files, execute the following command:
bfa_supportsave
Data collected by this command saves to subdirectory named bfa_ss_out. In Windows explorer, right-click this directory and select Send To > Compressed (zipped). This creates a zip file that you can send to your Support representative.
Build information for the HCM application. Select Help > About in HCM to display the
version, build identification, and build date.
Support data from the HCM application SupportSave feature.

Select Tools > SupportSave to generate a supportsave file. If HCM cannot connect to the agent, a message displays an error (Agent Support Save could not be collected) and explains that only a basic collection is possible. Messages also display that provide the location of the zip file created. By default, a zip file is created in the following location if installed the application in C:\Program Files\BROCADE\FCHBA: C:\Program Files\BROCADE\FCHBA\client\data\localhost\supportsave The zip file will have a name similar to the following: SupportSave_Basic_2008723_0_50_57.zip

Check for link problems by observing LED operation for HBA ports. LED operation other than expected or LEDs may indicate link problems. For example, all LEDS flashing amber for a port indicates that an invalid non-Brocade SFP may be installed. For details on HBA LED operation, refer to the specifications chapter in the Brocade HBA Installation and Reference Manual. If LEDs do not illuminate to indicate an active link, use appropriate Fabric OS and HBA diagnostic commands and HCM options in Table 4 on page 21. For additional diagnostics commands, refer to Chapter 3, Tools for Collecting Data for HCM and BCU commands and the Fabric OS Administrators Guide for Fabric OS commands. Also verify LED operation on switch port that is connected to an HBA port. Refer to the switch Hardware Reference Manual to analyze LED meaning. Common link problems can be caused by the following:
NOTE
Damaged fiber optic cables. (Note that damaged cables can also cause errors and invalid data
on links.)
20
Fiber optic cables may not be rated or compatible with HBA port speeds. Refer to Fibre Optic
Cable specifications in the Brocade Fibre Channel HBA Installation and Reference Manual.
Faulty switch or HBA SFPs. Verify if an SFP is the problem by connecting a different link to the
HBA port or, if convenient, replace the cable with a cable of known quality. If the errors or invalid data on the link still indicate a cable problem, the SFP may be faulty. Try replacing the SFP.
SCSI retries and timeouts determine communication between HBA and storage. Dropped
packets cause timeouts. Packets can drop because of SFP issues on HBA or switch - possibly the SFP is not compatible with HBA, but is compatible with switch or vice versa. You can run the BCU port --stats command to display port statistics, such as error and dropped frames. Table 4 lists HCM options and BCU commands, as well as Fabric OS commands that you can use to determine link status.
TABLE 4
Application
HCM
Tools to determine link status

Tool References
Port Statistics Loopback and PCI loopback test Fibre Channel ping, echo, and trace route tests Link Beaconing Port Properties SFP information BCU fcdiag and diag commands. Port commands, such as port --stats. switchShow portShow portStatsShow portErrShow fcpProbeShow fPortTest Chapter 3, Tools for Collecting Data
BCU Switch Fabric OS
Chapter 3, Tools for Collecting Data
Chapter 3, Tools for Collecting Data Fabric OS Administrators Guide Fabric OS Troubleshooting and Diagnostics Guide
21
22
Chapter
Tools for Collecting Data
In this chapter
For detailed information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data to provide support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Collecting data using host system commands . . . . . . . . . . . . . . . . . . . . . . . Collecting data using BCU commands and HCM . . . . . . . . . . . . . . . . . . . . . Collecting data using Fabric OS commands . . . . . . . . . . . . . . . . . . . . . . . . . Event logs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diagnostics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Collecting SFP data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Collecting port data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Authentication settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . QoS and target rate limiting settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Persistent binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23 24 24 25 28 30 34 40 45 46 48 48 50
For detailed information

This chapter provides basic instruction on tools useful for gathering information to isolate HBA problems. For more detailed information on using these tools, refer to the following publications:
Brocade Fibre Channel HBA Administrators Guide

The following chapters in this guide cover HBA HCM and BCU monitoring and diagnostics tools:
Monitoring Diagnostics and Troubleshooting Command Line Utility (BCU)
Fabric OS Troubleshooting and Diagnostics Guide

This guide provides detailed information on collecting troubleshooting information and isolating general SAN problems between the switch, host systems, and storage systems.
Fabric OS Command Reference Manual

Fabric OS diagnostic and monitoring commands.
Your host systems operating system user and administrators guides.

Host system diagnostics, logs, and system monitoring tools.
23
Data to provide support
Data to provide support

When problems occur requiring support assistance, provide a detailed description of the problem, as well as output collected from the following HCM and BCU tools:
Support Save Diagnostics Port logs Port statistics and properties HBA properties Host operating system commands
Collecting data using host system commands

The following table describes commands common to each supported operating system that you can use to gather information for troubleshooting problems. For details on these commands, refer to your systems online help and documentation.
NOTE
Output from all of these commands is captured using the Support Save feature.
TABLE 5
Task
Host system data collection commands

Linux
lspci -vv
Windows
In Windows registry location HKEY_LOCAL_MACHINE \SYSTEM\CurrentContro lSet\Enum\PCI devcon find pci\* msinfo32.exe Click the plus sign(+) next to Components to view hardware details. Windows Task Manager, tasklist.exe Windows Task Manager, tasklist.exe Windows Task Manager, perfmon.exe
VMware
lspci -vv, esxcfg-info -w
Solaris
prtdiag -v, prtconf -pv
List PCI devices
List installed HW details
lsdev
esxcfg-info -a
prtdiag -v, prtconf -pv
Process information Memory usage Performance monitoring
ps -efl, top top, vmstat -m iostat, vmstat, sar
ps -efl, top top, vmstat -m vmstat, VM Performance: esxtop [first type 'v', 'e' then enter vm# in the list down], Disk Performance: esxtop [type 'v' then 'd']. vmkload_mod -l vmkload_mod -l
ps -efl, top vmstat -s iostat -nx 1 5, vmstat, mpstat, sar
List for driver modules To check for Brocade Fibre Channel adapter (BFA) driver module
lsmod lsmod | grep bfa
driverquery driverquery /v | findstr bfad
modinfo modinfo | grep bfa
24
Collecting data using BCU commands and HCM
TABLE 5
Task
Host system data collection commands

Linux
dmesg, /var/log/message*
Windows
System Category in Windows Event Viewer (eventvwr.exe) systeminfo.exe
VMware
/var/log/message* , /var/log/vmkernel*, /var/log/vmkwarning*,/p roc/vmware/log cat /etc/vmware-release
Solaris
dmesg, /var/adm/message*
System log message location NOTE: For more information, refer to Host system logs on page 30. Show OS distribution info
(SuSE) cat /etc/SuSE-release,( RedHat) cat /etc/redhat-release /etc/bfa.conf
uname -a, cat /etc/release
BFA configuration file location
/etc/bfa.conf Windows Registry (HKEY_LOCAL_MACHINE \SYSTEM\CurrentContro lSet\Services\bfad\Para meters\Device), HBA Flash Windows Registry (HKEY_LOCAL_MACHINE \HARDWARE\DEVICEMA P\Scsi\Scsi Port x) /dev/bfa*
/kernel/drv/bfa.conf
BFA device file location
/dev/bfa*
(Release 1.0) /devices/pci*/pci*/ fibre-channel@0:dev ctl, (Release 1.1 and later) /devices/pci*/pci*/ bfa@0:devctl

You can collect a variety of information on installed Brocade HBAs, such as firmware version installed, operational status, port speed, WWN, PCI data, configuration data, flash status, and other details for troubleshooting using BCU commands, HCM menu options, and host operating system commands.
Using Support Save

The Support Save feature is an important tool for collecting debug information from the driver, internal libraries, and firmware. You can save this information to the local file system and send to support personnel for further investigation. Use one of the following options to launch this feature:
In HCM, launch Support Save through the Tools menu. Through BCU, enter the bfa_supportsave command. Through your internet browser (Internet Explorer 6 or later or Firefox 2.0 or later), you can
collect bfa_supportsave output if you do not have root access, do not have access to file transfer methods such as FTP and SCP, or do not have access to the Host Configuration Manager (HCM). A bfa_supportsave collection can also occur automatically for a port crash event. The Support Save feature saves the following information:
NOTE
25
HBA model and serial number. HBA firmware version. Host model and hardware revision. All support information. HBA configuration data. All operating system and HBA information needed to diagnose field issues Information about all HBAs in the system. Firmware and driver traces. Syslog message logs. Windows System Event log .evt file. HCM GUI-related engineering logs Events HBA configuration data Environment information Data.xml file Vital CPU, memory, network resources HCM Agent (logs, configuration) Driver logs (bfa_supportsave output) Core files
Master and Application logs are saved when Support Save is initiated through HCM, but not through BCU.
NOTE
Initiating Support Save through HCM

Initiate Support Save by selecting Tools > Support Save. Messages display during the Support Save operation that provide the location of the directory where data is saved. If you are initiating Support Save from a remote management station and receive a warning message that support files and Agent logs could not be collected, the HCM Agent is unavailable on the remote host. Select Tools > Backup to backup data and configuration files manually. For more information and additional options for using this feature, refer to the Brocade Fibre Channel HBA Administrators Guide.
Initiating Support Save through BCU commands

Use the bfa_supportsave command to Initiate Support Save through the BCU:
bfa_supportsave {output_dir]
where: [output_dir] An optional parameter that specifies the directory where you want output saved. If not specified, output is saved as a directory in the current working directory as bfa_ss_out.
26
Messages display as the system gathers information. When complete, an output file and directory display. The directory name specifies the date when the file was saved. For more information on the bfa_supportsave feature, refer to the Host Connectivity Manager (HCM) Administrators Guide.
Initiating Support Save through the internet browser

Initiate bfa_supportsave through an internet browser. 1. Open an Internet browser and type the following URL:
https://localhost:34568/JSONRPCServiceApp/SupportSaveController.do
where localhost is the IP address of the server from which you want to collect the bfa_supportsave information. 2. Log in using the factory default user name (admin) and password (password). Use the current user name and password if they have changed from the default, The File Download dialog box displays, prompting you to save the supportSaveController.do file. 3. Click Save and navigate to the location where you want to save the bfa_supportsave file.
Initiating Support Save through a port crash event

If the port crashes and triggers a port crash event, support save data is collected at a system-wide level. An Application Log message is generated with the following message:
Port Crash Support Save Completed
Port crash events have a CRITICAL severity and you can view the details in the Master Log and Application Log tables in HCM. For more information on these logs, refer to HCM logs on page 32.
Support Save differences

Following are differences in data collection for the HCM, BCU, and browser applications of bfa_supportsave:
BCU
Collects only driver-related logs and configuration files.
Browser
Collects driver-related and HCM Agent logs and configuration files.
HCM
Collects, HCM, driver-related, and HCM Agent logs and configuration files.
bfa_supportsave collection on a port crash event

If the port crashes and triggers a port crash event, support save data is collected at a system-wide level. An Application Log message is generated with the following message:
Port Crash Support Save Completed
27
Collecting data using Fabric OS commands
Port crash events have a CRITICAL severity and you can view the details in the Master Log and Application Log tables in HCM.
HBA data collection using HCM

Use the HBA Properties panel in HCM to display information about the HBA, such as its WWN, disabled or enabled status, temperature, installed driver name, driver version, firmware version, and BIOS version. This panel also displays PCI information for the HBA, such as vendor ID, device ID, subsystem ID, current number of lanes, and PCI generation. 1. Launch HCM. 2. Select an HBA in the device tree 3. Click the Properties tab in the right pane.
Collecting data using BCU commands

Use the BCU adapter command to list and query available HBAs seen by the driver. The bcu adapter --list command lists all HBAs on the system with a brief summary of information such as model number, serial number, and HBA number. Enter the following command:
adapter --list
where: list Lists all adapters in the system. For each adapter in the system, a brief summary line is displays.
The adapter --query command displays adapter information, such as the current version of the HBA (chip revision) and driver (fw version), maximum port speed, model information, serial number, number of ports, PCI information, pwwn, nwwn, hardware path, and flash information (such as firmware version).
adapter -query <ad_id>
where: ad_id ID of the adapter (HBA) for which you want to query.

Use the following Fabric OS commands on attached Brocade switches to gather information and help isolate connectivity and other problems between the HBA, switch, and storage ports. For details on using these commands, refer to the Fabric OS Command Reference Manual.
authUtil
Use this command to display and set local switch authentication parameters.
cfgShow
Displays zone configuration information for the switch. You can use command output to verify target ports (by port WWN) and LUNs that are intended to be accessible from the HBA.
28
fcpProbeShow
Use this command to display the Fibre Channel Protocol daemon (FCPd) device probing information for the devices attached to a specified F_Port or FL_Port. This information includes the number of successful logins and SCSI INQUIRY commands sent over this port and a list of the attached devices.
nsShow
Use this command to display local NS information about all devices connected to a specific switch. This includes information such as the device PID, device type, and the port and node WWN.
zoneshow
Use this command without parameters to display all zone configuration information (both defined and enabled).
portErrShow
Use this command to display an error summary for all switch ports.
portLogShow
Use this command to display the port log for ports on a switch.
portLogShowPort
Use this command to display the port log for a specified switch port.
portPerfShow
Use this command to display throughput information for all ports on the switch.
portStatsShow
Use this command to display hardware statistics counters for a specific switch port.
portShow
Use this command to display information and status of a specified switch port, including the speed, ID, operating state, type, and WWN.
SecAuthSecret
Use this command to manage the DH-CHAP shared secret key database used for authentication.This command displays, sets, and removes shared secret key information from the databases
sfpShow
Use this command to display detailed information about specific SFPs installed in a switch.
switchShow
Use this command to display switch and port information. Output may vary depending on the switch model. Use this information to determine the fabric port WWN and PID connected to an HBA port. Also display topology, speed, and state of each port on the switch.
29
Event logs
Event logs
Event messages that occur during HBA and driver operation are important tools for isolating and resolving problems. Messages provide descriptions of the event, severity, time and date of the event, and in some cases, cause and recommended actions. These messages are captured in logs. Monitoring events in these logs allows early fault detection and isolation on a specific HBA. The following types of logs are available:
Host operating system logs HCM logs
Host system logs

Brocade HBA event messages are captured in host system log files. All messages related to the Brocade HBA are identified in these logs by BFAD or bfa (Brocade Fabric Adapter). These messages contain the following information:
Message ID Message text Message arguments Severity level Cause Recommended action
Brocade HBA event message files are installed in the HBA driver installation directory for each supported operating system. Table 8 provides the location of the message files for each system.
TABLE 6
Linux VMware Solaris Windows
Message catalog location

Catalog Location
/opt/bfa /opt/bfa /opt/bfa aen.zip loaded to your driver installation directory. Unzip this file to obtain all message catalog files.
Operating System
Table 7 describes the logs for each supported operating system, where the logs are stored, and how to view them.
TABLE 7
Solaris Windows
System Event Logs

Log Name
Syslog Event Log
Operating System
Location
/var/adm/messages Not applicable
Viewing Message Log

dmesg command System category in Event Viewer (eventvwr.exe)
30
Event logs
TABLE 7
Linux
1VmWare
System Event Logs

Log Name
Messages Log Messages Log
Operating System
Location
/var/log/message /var/log/message* , /var/log/vmkernel*, /var/log/vmkwarning*,/proc/v mware/log
Viewing Message Log

dmesg command dmesg command
1.
For ESX Server Console operating system. For Guest system, refer to information in Windows or Linux.
You can view all event messages that can display for a Brocade HBA by viewing HTML files that are loaded to your system as the driver package installs. These files contain all message information that can display on system logs for the Brocade HBA. View these files through your internet browser. Table 8 provides the location of the Brocade HBA message files for each supported system.
TABLE 8
Linux VMware Solaris Windows
Message file location

File Location
/opt/bfa /opt/bfa /opt/bfa aen.zip loaded to your driver installation directory. Unzip this file to obtain all message catalog files.
Operating System
Table 9 lists the message file names and content.
TABLE 9
Event message files

Content
Adapter events, such as adapter added or removed. Audit events, such as authentication enabled or disabled for base port. IO controller (IOC) events, such as IOC heartbeat status and IOC enabled or disabled. Initiator-target nexus events, such as target is online for initiator, has lost connection with initiator, or is disconnected by initiator. Logical port events, such as a logical port is created, deleted, online, offline, has lost fabric connectivity, or the ports login has failed. Physical port events, such as port online or offline, SFP removed from port, or authentication successful or unsuccessful for base port.
Event Catalog File

bfal_aen_adapter.html bfal_aen_audit.html bfal_aen_ioc.html
bfal_aen_itnim.html
bfal_aen_lport.html
bfal_aen_port.html
31
Event logs
TABLE 9
Event message files

Content
Remote port (RPORT) events, such as a remote port is online, offline, or disconnected, or the QoS priority or flow ID changed. List of error codes and meanings for adapter, port, L_Port, R_Port, itnim, audit, and IOC events.
Event Catalog File

bfal_aen_rport.html
hba_error_codes.doc
Adjust the logging level, or the types of messages logged to your system log that relate to HBA driver operation, using the following HCM options and BCU commands.
Adjust logging level using this BCU command.

log -level <port_id> [<Trace | Debug | Info | Warn| Error | Fatal>]
Adjust logging level using the HCM.

Select Configure > HCM Logging Levels. Select the a level on the Agent Communication Log and HCM Debug Log lists. Values are Trace, Debug, Info, Warning, Error, and Fatal. For details on adjusting logging levels, refer to the Brocade Fibre Channel HBA Administrators Guide.
NOTE
HCM logs
You can view data about HBA operation through the following HCM logs. Both of these logs display on the bottom of the HCM main window. Click the Master Log or Application Log to toggle between logs.
The Master Log displays informational and error messages during HBA operation. This log
contains the severity level, event description, date and time of event, the function that reported the event (such as a specific HBA port or remote target port), WWN of device where event occurs, and other information.
The Application Log displays informational and error messages related to HBA discovery or
HCM application issues.
Master Log
The Master Log displays event information in seven fields:
Sr No.
Sequence number that event occurred in ascending order.
32
Event logs
Severity
Event severity level (informational, minor, major, or critical).
Critical-level messages indicate that the software has detected serious problems that will eventually cause a partial or complete failure of a subsystem if not corrected immediately. Examples of these could be a power supply failure or rise in temperature Major messages represent conditions that do not impact overall system functionality significantly. Examples of these could be timeouts on certain operations, failures of certain operations after retries, invalid parameters, or failure to perform a requested operation. Minor messages highlight a current operating condition that should be checked or it might lead to a failure. Information-level messages report the current non-error status of the system components; for example, the online and offline status of a fabric port.
WWN
World Wide Name of HBA where event occurred.
Category
The category or type of event. Categories define the component where events occur:
Adapter - Events relating to the HBA (Adapter). Port - Events relating to a specific port on the HBA. LPORT - Events relating to a specific logical port. RPORT - Events relating to a specific remote initiator or target port. ItNIM - Events relating to an initiator-target nexus. Examples of these include end to end target discovery, initiator target connectivity, and loss of connectivity. Audit - Audit events. IOC - Driver and firmware events involving the I/O controller on the HBA.
Subcategory
Subcategory of main category.
Description, Date, and Time

Brief description of event and date and time when event occurred. You can block events from display in the Master Log by severity, category, and WWN of HBA. To display the Master Log Filter dialog box, click the Filter button in the Master Log section of the main HCM screen. Select areas that you want to filter and click OK.
Application Log
The Application Log displays all application-related informational and error messages, as well as the following attributes.
Date and time the message occurred. Severity of the message. Description of the message. The agent IP address.
33
Statistics
Syslog support
You can configure the HCM agent to forward events to a maximum of three syslog destinations using the Syslog option on the HCM Configure menu. These events will display in the operating system logs for systems such as Solaris and Linux. For procedures to configure syslog destinations, refer to the Brocade Fibre Channel HBA Administrators Guide.
Windows Event Log support

HBA events are logged in the Windows Event Log by the HCM agent. Event messages appear with the date and time the event occurred and the source as hcmAgent. HBA events are logged under the HCM_Event category. Event types display with each event to note severity level. From most to least severity, NT events are error, warning, and information. Driver errors are critical, error, warning, and info.
Statistics
You can access a variety of statistics using BCU commands and HCM. Use these statistics to monitor HBA performance and traffic between the HBA and LUNs and isolate areas that impact performance and device login. You can display statistics for the following:
HBA ports IO controller Virtual ports (vport) Logical ports (lport) Remote ports (rport) FCP initiator mode Fabric (BCU only) Targets Security authentication
This section provides an overview of these statistics and how to access them. For more detail, refer to the Brocade Fibre Channel HBA Administrators Guide.
Port statistics
Use BCU and HCM to display a variety of port statistics, such as transmitted and received frames and words, received loop initialization primitive (LIP) event counts, error frames received, loss of synchronization, link failure and invalid CRS counts, and end of frame (EOF) errors. Use these statistics to isolate link and frame errors. For example, loss of synch and loss of signal errors indicate a physical link problem. To resolve these problems, check cables, SFPs on the HBA or switch, and patch panel connections.
34
Statistics
Displaying statistics through BCU

Use the port --stats BCU command to display statistics for a specified HBA port.
port --stats <port_id>
where: port_id ID of the port for which you want to display statistics.
Displaying statistics through HCM

Use the Port Statistics dialog box to monitor a variety of port data. 1. Launch the HCM. 2. Select the base HBA port from the device tree window. 3. Click Monitor > Statistics > Port Statistics.
IOC statistics
Use BCU and HCM to display port-level statistics for the I/O controller through the BCU and HCM. The I/O controller refers to the firmware entity controlling the port. The following types of IOC statistics are displayed:
IOC driver IOC firmware Firmware IO Firmware port FPG Firmware port PHYSM Firmware port LKSM Firmware port SNSM
Displaying IOC statistics through BCU

Use the ioc -stats command to display IOC statistics.
ioc -stats <ioc_id>
where: ioc_id ID of the IOC controller for which you want to display statistics.
Displaying IOC statistics through HCM

Use the IOC Statistics dialog box to monitor a variety of port data. 1. Launch the HCM. 2. Select the base HBA port from the device tree window. 3. Click Monitor > Statistics > IOC Statistics.
35
Statistics
Fabric statistics
Use BCU and HCM to display statistics for fabric login (FLOGI) activity and fabric offlines and onlines detected by the port. Use these statistics to help isolate fabric login problems. Following are two examples of how to use these statistics for troubleshooting:
If the HBA is not showing in the fabric, check the FLOGI sent and FLOLGI accept statistics. If the
counts do not match, the switch or fabric may not be ready to respond. This is normal as long as it does not persist. If the problem persists, this could indicate a problem in the fabric or a protocol issue between the HBA and fabric.
If fabric offline counts increase and fabric maintenance is not being done, this may indicate a
serious fabric problem. Slow fabric performance or hosts unable to address storage could also be seen.
Displaying fabric statistics through BCU

Use the fabric --stats command to display fabric statistics.
fabric -stats <port_id>
where: port_id ID of the HBA port for which you want to display statistics.
Displaying fabric statistics through HCM

Use the Fabric Statistics dialog box to monitor a variety of port data. 1. Launch the HCM. 2. Select the base HBA port from the device tree window. 3. Click Monitor > Statistics > Fabric Statistics.
Remote port statistics

Remote port statistics can help isolate end-to-end login problems. Use HCM and BCU to display statistics for the following:
Port login (PLOGI) activity Authentication and discovery (ADISC) activity Logout (LOGO) activity RCSNs received Process logins (PRLI) received Hardware abstraction layer (HAL) activity
As an example of using these statistics for troubleshooting, if the host cannot see the target, you can verify that the remote port (rport) is reporting itself online by comparing the rport offline and rport offline statistics. The rport online counter should be one greater than the rport offline counter. If not, clear the counters and retry connecting to the remote port. Verify the rport online and rport offline statistics again.
36
Statistics
Displaying target statistics (HCM)

Use the Target Statistics dialog box to display target statistics. 1. Launch the HCM. 2. Select the base HBA port from the device tree window. 3. Click Monitor > Statistics > Remote Port Statistics > Target Statistics.
Displaying remote port statistics (BCU)

Use the rport --stats command to display remote port statistics:
rport -stats <port_id> <rpwwn> [-l lpwwn]
where: port_id lpwwn rpwwn ID of the port for which you want to display rport statistics. Displays the logical port world wide name. This is an optional argument. If the -l lpwwn argument is not specified, the base port is used. Displays the remote ports port world wide name.
FCIP initiator mode statistics

Use HCM and BCU to display Fibre Channel Protocol Initiator Mode (FCP IM) statistics such as the number of online and offline remote ports, process login (PRLI) activity, and HAL statistics.
Displaying FCIP initiator mode statistics through BCU

Use the fcpim --stats command to display FCIP initiator mode statistics.
fcpim --stats <port_id> <rpwwn [-l <lpwwn>]
where: port_id lpwwn rpwwn ID of the port for which you want to display statistics. Logical port world wide name. This is an optional argument. If the -l lpwwn argument is not specified, the base port is used. Remote port world wide name.
Displaying FCIP initiator mode statistics through HCM

Display statistics by selecting Monitor > Statistics > Remote port statistics > FCP IM Statistics.
Logical port statistics

Use HCM and BCU to display logical port statistics for the following:
Name server (NS) port logins (plogin) activity Register symbolic port name (RSPN_ID) identifier activity Register FC4 type identifier (RFT_ID) activity
37
Statistics
Register FC4 type identifier (RFT_ID) activity Get all port ID requests for a given FC4 type (NS_GID_FT) activity Retries Timeouts
Use these statistics to help determine if the HBA is not registering with the name server or cannot access storage. Following are examples of how these statistics indicate these problems:
If name server port login (NS PLOGI) error rejects and unknown name server port login
response (NS login unknown rsp) errors increase, then the HBA most likely cannot log in to the name server.
If name server register symbolic port name identifier (NS RSPN_ID) or name server register
symbolic port name identifier response (NS RFT_ID rsp) errors or rejects (NS RFT_ID rejects) are increasing, the HBA has a problem registering with the name server.
If name server get all port ID response NS GID_FT rsp), rejects (NS_GID FT rejects), or unknown
responses (NS_GID FT unknown rsp) are increasing, the HBA has a problem querying the name server for available storage.
Displaying logical port statistics through HCM

Display logical port statistics by selecting Monitor > Statistics > Logical Port Statistics OR Right-click a logical port from the device tree and select Logical Port Statistics.
Displaying logical port statistics through BCU

Use the lport --stats command to display logical port statistics.
lport -stats <port_id> [-l lpwwn]
where: port_id lpwwn ID of the port for which you want to display statistics. Logical port world wide name for which you want to display statistics. This is an optional argument. If the -l lpwwn argument is not specified, the base port is used. Remote port world wide name for which you want to display statistics.
rpwwn
Virtual port statistics

Use HCM and BCU to display logical port statistics for fabric discovery (FDISK) activity, logouts (LOGO) activity, NPIV support, number of fabrics online and offline, and fabric cleanups. Use these statistics to isolate NPIV login problems. Following are examples of what to check if virtual devices are not listed in the name server:
If FDISK sent and FDISK accept statistics do not match, the fabric or switch may not be ready.
This is normal as long as it does not persist. If it does persist, there may be a problem in the fabric or a protocol issue between the HBA and fabric. Note that in this case FDISK retries also increase.
38
Statistics
Check the No NPIV support statistics to verify that NPIV is supported and enabled on the
switch.
Displaying virtual port statistics through HCM

Display statistics by selecting Monitor > Statistics > Virtual Port Statistics OR Right-click a virtual port on the device tree and select Virtual Port Statistics.
Displaying virtual port statistics through BCU

Use the vport --stats command to display statistics.
vport --stats <port_id> vpwwn
where: port_id vpwwn ID of the port for which you want to display rport statistics. Displays the statistics for the virtual port by its WWN. If no part WWN is specified, the information provided is for the base vport.
Quality of service (QoS) statistics

Use HCM and BCU to display QoS statistics for individual ports. You can display statistics for fabric login (FLOGI) activity, exchange link parameter (ELP) activity, and received QOS registered state change notifications (RSCNs).
Displaying QoS statistics through HCM

Use the QOS Statistics dialog box to display QoS statistics. 1. Launch the HCM. 2. Select the base HBA port from the device tree window. 3. Click Monitor > Remote Port Statistics > QOS Statistics.
Displaying QoS statistics through BCU

Use the qos --stats command to display remote port statistics.
vport --stats <port_id> vpwwn
where: port_id ID of the port for which you want to display rport statistics.
39
Diagnostics
Diagnostics
Diagnostics, available through BCU commands and HCM, evaluate the integrity of HBA hardware and end-to-end connectivity in the fabric. All of these diagnostics can be used while the system is running.
Beaconing
Initiate beaconing on a specific HBA port to flash the port LEDs and make it easier to locate the HBA in an equipment room. Initiate link beaconing to flash the LEDs on a specific HBA port and the LEDs on a connected switch port to verify the connection between HBA and switch. When you initiate link beaconing, commands are sent to the other side of the link. When the remote port receives these commands, that ports LEDs flash. The remote port sends a command back to the originating port. When that port receives this command, the ports LEDs flash.
NOTE
To initiate link beaconing, this feature must be available on the connected switch. Toggle beaconing on and off and set beaconing duration using the BCU or HCM.
Enabling beaconing through BCU

Use the diag --beacon command to enable beaconing for a specific HBA port.
diag --beacon <port_id> <on | off> [<duration>]
where: port_id duration ID of the port for which you want to enable beaconing. Length of time between blinks.
Use the fcdiag --linkbeacon command to enable link beaconing.

fcdiag --linkbeacon <portid> {on | off} [<duration>]
where: port_id on | off duration ID of the port for which you want to run a link beacon test. Toggle on or off. If turned on, you can specify duration. Length of time between blinks.
Enabling beaconing through HCM

Enable link beaconing by selecting Configure > Beacon, then click the Beacon Link check box, the Beacon Port check box, or both.
40
Diagnostics
Loopback tests
Use the BCU or the HCM to perform a loopback test for a specific port. Loopback tests require that you disable the port. The following loopback tests are available:
Internal
Random data is sent to the HBA port, then returned without transmitting through the port. The returned data is validated to determine port operation. Errors may indicate an failed port.
External
For this test, a loopback connector is required for the port. Random data is sent to the HBA port. The data transmits from the port then returns. The returned data is validated to determine port operation. Errors may indicate a failed port.
Performing loopback tests through BCU

Use the diag --loopback BCU command test to verify port function through a loopback test.
bcu diag --loopback <port_id> [-t <loopback_type>][<duration>][-s <speed>] [-c <frame_count>] [-p <pattern>]
where: port_id loopback type speed duration frame count -p pattern ID of the port that you want to run the test. internal, external, serdes For 8 Gbps HBA, this is 2, 4, or 8. For 4 Gbps HBA, this is 1, 2 or 4. Length of time between blinks. Integer from 0- 4,294,967,295. Default is 8192. Hex number. Default value is A5A5A5A5.
Performing loopback tests through HCM

Use the Port Tests tab on the Diagnostics dialog box to perform a loopback test. 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Port Tests tab. 4. Select Loopback Test. You can modify the following test parameters
Subtest: three options: Internal, Serdes, and External. Link Speed: For 8G HBA: 2G, 4G and 8G; For 4G HBA: 1G, 2G and 4G Frame Count: Integer from 0- 4,294,967,295. Default value is 8192 Data Pattern: 8 Hex number. Default value is A5A5A5A5
5. Click Start.
41
Diagnostics
PCI loopback test

Use the BCU or HCM to perform a PCI loopback test for a specific port. In this test, a data pattern is sent from the host to HBA firmware through the PCI bus. The returned data is validated to determine PCI operation.
Performing loopback tests through BCU

Use the diag --pciloopback BCU command to perform a PCI loopback test.
bcu diag --pciloopback <port_id> [-p <pattern>] [-c <frame_count>]
where: port_id pattern frame count ID of the port from which you want to run the test. Hex number. Integer.
Performing loopback tests through HCM

Use the Port Tests dialog box to perform a PCI loopback test. 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Port Tests tab. 4. Select PCI Loopback Test. You can modify the following parameters
Frame count: Integer Data pattern: Hex number. Test cycle: The number should be positive and the default is 1.
5. Click Start.
Memory test
Use the BCU or the HCM to perform a memory test for the HBA.
NOTE
Performing the Memory test disables the HBA.
Performing a memory test through BCU

Use the diag --memtest command to test the adapters memory blocks.
diag --memtest <ad_id>
where: ad_id Serial number of adapter.
42
Diagnostics
Performing a memory test through HCM

Use the Port Tests dialog box to perform a memory test. 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Port Tests tab. 4. Select Memory Test. 5. You can specify the test cycle using a positive number. 6. Click Start.
HBA temperature
Use the BCU diag --tempshow command to read the adapters temperature sensor registers.
diag --tempshow <ad_id>
ad_id - Serial number of adapter.
Ping end points

Use the BCU and HCM to ping a Fibre Channel end point from an HBA port to determine the basic connectivity to the remote port and monitor network latency.
Ping end points through BCU

Use the fcdiag --fcping BCU command to test the connection to a Fibre Channel end point.
fcdiag -fcping <port_id> <rpwwn> [-l lpwwn]
where: port_id rpwwn lpwwn ID of the HBA port from which you want to ping the remote port. Remote port WWN that you want to ping. Logical port WWN. 0 indicates the base port.
Ping end points through HCM

Use the Protocol tab on the Diagnostics dialog box to test the connection to Fibre Channel end points. 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Protocol Tests tab. 4. Select FC Ping Test. 5. Select the HBA port and target that you wish to ping. 6. Enter a test cycle if applicable.
43
Diagnostics
7.
Click Start.
Trace route
Use the BCU and HCM to report the SAN path between the HBA and remote end point.
Trace the route through BCU

Use the fcdiag --traceroute BCU command to trace the route between end points.
fcdiag -fctraceroute <port_id> <rpwwn> [-l lpwwn]
where: port_id rpwwn lpwwn ID of the port from which you want to trace the route. Remote port WWN that you want to ping. Logical port WWN. 0 indicates the base port
Trace the route through HCM

Use the FC Trace Route feature to trace the route between the HBA port and a target port: 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Protocol Tests tab. 4. Select FC Trace Route. 5. Select the HBA port and target for which you wish to trace the route. 6. Enter a test cycle if desired. 7. Click Start.
Echo test
Use the BCU and HCM to initiate an echo test between the HBA port and a Fibre Channel end point. This sends an ECHO command and response sequence between the HBA port and target port to verify connection with the target.
Perform echo test through BCU

Use the fcdiag --fcecho BCU command to initiate an echo test between the HBA and remote port.
fcdiag -fcecho <port_id> <rpwwn> [-l lpwwn]
where: port_id rpwwn lpwwn ID of the port for which you want to perform the test. Remote port WWN that you want to ping. Logical port WWN. 0 indicates the base port; otherwise.
44
Collecting SFP data
Perform echo test through HCM

Use the Echo Test feature to initiate an echo test between the HBA port and a Fibre Channel end point: 1. Launch the HCM. 2. Select Configure > Diagnostics 3. Click the Protocol Tests tab. 4. Select Echo Test. 5. Select the HBA port and target port for the test. 6. Enter a test cycle if applicable. 7. Click Start.
SCSI test
Use the fcdiag --scsitest BCU command to test the SCSI link between the HBA and remote port.
fcdiag -scsitest <port_id> <rpwwn> [-l lpwwn]
where: port_id rpwwn lpwwn ID of the port for which you want to test the SCSI link. Remote port WWN that you want to ping. Logical port WWN. 0 indicates the base port; otherwise.
Collecting SFP data

This section provides an overview of BCU commands and HCM features that provide information, on small form factor pluggable (SFP) transceivers.
SFP diagnostics
SFP diagnostics provide detailed information on the SFP transceiver for a selected port, such as its health status, port speed, connector type, minimum and maximum distance, as well as details on the extended link.
Initiating diagnostics through BCU

Use the diag --sfpshow BCU command to display detailed attributes for a specific SFP transceiver.
diag --sfpshow <port_id>
where: port_id ID of the port for which you want to display SFP attributes.
45
Collecting port data
Initiating diagnostics through HCM

Use the port SFP properties dialog box to display properties for a selected small form-factor pluggable (SFP) transceiver. 1. Launch the HCM. 2. Select a port in the device tree. 3. Click the SFP tab in the right pane.
Port power on management (POM)

Use the Port POM properties panel to monitor the SFP attributes. A notification is given for any parameters that are not within the configured power, temperature, voltage, and current specification policy. 1. Select a port in the device tree. 2. Click the POM tab in the right pane.

This section provides an overview of BCU commands and HCM features that provide information on HBA ports. such as port WWN, node WWN, port type, configured speed, operating speed, configured topology, operating topology. link and port beaconing state, and other information.
Base port properties

Use the Base Port Properties panel to display information about a selected base HBA port, such as port number, WWN, node WWN, offline or online state, role of port (for example, FCP initiator), Fibre Channel address, and WWN of the attached switch. 1. From the device tree, select a base port. 2. In the right panel, click the Base Port Properties tab.
Remote port properties

Use the Remote Port Properties panel to display properties that are associated with the remote port, such as WWN, node WWN, port name, Fibre Channel address, frame data field size, online or offline state, role (such as target or initiator), remote device information, QoS priority, QoS flow, and target ID. 1. From the device tree, select a remote port. 2. Click the Remote Port Properties tab in the right panel. If it is a target port, two tabs display in the right pane: Properties and LUNs.
46
Logical port properties

Use the LPorts Properties panel to display properties associated with a logical port, such as port and node WWN, Fibre Channel address, online or offline state, and name server activity. 1. From the device tree, select a logical port. 2. Click the LPORTs Properties tab in the right panel.
Virtual port properties

Use the Virtual Port Parameters properties panel to display the properties associated with a virtual port, such as port and node WWN, Fibre Channel address, offline or online state, role (such as FCP initiator), and attached switch WWN. 1. From the device tree, select a virtual port. 2. The virtual port parameters property panel displays.
Port log
Use the debug --portlog BCU command to display a log of Fibre Channel frames and other main control messages that were sent out and received on a specific port. You can use this information to isolate HBA and Fibre Channel protocol problems.
debug --portlog <port_id>
where: port_id The ID of the port for which you want to display the port log.
If the port log is disabled, a warning message displays. Use the debug -portlogctl command to enable and disable the port log.
NOTE
Port list
Use the port --list BCU command to list all physical ports on the HBA along with their physical attributes, such as PWWN, Fibre Channel address, port type, speed, and state.
port --list <port_id>
where: port_id ID of the port for which you want to display information.
Port query
Use port --query BCU command to display port information, such as WWN, NWWN, state, current and configured speed, topology, received and transmitted BB_Credits, and beacon status.
port --query <port_id>
port_id - ID of the port for which you want to display information.
47
Authentication settings
Port speed
Use port --speed BCU command to display the current port speed setting, such as 1, 4, or 8 Gbps.
port --speed <port_id <1|2|4|8|auto>
where: port_id <1|2|4|8|auto> ID of the port for which you want to display port speed. The speed settings, with auto being autosensing mode.
Authentication settings
Use the Brocade CLI utility (BCU) or the HCM GUI to display the HBA authentication settings and status.
Displaying settings through HCM

Use the Fibre Channel Security Protocol Configuration dialog box display authentication settings: 1. Select a port from the device tree. 2. Select Configure > FC-SP > Authentication. The Fibre Channel Security Protocol Configuration dialog box displays. This displays the current CHAP secret, hashing algorithm, and group value.
Displaying settings through BCU

Use the BCU auth --show command to display authentication settings.
auth --show <port_id>
where: port_id ID of the port for which you want to display authentication settings.
QoS and target rate limiting settings

Target rate limiting throttles the Fibre Channel Protocol (FCP) read traffic rate to slow-draining targets to reduce or eliminate network congestion and alleviate I/O slowdowns to faster targets. Quality of Service (QoS) works in conjunction with the QoS feature on Brocade switches to assign traffic priority (high, medium (default), low) to a given source and destination traffic flow. Use the following BCU commands and HCM to determine current status and configuration for these features.
48
QoS and target rate limiting settings
BCU commands
Use the following BCU command to determine Target Rate Limiting speed and enabled status.
ratelim --query <port-id>
where: port_id ID of the port for which you want to display target rate limiting settings.
Use the following BCU command to display QoS and target rate limiting enabled status and
target rate limiting default speed.
port --query <port-id>
where: port_id ID of the port for which you want to display port information.
Use the following command to display QoS status and information for a port.
bcu qos --query <port_id>
where: port_id ID of the port for which you want to display target rate limiting settings.
Use the following command to determine operating speed of the remote port, QoS priority, and
target rate limiting enforcement:
bcu rport --query
where: port_id Specifies the ID of the port for which you want to query attributes of a remote port.
HCM
Use the Port Properties panel in HCM to display configured QoS parameters.
To open the Port Properties panel: 1. Select a port in the device tree. 2. Click the Properties tab in the right pane.
Use the Remote Port Properties panel in HCM to display information on target rate limiting and
QoS for the remote port. To open the Remote Port Properties panel: 1. From the device tree, select a remote port (target or initiator). 2. Click the Remote Port Properties tab in the right pane.
49
Persistent binding
Persistent binding
Persistent binding is a feature of Fibre Channel (FC) host bus adapters that enables you to permanently assign a system SCSI target ID to a specific FC device, even though the devices ID on the FC loop may be different each time the FC loop initializes. Persistent binding is available in the Windows and VmWare environments only. Use the HCM or BCU to display target ID mapping for an HBA port. BCU Use the pbind --list BCU command to query the list of mappings for persistent binding on a specific port.
pbind --list <port_id> <pwwn>
where: port_id pwwn ID of the port for which you want to query mappings. Port World Wide Name
HCM Use the Persistent Binding dialog box to determine SCSI target ID mappings, perform the following steps: 1. Launch the HCM. 2. Select an HBA or port from the device tree. 3. Select Configure > Persistent Binding. You can also select and right-click on an HBA or port in the device tree and select Persistent Binding from the list.
50
Chapter
Performance optimization
This chapter provides information and tools for optimizing your HBA performance.
In this chapter
Linux tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solaris tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Windows tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VMware tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51 52 52 53
Linux tuning
Linux disk I/O scheduling reorders, delays, and merges requests for disk I/O to achieve better throughput and lower latency than would happen if all the requests were sent straight to the disk. Linux 2.6 has four different disk I/O schedulers: noop, deadline, anticipatory and completely fair queuing. Enabling the noop scheduler avoids any delays in queuing of I/O commands. This helps in achieving higher I/O rates by queuing multiple outstanding I/O requests to each disk. To enable the noop scheduler, run the following commands on your system.
for i in /sys/block/sd[b-z]/queue/scheduler do echo noop > $i done
You must disable the default scheduler because it is not tuned for achieving the maximum I/O performance. For performance tuning on Linux, refer to the following publications
NOTE
Workload Dependent Performance Evaluation of the Linux 2.6 IO Schedulers

Heger, D., Pratt, S., Linux Symposium, Ottawa, Canada, July 2004
Optimizing Linux Performance

HP Professional Books, ISBN: 0-13-148682-9
Performance Tuning for Linux Servers

Sandra K. Johnson, Gerrit Huizenga, Badari Pulavarty, IBM Press, ISBN: 013144753X
Linux Kernel Development

Robert Love, 2nd edition, 2005
51
Solaris tuning
Solaris tuning
To increase I/O transfer performance, set the following parameters on your system:
Set the maximum device read/write directive (maxphy). Set the Fibre disk maximum transfer parameter (ssd_max_xfer_size).
Please refer to Sun StorageTek SAM File System Configuration and Administration Guide document for details of the two parameters.
Windows tuning
Windows tuning involves configuring the driver and operating system tunable parameters.
Driver tunable parameters

You can manipulate several driver parameters to increase performance.
bfa_lun_queue_depth (outstanding I/O requests per LUN)

The driver uses a default LUN queue depth value of 32. This value is chosen to reflect the average operating I/O load in most scenarios. Storport manages the I/O throttling at per-LUN level to guarantee the configured queue depth. During performance testing with specific high-end array LUNs, it may be necessary to increase this I/O queue depth to a much larger value. Microsoft recommends a value of 96 during high-performance testing scenarios. If the queue depth is not sufficient, then the user will notice performance degradation. The queue depth can be configured at the following registry location with any value within the range 1 254. HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\bfad\Parameters\Device\bfa_lun_ queue_depth
Interrupt moderation using interrupt coalescing feature

Moderating interrupts can often result in reduced CPU load on the host but, unless interrupt moderation is performed intelligently, the CPU savings might increase latency. The default values for the fibre channel port interrupt attributes are configured as below.
Interrupt coalescing
Default: ON
Interrupt delay
Default: 1 micro second Valid Range: 0 1125 micro seconds (Note that the value of 0 disables the delay timeout interrupt.)
Interrupt latency
Default: 1 micro second
52
VMware tuning
Valid Range: 0 225 micro seconds (Note that the value of 0 disables the latency monitor timeout interrupt.)
Interrupt Coalescing
When this feature is turned off, I/O completion requests are not coalesced by the firmware. While this helps reduce I/O latency, the host CPU will frequently be interrupted, leading to a slower system response under heavy I/O load (more than 7000 I/Os per second). When this feature is turned on, the HBA will not interrupt the host until Interrupt delay duration. Interrupt delay, together with Interrupt latency, helps to reduce the number of interrupts that the host CPU processes per second, leading to improved overall CPU utilization. However, if the number of interrupts handled between the Interrupt delay period is relatively smaller, then this will result in performance degradation as the I/O completion process has slowed down. The BCU ioc -intr command can be used to configure these interrupt attributes for the desired port.
bcu ioc --intr <ioc_id> <--coalesce|-c> {on | off} [<Latency> <Delay>]
Message signaled interrupts (MSI-X)

All Brocade HBAs support MSI-X, an eXtended version of MSI defined in PCI 3.0 specification. MSI-X helps improve overall system performance by contributing to lower interrupt latency and improved host CPU utilization. MSI-X is supported in Windows Vista and Windows Server 2008. To enable MSI-X, set the following registry key value to 0.
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\bfad\Parameters\Device\msix_ disable
OS tunable parameters
Please see the section Storage Stack Drivers in Disk Subsystem Performance Analysis for Windows Server 2003 optimizations located on the following website. http://download.microsoft.com Please see the sections Performance Tuning for Storage Subsystem and I/O Priorities in Performance Tuning Guidelines for Windows Server 2008 located on the following website. http://www.microsoft.com
VMware tuning
For performance tuning on VMware, refer to the following publications on the VMware website at www.vmware.com:
Performance Tuning Best Practices for ESX Server 3. Refer to the following sections: - Storage Performance Best Practices - Related Publications Fibre Channel SAN Configuration Guide. Refer to Using ESX Server with SAN: Concepts.
53
VMware tuning
54
Index
A
adapter list command, 28 adapter query command, 28 application log, 32, 33 authentication settings, 48
B
BCU commands adapter list, 28 adapter query, 28 port list, 47 port query, 47 port speed, 48 to collect data, 25 beaconing, 40 enabling through BCU, 40 enabling through HCM, 40
C
collecting data using BCU, 28 collecting data using event logs, 30 collecting data using Fabric OS commands, 28
D
data collecting data with BCU and HCM, 25 collecting using host commands, 24 data to provide support, 24 device manager, 15
diagnostics, 40 beaconing, 40 enabling through BCU, 40 enabling through HCM, 40 echo test enabling through BCU, 44 enabling through HCM, 45 HBA, 40 HBA temperature, 43 loopback tests, 41 enabling through BCU, 41 enabling through HCM, 41 memory test, 42 enabling through BCU, 42 enabling through HCM, 43 PCI loopback tests, 42 enabling through BCU, 42 enabling through HCM, 42 ping end points, 43 enable through HCM, 43 enabling through BCU, 43 SCSI test, 45 SFP, 45 enable through BCU, 45, 46 enable through HCM, 46 trace route, 44 enable through HCM, 44 enabling through BCU, 44 driver install errors, 13 driver installation verify using Linux commands, 15 verify using Solaris commands, 16 verify using VMware command, 16 driver tunable parameters for Windows, 52, 53
E
echo test, 44 enable through HCM, 45 enabling through BCU, 44 error message when removing Windows driver, 13 error when rolling back driver, 14
55
event logs, 30 HCM, 32 host system, 30 host system logs adjust logging level, 32 syslog support, 34 Windows event log support, 34 event message files, 31
I
information gathering, 2 installation confirming driver installation, 14 driver errors, 13 problems, 12 verifying, 12 IOC statistics, 35 displaying through BCU, 35 displaying through HCM, 35 isolating problems, 5
F
Fabric OS commands, 28 fabric statistics, 36 displaying through BCU, 36 displaying through HCM, 36 FCIP initiator mode statistics, 37 displaying through BCU, 37 displaying through HCM, 37 Fibre Channel links verifying, 20 files needed for bfad.sys message, 13
L
Linux tuning, 51 log application, 32 master, 32 logical port properties, 47 logical port statistics, 37 displaying through BCU, 38 displaying through HCM, 38 logs, 32 application, 33 event, 30 HCM, 32 host system, 30 master log severity levels, 32 port, 47 syslog support, 34 Windows event log support, 34 loopback tests, 41 enabling through BCU, 41 enabling through HCM, 41
H
HBA diagnostics, 40 fabric OS support, viii operating system support, viii PWWN, xiii serial number, xii storage support, viii supported models, viii switch support, viii HBA event message files, 31 HBA memory test, 42 HBA properties, 15 HBA properties panel, 28 HBA statistics, 34 HCM cannot connect with agent, 17 HCM logs, 32 HCM options to collect data, 25 host commands for collecting data, 24 host freezes or crashes, 16 host system logs, 30
M
master log, 32 master log severity levels, 32 memory test, 42 enabling through BCU, 42 enabling through HCM, 43
O
operating system support, viii
56
P
PCI loopback tests, 42 enabling through BCU, 42 enabling through HCM, 42 performance optimization Linux tuning, 51 Solaris tuning, 52 VMware tuning, 53 Windows tuning, 52 persistent binding settings, 50 ping end points diagnostics, 43 enable through HCM, 43 enabling through BCU, 43 port data, 46 port list command, 47 port log, 47 port properties base, 46 logical, 47 remote, 46 virtual, 47 port properties panel, 46 port query command, 47 port speed command, 48 port statistics, 34 enable through BCU, 35 enable through HCM, 35
problem driver event messages, 11 errors when installing driver, 10 fabric authentication failures, 8 failed to connect to agent error, 11 files needed for bfad.sys message, 13 general, 5 HBA not in fabric, 10 HBA not registering with name server, 11 HBA not reported under PCI subsystem, 6 HCM fails connection with agent, 17 host not booting from remote LUN, 10, 14 host system freezes, 10, 16 I/Os not failing over on path failure, 9 installation, 12 installer program does not autorun, 10, 13 isolating, 5 loss of sync and signal, 8 LUN not visible, 9 maximum IOPS numbers low, 9 no HBAs reported, 6 operating system errors, 10 port link not up, 7 QoS performance issues, 10 target not visible, 9 unable to create NPIV ports, 9 virtual devices not in name server, 10 problem information, 2 properties panel for HBA, 28
Q
QoS settings, 48 QoS statistics, 39 displaying through BCU, 39 displaying through HCM, 39
R
remote port properties, 46 remote port statistics, 36 displaying through BCU, 37 displaying through HCM, 37 resolving installation problems, 12
S
SCSI target ID mappings, 50
57
SCSI test, 45 serial number location, xii SFP diagnostics, 45 enable through BCU, 45 enable through HCM, 46 Solaris tuning, 52 statistics fabric, 36 displaying through BCU, 36 displaying through HCM, 36 FCIP initiator mode, 37 displaying through BCU, 37 displaying through HCM, 37 IOC, 35 displaying through BCU, 35 displaying through HCM, 35 logical port, 37 displaying through BCU, 38 displaying through HCM, 38 port, 34 display through BCU, 35 display through HCM, 35 QoS displaying through BCU, 39 displaying through HCM, 39 remote port, 36 displaying through BCU, 37 displaying through HCM, 37 virtual port, 38 displaying through BCU, 39 displaying through HCM, 39 statistics for HBAs, 34 support data to provide, 24 support save differences between HCM, BCU, and browser, 27 on port crash event, 27 using, 25 using through BCU, 26 using through browser, 27 using through HCM, 26 using through port crash event, 27 syslog support, 34
trace route, 44 enable through HCM, 44 enabling through BCU, 44 troubleshooting gathering information, 2 general problems, 5 introduction, 1 using this manual, 1
V
virtual port properties, 47 virtual port statistics, 38 virtual port statsitcs displaying through BCU, 39 displaying through HCM, 39 VMware tuning, 53
W
Windows driver tunable parameters, 52 Windows event log support, 34 Windows tuning, 52 WWPN of HBA, xiii
T
target rate limiting settings, 48 target statistics, 37 technical help for product, xii temperature diagnostics, 43
58

HBA Troubleshooting Guide

Uploaded by

Copyright:

Available Formats

HBA Troubleshooting Guide

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

HBA Troubleshooting Guide

Uploaded by

Copyright:

Available Formats

53-1000885-02 January 30, 2009

Fibre Channel HBA

Brocade Communications Systems, Incorporated

About this Document

Tools for Collecting Data

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

About this Document

How this document is organized

Chapter 1, Introduction to troubleshooting provides an introduction and approach to

Chapter 2, Isolating Problems provides information on common HBA problems and

Chapter 4, Performance optimization contains guidelines for optimizing HBA performance on

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Supported hardware and software

Install only Brocade-branded SFPs in these HBAs.

Fabric OS and switch support

Host operating system support

Windows Vista (HCM only) Windows XP (HCM only)

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Whats new in this document

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Command syntax conventions

[] variable ... value |

Notes, cautions, and warnings

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Notice to the reader

Referenced Trademarks and Products

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Other industry resources

Getting technical help

Description of any troubleshooting steps already performed and the results.

HBA Properties tab in HCM.

BCU adapter --list command.

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Brocade BIOS Configuration Utility.

HCM Port Properties dialog box.

The following BCU commands:

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

How to use this manual for troubleshooting

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Gathering problem information

Gathering problem information

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Gathering problem information

Topology map details

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Gathering problem information

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Chapter 3, Tools for Collecting Data in this manual.

Fabric OS Administrators Guide

Fabric OS Command Reference Manual.

Fabric OS Troubleshooting and Diagnostic Guide.

Your hosts operating system documentation and help system.

Brocade Fibre Channel HBA Troubleshooting Guide 53-1000885-02

Troubleshooting general problems

HBA not reported under servers PCI subsystem.