Learnk8s’ Post

Learn how changing the kubelet root directory can cause issues with CSI drivers and other components, and using a special mount for ephemeral storage in Kubernetes is a better approach. More: https://lnkd.in/dtKdZfEG

Kubectl cp Command: How to Copy Files From Kubernetes Pods

In Kubernetes, weight is an integer between 1 and 100 that can be assigned to preferred rules. The Kubernetes scheduler prefers a node with the highest calculated weight value. For each node that matches a preference, its total affinity weight increases by the set amount. The node with the highest overall weight is allocated the pod. Weight can also be set to modify the influence of a resource type in the node score. For example, you can set the MostAllocated strategy for the NodeResourcesFit plugin using a scheduler configuration like this: apiVersion: https://lnkd.in/dBVM4Gsz kind: KubeSchedulerConfiguration profiles: - pluginConfig: - args: scoringStrategy: resources: - name: cpu weight: 1 name: NodeResourcesFit Weight-based routing can also be used to split traffic between versions of an application. This technique can simplify A/B testing, where a subset of users receive a new version of an application while the remaining users receive the current version.

When working with containers I find it works best to try and build them to be as stateless as possible. You should be able to pull in the data needed at runtime while the container instance itself is totally replaceable. You will need persistent volumes in some cases though. The terms Persistent Volume (PV) and Persistent Volume Claim (PVC) in Kubernetes often confuse people. PVs are more associated with the actual storage which PVCs are what ties an entity to a particular storage instance. As with most things i find it's better to see some examples to help understand things. Here is an article with examples of PV and PVC from Ashhadali that should be able to help.

Interesting performance comparison: ARM vs RISC-V with Docker and Kubernetes Reference paper: https://lnkd.in/dt9ygqMQ

This article explains the role of Kube-Proxy in mapping Kubernetes Services to Pods, its different modes, and how it works. It also provides a step-by-step example of inspecting IPtables rules for a ClusterIP Service. More: https://lnkd.in/gujamfj3

What is your favorite 🌟 container runtime? During one of our recent projects, we discovered that container runtimes can have a massive response time and infrastructure impact. ✅ Our lesson learned is to research if a better option is available instead of taking suggested container runtimes for granted. Performance comparison CRUN vs RUNC There is a dramatic performance benefit when using CRUN #CRUN is twice as fast as RUNC #CRUN consumes 50 times less memory #CRUN can run with 1 PID #RUNC can't be limited to 1 PID #Container #CRUN #RUNC https://lnkd.in/eZRVhyEF

Kompare is a Go CLI runner for comparing two clusters. It compares two Kubernetes clusters using kubeconfig to connect to them and compare existing objects in the two clusters based on flexible criteria passed via the command line. More: https://lnkd.in/g4QMRSHm

Join Saiyam Pathak‎ as he deploys vCluster onto a Kubernetes cluster with Sqlite as a backend and then shows how to use the latest vCluster PRO feature to create a virtual cluster with embedded ETCD as the backend in this quick video!  https://lnkd.in/gsmHhj7N