s0sta
EN NL عربي
Filters
How to search
  • Plain words find names, texts and examples.
  • "two words" finds the exact phrase.
  • --name or -x finds an option.
  • #word looks only inside commands.
  • @grep finds lessons and labs that use grep.
  • Filters: cat:network · in:labs · in:notes · level:rhcsa · ch:8 · root:yes. Put ! before a word to leave it out. OR between words means one of them.
  • ↑ ↓ to move · Enter to open · Esc to close · / to search

KubernetesCloud / DevSecOps series

Fix: Kubernetes CrashLoopBackOff

CrashLoopBackOff means the container starts, crashes, and Kubernetes keeps restarting it with an increasing delay. The status tells you it is failing; the logs and events tell you why. This is the most common pod problem, and a CKA staple.

Quick reference

Diagnose — Why did it crash?

  • kubectl logs <pod> --previous — Logs of the crashed container, not the new one.
  • kubectl describe pod <pod> — Events: OOMKilled, failed probe, mount error.
  • Check the exit code — 137 = OOM/killed, 1 = app error, 0 = exited cleanly.

Fix — Match the cause

  • App error in logs — Fix the config, missing env, or bad connection string.
  • OOMKilled (137) — Raise the memory limit, or fix the leak.
  • Failing liveness probe — The probe is too strict or the port is wrong.

Prevent — Fewer crash loops

  • Realistic probes — Give slow apps an initialDelaySeconds.
  • Right-size memory — Set requests and limits from real usage.
  • Fail fast with a clear log — Make the app say why it quit.

Explained

Diagnose

kubectl logs <pod> --previous is the key command. In a crash loop the current container has just started and has nothing useful; --previous shows the output of the one that actually crashed. The real error is almost always here: a missing environment variable, a refused database connection, a bad config file. kubectl describe pod <pod> shows the Events at the bottom, which catch the causes that are not in the app log: OOMKilled, a failing liveness probe, or a volume that would not mount. The exit code is a strong hint. 137 means killed, usually out of memory; a code of 1 or 2 is an application error; 0 means it exited cleanly, which is wrong for a long-running service.

Fix

An app error in the logs: fix the config, the missing secret, or the connection string it cannot reach. OOMKilled: raise the memory limit, or fix the leak that makes it grow. A failing liveness probe: the probe may be too strict, or pointing at the wrong port or path, killing a healthy-but-slow app.

Prevent

Realistic probes with an initialDelaySeconds for apps that take time to start. Right-sized memory requests and limits, based on real usage. An app that fails fast and logs clearly why, so the next crash loop takes 30 seconds to diagnose.

In one line

Read --previous, not the restart count. The crashed container's log holds the answer.

The first command

kubectl logs <pod> --previous

More notes: Kubernetes

Open in the notes app: search, filters, copy This note on GitHub