CrashLoopBackOff indicates that a container in a Deployment-managed Pod keeps crashing immediately or shortly after startup. Kubernetes restarts it, then waits longer between attempts to avoid continuously consuming system resources. section 3.2.3
A crash includes more than an application error:
- Treats any terminated container process as a crash, even when it exits successfully. section 3.2.3
- Restarts containers that occasionally fail; frequent failures trigger progressively delayed attempts such as 10, 20, and 40 seconds. section 3.2.3
- Points toward startup failure or configuration error rather than a scheduling problem caused by unavailable resources. section 3.2.3
Deployment troubleshooting
A timeserver Deployment may show a Pod repeatedly restarting instead of reaching Running. Inspect its logs, including the previous container instance, to find why startup fails; use a Job for an intentional once-off task instead. section 3.2.3
kubectl logs -p $POD_NAME
Use previous-container logs when the current crashing instance has already terminated. section 3.2.3
