k8s-health-check

name	k8s-health-check
description	Use when asked to check cluster health, "what's wrong in k8s", "check pods", or "/k8s-health". Runs a structured Kubernetes health check across nodes, workloads, and recent events.

k8s-health-check

When to use

• "check cluster health" / "/k8s-health"
• "what's wrong in Kubernetes?"
• "why are pods crashing?"
• "check the namespace "

Inputs

• Optionally: namespace (default: all namespaces), cluster context, or a specific workload name.

Procedure

1. Confirm context. Verify the target cluster before running any commands:

   kubectl config current-context
   

   If it doesn't match the user's intent, ask before proceeding.

2. Cluster-level checks (run in parallel):

   kubectl get nodes -o wide                          # node status + roles
   kubectl top nodes                                  # CPU/mem pressure
   kubectl get events --all-namespaces \
     --field-selector type=Warning \
     --sort-by='.lastTimestamp' | tail -20            # recent warnings
   

3. Workload checks:

   # Not-running pods
   kubectl get pods -A --field-selector='status.phase!=Running,status.phase!=Succeeded'
   
   # Recent restarts (>3 restarts is a signal)
   kubectl get pods -A --sort-by='.status.containerStatuses[0].restartCount' | tail -20
   
   # Pending pods (scheduling issues)
   kubectl get pods -A --field-selector=status.phase=Pending
   

4. If a specific pod is unhealthy:

   kubectl describe pod <name> -n <ns>    # events, resource limits, node assignment
   kubectl logs <pod> -n <ns> --tail=50  # recent log output
   kubectl logs <pod> -n <ns> --previous # logs from crashed container
   

5. Report:

   ## Kubernetes Health Check — <context> — <timestamp>
   
   ### Nodes
   - X/Y nodes Ready
   - <any NotReady nodes and reason>
   
   ### Workload issues
   | Namespace | Pod | Status | Restarts | Issue |
   |---|---|---|---|---|
   | <ns> | <pod> | CrashLoopBackOff | 12 | OOMKilled |
   
   ### Recent warnings
   - <top 3-5 warning events>
   
   ### Recommended actions
   - [ ] <describe pod X for OOMKilled details>
   - [ ] <increase memory limit or fix leak>
   

Guardrails

• Always confirm kubectl config current-context matches the intended cluster.
• Never run kubectl delete or any mutating command without explicit user confirmation.
• kubectl exec into pods only if the user explicitly asks.
• If kubectl top is unavailable (metrics-server not installed), note it and skip resource pressure checks.