Q34: How do you systematically troubleshoot CrashLoopBackOff, ImagePullBackOff, and Pending Pods? - DevOps & Kubernetes (50 Qs)

Step-by-Step Production Troubleshooting Guide

When a Pod fails to reach Running state, execute this diagnostic flow before making code or configuration changes.

scss

[ Pod Status Abnormal ]
          |
          +---> [ Pending ] --------> 1. kubectl describe pod (Check Scheduler Events, CPU/Mem, Affinity)
          |
          +---> [ ImagePullBackOff ] -> 2. Check Image Name, Tag, and imagePullSecrets
          |
          +---> [ CrashLoopBackOff ] -> 3. kubectl logs --previous (Inspect crash stack trace)

Diagnostic Checklist Matrix

Error State	Root Cause	Immediate Diagnostic Command	Fix Action
`Pending`	Node resources exhausted, node selector mismatch, or unfulfilled PVC.	`kubectl describe pod <pod>	grep -A 10 Events:`
`ImagePullBackOff`	Tag typo, non-existent repository, or registry authentication failure.	`kubectl describe pod <pod>`	Verify image tag; check `imagePullSecrets` for private registries (ECR/DockerHub).
`CrashLoopBackOff`	Container starts, encounters fatal error (exit code 1 or 137), and exits.	`kubectl logs <pod> --previous`	Inspect application logs just before crash; verify DB connection strings and config files.
`CreateContainerConfigError`	Missing ConfigMap or Secret referenced in Pod spec.	`kubectl describe pod <pod>`	Create the missing ConfigMap/Secret or fix key names.

Pro-Tip: Ephemeral Debug Containers

For distroless or hardened containers without a shell, attach an interactive debug container directly into the pod’s network and process namespace:

bash

kubectl debug -it <crashing-pod> --image=nicolaka/netshoot --target=app-container