Upgrading Kubernetes

Keeping your clusters up to date can bring you security and stability but also a whole lot of anxiety. Good cluster hygiene is a must in k8s world but there are a whole lot of challenges with patching and upgrading the thing that runs your production systems! In this week's issue we are going to go through some sane paths and good tips on how to upgrade Kubernetes safely.

Issue #160

This article lays out the various strategies you can employ to upgrade your clusters. This article does a great job of going through the pros and cons of in-place, blue/green, rolling, and canary upgrades. They go into depth on all of the aforementioned strategies and really is the place to start when picking a strategy that is right for your org. ✅ ⭕

This article from our friends over at Pixie is a good resource for understanding all the different pieces of your k8s cluster that can / need to get upgraded at any given time. It can get a little overwhelming on where to start but this article does a good job of breaking down the steps you need to take and some sane approaches to it.

At the core of every cluster is etcd, the database that keeps everything running. Upgrading this can be a bit scary and daunting, and Kevin over at Teleport walks us through the complications that arise in interview format. The deeper you go the more your head spins and the more you appreciate how complex but yet simple k8s actually is. 😨

Lets not forget that kubeadm is a great option these days for rolling your own cluster. This handy guide walks us through exactly how to upgrade aforementioned clusters, remember networking / control plane / nodes the list goes on! It's crazy how far we have come that a tool as nice as kubeadm is built into the k8s ecosystem. 📚

The folks over at flant walk us through their experience with upgrading 150+ k8s clusters. There are some specific issues they hit that might not apply to all of us but great to read and try and think about similar problems we all might face in our specific setups. 🐜

Upgrading managed clusters can be just as tricky as clusters that you have rolled by yourself. This thoughtful writeup from the folks over at Rafay goes through upgrading nodes, DNS and much more, this is a must read before you upgrade a EKS cluster.

Important to have the conversation around OSS at big companies and making sure it's aligned with the communities best interest…fascinating conversation about OSS we all should pay attention to.