upgrade-cluster

sagaragas 的头像作者 sagaragas

Upgrade Talos Linux, Kubernetes version, or cluster components safely

📖 中文导读:Upgrade Talos Linux, Kubernetes version, or cluster components safely 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: upgrade-cluster description: Upgrade Talos Linux, Kubernetes version, or cluster components safely

Upgrade Cluster Skill

Safely upgrade Talos Linux, Kubernetes, and cluster components.

When to Use

  • User wants to upgrade Kubernetes version
  • User wants to upgrade Talos Linux
  • User wants to upgrade Cilium or other core components
  • Security patches needed

Pre-Upgrade Checklist

  1. Check current versions:

    talosctl -n 172.16.1.50 version
    kubectl version
    cilium version
    flux version
    
  2. Verify cluster health:

    kubectl get nodes
    talosctl -n 172.16.1.50 health
    flux get ks -A | grep -v "Applied"
    
  3. Check for pending HelmReleases:

    flux get hr -A | grep -v "Release"
    

Talos Upgrade Process

  1. Update Talos version pin in talos/talenv.yaml
  2. Generate new configs:
    task talos:generate-config
    
  3. Upgrade control plane nodes one at a time:
    task talos:upgrade-node IP=172.16.1.50
    # Wait for node to rejoin
    kubectl get nodes -w
    task talos:upgrade-node IP=172.16.1.51
    task talos:upgrade-node IP=172.16.1.52
    
  4. Upgrade worker nodes:
    task talos:upgrade-node IP=172.16.1.53
    

Kubernetes Upgrade Process

  1. Update Kubernetes version pin in talos/talenv.yaml
  2. Regenerate configs:
    task talos:generate-config
    
  3. Apply to control plane:
    task talos:upgrade-k8s
    

Component Upgrades (GitOps)

Components upgrade automatically via Renovate PRs:

  • Review Renovate PR for the component
  • Check release notes for breaking changes
  • Merge PR - Flux will reconcile

Manual Component Upgrade

  1. Update version in relevant HelmRelease
  2. Commit and push
  3. Monitor: flux get hr <name> -n <namespace> -w

Rollback Procedures

Talos Rollback

talosctl -n <node-ip> rollback

Flux HelmRelease Rollback

flux suspend hr <name> -n <namespace>
# Revert git commit
git revert HEAD
git push
flux resume hr <name> -n <namespace>

Post-Upgrade Verification

kubectl get nodes -o wide
flux get ks -A
flux get hr -A
cilium status