https://dev.to/yash_sonawane25

Learn how to debug your first Kubernetes crash like a pro (even if you’re just starting out).

You deployed your app to Kubernetes. Everything looked fine. And then—boom—it crashed. No logs, no traffic, just a mysterious failure.

Don’t worry. We’ve all been there. In this beginner-friendly guide, I’ll walk you through exactly how to debug a Kubernetes app crash step-by-step.

🧠 Step 1: Understand the Crash Types

Before jumping in, it helps to know what kind of crash you're dealing with:

  • CrashLoopBackOff: Your container is starting, failing, and retrying in a loop.
  • ImagePullBackOff: Kubernetes can’t find or pull your container image.
  • OOMKilled: Your app used too much memory and was killed.
  • Completed: Your container exited normally (common in jobs).

Each type tells a story. Let’s decode it.

🔍 Step 2: Check Pod Status 

kubectl get pods

Look at the STATUS column. If you see something like CrashLoopBackOff, grab the pod name and inspect it:

kubectl describe pod

This gives you events, restarts, reasons, and more. Look for lines like:

Last State:   Terminated
Reason:       OOMKilled
Exit Code:    137

📄 Step 3: Check Pod Logs

Logs are your best friends.

kubectl logs

If the pod has multiple containers:

kubectl logs  -c

Look for stack traces, uncaught exceptions, or config errors.

💡 Use --previous flag if the container has already restarted.

kubectl logs  --previous

🔧 Step 4: Exec Into the Pod (If It’s Still Running)

Sometimes it's easier to debug from inside the pod:

kubectl exec -it  -- /bin/bash

Check environment variables, file paths, dependencies:

env
ls /app
cat config.yaml

📦 Step 5: Validate Your Deployment

Check for errors in your manifest:

  • Wrong image name/tag?
  • Missing environment variables?
  • Wrong command/entrypoint?
  • CPU/memory limits too low? 
kubectl get deployment  -o yaml

🧯 Step 6: Check Events & Nodes

Maybe the issue isn't your app. Maybe it's the cluster.

kubectl get events --sort-by=.metadata.creationTimestamp

Also check node status:

kubectl get nodes

A node could be out of resources or in NotReady state.

🛠️ Step 7: Use Liveness & Readiness Probes

If you’re not already using them, add these to your deployment:

livenessProbe:
  httpGet:
    path: /health
    port: 8080
  initialDelaySeconds: 10
  periodSeconds: 5

readinessProbe:
  httpGet:
    path: /ready
    port: 8080
  initialDelaySeconds: 5
  periodSeconds: 3

This helps Kubernetes know when to restart your app or stop sending traffic.

✅ Final Thoughts

Debugging in Kubernetes takes practice. Use the built-in tools: kubectl logs, describe, exec, events.

Start with what Kubernetes is telling you, then dig deeper. And always keep your YAML clean, validated, and version-controlled.

If you found this helpful, share it with a fellow beginner and follow me on Dev.to for more DevOps guides. Happy debugging! 🛠️🐳☸️