Kubernetes CrashLoopBackOff: How to Diagnose and Fix It
CrashLoopBackOff means a container starts, exits, and Kubernetes keeps restarting it with growing back-off. It's a symptom — the real cause is in the events and the previous container's logs.
Step 1 — Read the events and last state
kubectl describe pod <pod>Scroll to Last State and the Events at the bottom. Note the Reason and Exit Code — they point straight at the cause (for example Reason: OOMKilled, Exit Code: 137 means out of memory).
Step 2 — Read the crashed container's logs
kubectl logs <pod> --previousThe --previous flag is essential — it shows the logs of the container instance that crashed, not the fresh one. This is where an application stack trace or "config not found" error appears.
Step 3 — Match the symptom to the cause
- Application error on startup — a stack trace in the logs; fix the code or the input it's choking on.
- Missing config / secret / env var — the app exits because a required value isn't mounted; check the ConfigMap/Secret and volume mounts.
- OOMKilled (exit 137) — the container exceeded its memory limit; raise
resources.limits.memoryor fix the leak. - Failing liveness probe — the app is fine but the probe path/port/timing is wrong, so the kubelet keeps killing it; fix the probe.
- Bad command/entrypoint or missing dependency — exit code 127/126; correct the command or image.
Step 4 — Fix and confirm
Apply the fix, then watch it recover:
kubectl get pods -wThe pod should move to Running and stay there, with the restart count holding steady.
How Tech Matrix solves this in ~60 seconds
CrashLoopBackOff has five common causes and the slow part is correlating the exit code, events, probe config and logs. Tech Matrix reads the pod's describe output, the previous-container logs and the manifest, and tells you which of the five it is — and the exact fix — instead of you grepping across kubectl output.
Frequently asked questions
The container starts, exits, and Kubernetes restarts it repeatedly with an increasing back-off delay. The crash itself has a separate cause shown in the events and previous logs.
Run 'kubectl describe pod
Exit 137 is OOMKilled — the container exceeded its memory limit. Raise resources.limits.memory or fix the memory leak.