troubleshoot

name	troubleshoot
description	Troubleshoot a specific infrastructure component — resolve namespace, check pods, events, logs, and Argo CD sync status
argument-hint	<component>

Troubleshoot Component

Diagnose issues with $ARGUMENTS[0].

Steps

1. Resolve component → namespace:

   Component	Namespace
   cilium	kube-system
   istio, istiod	istio-system
   argocd	argocd
   prometheus, grafana, loki, tempo, otel-collector	monitoring
   cert-manager	cert-manager
   sealed-secrets	sealed-secrets
   keycloak	platform-auth-prod
   backstage	backstage

   If component is an app name, check app-prod (per-app namespaces such as app-kensan may also exist).

2. Check Argo CD Application status:

   kubectl get application $ARGUMENTS[0] -n argocd -o yaml
   

   • Look at sync status, health status, and any error conditions

3. Check pods:

   kubectl get pods -n <namespace> -l app.kubernetes.io/name=$ARGUMENTS[0]
   

   • If no label match, fall back to kubectl get pods -n <namespace>

4. Check events:

   kubectl get events -n <namespace> --sort-by='.lastTimestamp' | tail -20
   

5. Check logs (if pods exist but unhealthy):

   kubectl logs -n <namespace> <pod-name> --tail=50
   

6. Diagnosis:

   • Summarize findings
   • Suggest remediation steps
   • If GitOps-related, remind that fixes should go through Git (not kubectl apply)