CKA Troubleshooting Cheat Sheet

Golden Rule

1. kubectl get
2. kubectl describe
3. kubectl logs
4. kubectl exec
5. kubectl get events

This solves a large percentage of troubleshooting questions.

Pending Pod

Think:

Scheduler
Resources
Taints
PVC
Affinity
NodeSelector

Commands:

kubectl describe pod <pod>
kubectl get nodes
kubectl get events --sort-by='.lastTimestamp'

Look for:

• FailedScheduling
• Insufficient CPU
• Insufficient Memory
• Taints
• PVC issues

CrashLoopBackOff

Think:

Application problem
Container starts then crashes

Commands:

kubectl describe pod <pod>
kubectl logs <pod>
kubectl logs <pod> --previous

Look for:

• Application errors
• OOMKilled
• Misconfiguration
• Missing files

ImagePullBackOff

Think:

Image Name
Tag
Registry
imagePullSecrets

Commands:

kubectl describe pod <pod>

Look in Events:

• Failed to pull image
• ErrImagePull
• Authentication failed

Common Causes:

• Wrong image name
• Wrong tag
• Image doesn't exist
• Missing imagePullSecret
• Registry unavailable

Service Not Working

Think:

Endpoints
Selectors
Pod Labels

Commands:

kubectl get endpoints <svc>
kubectl describe svc <svc>
kubectl get pods --show-labels

Most Common Cause:

Selector does not match Pod labels

PVC Pending

Think:

PV exists?
StorageClass?
Capacity?
AccessModes?

Commands:

kubectl describe pvc <pvc>
kubectl get pv

Common Causes:

• No PV available
• StorageClass mismatch
• Capacity mismatch
• AccessModes mismatch

Taints and Tolerations

Error Example:

node(s) had taint {gpu=true:NoSchedule}

Think:

Taint → Toleration

Commands:

kubectl describe node <node>
kubectl get pod <pod> -o yaml

Toleration Example:

tolerations:
- key: "gpu"
  operator: "Equal"
  value: "true"
  effect: "NoSchedule"

Node Not Ready

Commands:

kubectl describe node <node>
kubectl get events

Check:

• MemoryPressure
• DiskPressure
• PIDPressure
• NetworkUnavailable
• Ready condition

Commands to Memorize

kubectl describe pod <pod>
kubectl logs <pod>
kubectl logs <pod> --previous

kubectl describe deploy <deploy>
kubectl get rs

kubectl describe pvc <pvc>
kubectl get pv

kubectl get endpoints <svc>
kubectl describe svc <svc>

kubectl describe node <node>

kubectl get all -n <namespace>
kubectl get pods --show-labels

kubectl run busybox \
  --image=busybox \
  --restart=Never -- sleep 3600

CKA Status → First Thought

Pending
→ Scheduler

CrashLoopBackOff
→ Logs

ImagePullBackOff
→ Image / Registry

Service Failing
→ Endpoints

PVC Pending
→ PV / StorageClass

Node NotReady
→ Node Conditions

FailedScheduling
→ Resources / Taints