CrashLoopBackOff

Your container keeps crashing and Kubernetes keeps restarting it.

Why this happens

Kubernetes expects containers to run continuously. When a container exits (whether from an error or intentionally), the kubelet restarts it with exponential backoff delays (10s, 20s, 40s, up to 5 minutes). CrashLoopBackOff is the status shown during these increasing wait periods between restart attempts.

Fix 1: Check the logs

kubectl logs my-pod
kubectl logs my-pod --previous  # Logs from the crashed container

Fix 2: Check the pod events

kubectl describe pod my-pod

Look at the “Events” section and “Last State” for exit codes.

Fix 3: Common causes

• Exit code 1 — application error. Check logs.
• Exit code 137 — out of memory (OOMKilled). Increase memory limits.
• Exit code 0 — container exited successfully but K8s expects it to keep running. Make sure your app doesn’t exit.

resources:
  limits:
    memory: "512Mi"  # Increase if OOMKilled
  requests:
    memory: "256Mi"

Fix 4: Debug with a shell

kubectl run debug --image=myimage --command -- sleep 3600
kubectl exec -it debug -- /bin/sh

Alternative solutions

Override the entrypoint to keep the container alive for debugging:

kubectl run debug --image=myimage --overrides='{"spec":{"containers":[{"name":"debug","command":["sleep","infinity"]}]}}'

Check liveness/readiness probes — misconfigured probes can kill healthy containers. Temporarily remove probes to confirm the app starts correctly.

Prevention

• Always test your container image locally with docker run before deploying to Kubernetes.
• Set appropriate resource requests/limits based on actual usage metrics.

Related: Kubernetes: OOMKilled fix · Docker vs Kubernetes