vixens-troubleshoot

To troubleshoot CrashLoopBackOff and OOMKilled pods in a Kubernetes homelab, run read-only kubectl commands to assess pod health and resource limits. You can then resolve the issue by committing configuration changes to Git and letting ArgoCD reconcile the deployment.

ArgoCD applications stuck OutOfSync can be fixed using GitOps by committing the desired state changes to your Git repository. You can then force an ArgoCD refresh to trigger reconciliation and bring the cluster state back into sync.

The best way to debug ingress controller errors and service accessibility problems in Kubernetes is to run read-only kubectl diagnostics to check service routing and ingress configurations. Fixes are then applied by updating Git and allowing ArgoCD to reconcile the manifests.

You should not use kubectl to directly modify resources when troubleshooting a Talos Linux cluster. Instead, run read-only kubectl commands for diagnostics and apply all fixes by committing changes to Git, maintaining strict GitOps principles for cluster reconciliation.

To safely restart applications and force ArgoCD refreshes for stuck applications, follow established recovery procedures that trigger reconciliation through Git updates. This ensures applications restart without violating GitOps principles or destabilizing the cluster state.

Pending pods in Kubernetes are typically caused by resource constraints or unschedulable nodes, diagnosed via read-only kubectl commands. GitOps workflows resolve them by committing adjusted resource requests or node affinity rules to Git for ArgoCD to reconcile.