← All Articles ·

Getting Started with CKA: A Practical Study Guide

Everything you need to know to prepare for the Certified Kubernetes Administrator exam — from someone who has helped hundreds of engineers pass it.

PT
Puru Tuladhar
kubernetesckacertificationdevops

The Certified Kubernetes Administrator (CKA) exam is one of the most valuable certifications in the cloud-native ecosystem. Unlike multiple-choice exams, the CKA is a performance-based test where you solve real problems in a live Kubernetes cluster under time pressure. That makes it both challenging and genuinely meaningful.

After helping hundreds of engineers across Nepal’s top organizations prepare for and pass the CKA, here’s my distilled guide.

What the CKA Actually Tests

The exam covers five core domains:

  • Cluster Architecture, Installation & Configuration (~25%) — bootstrapping clusters with kubeadm, RBAC, certificates
  • Workloads & Scheduling (~15%) — Deployments, DaemonSets, resource limits, taints, node affinity
  • Services & Networking (~20%) — Services, Ingress, NetworkPolicies, CoreDNS
  • Storage (~10%) — PersistentVolumes, StorageClasses, volume modes
  • Troubleshooting (~30%) — This is the biggest domain and the one most candidates underestimate

The troubleshooting weight is the key insight. You can memorize every manifest spec perfectly and still fail if you can’t debug a broken cluster quickly.

Your Study Environment

Before touching any curriculum, set up your practice environment. Options:

# Option 1: Local with kind
kind create cluster --name cka-prep

# Option 2: Local with minikube (multi-node simulation)
minikube start --nodes 3 --driver=docker

# Option 3: killer.sh (the official CKA simulator — highly recommended)
# Included with your exam registration

Killer.sh deserves special mention. The two included practice sessions are harder than the real exam, which is exactly what you want. Do each session twice.

The Kubectl Muscle Memory Trap

Most candidates spend too much time on YAML and not enough on kubectl. The exam is open-book — you have access to kubernetes.io/docs — but you don’t have time to look up every flag.

These commands should be instinctive:

# Generate YAML without applying
kubectl run nginx --image=nginx --dry-run=client -o yaml > pod.yaml

# Scale and rollout
kubectl scale deploy my-app --replicas=5
kubectl rollout undo deploy/my-app
kubectl rollout status deploy/my-app

# Debug fast
kubectl describe pod <name> | tail -20
kubectl logs <pod> --previous
kubectl exec -it <pod> -- /bin/sh

# Node management
kubectl cordon node01
kubectl drain node01 --ignore-daemonsets --delete-emptydir-data
kubectl uncordon node01

Set this alias at the start of every exam session:

alias k=kubectl
export do="--dry-run=client -o yaml"

Troubleshooting Methodology

When a cluster component is broken, work top-down:

  1. Check pod status — k get pods -A to spot anything not Running/Completed
  2. Check node status — k get nodes for NotReady nodes
  3. Check system pods — kube-system namespace is almost always involved
  4. Check kubelet — systemctl status kubelet on the affected node
  5. Check static pod manifests — /etc/kubernetes/manifests/ for control plane issues
  6. Check certificates — openssl x509 -in /path/to/cert -text -noout

The exam frequently breaks clusters by tampering with static manifests or kubeconfig files. Know these paths cold.

Time Management on Exam Day

You have 2 hours for ~17 questions, weighted by difficulty (shown as % per question). Recommended approach:

  1. First pass — answer everything you can complete in under 4 minutes
  2. Flag and skip — don’t let one hard question sink 3 easy ones
  3. Second pass — tackle the heavy-weight questions you skipped
  4. Don’t perfect — a working solution beats a polished incomplete one

The One Thing Most People Get Wrong

Candidates practice in single-node clusters then get surprised by multi-context exam environments. The CKA gives you multiple clusters and you must switch between them using kubectl config use-context at the start of each question.

Miss the context switch = wrong cluster = all work wasted.

Read every question fully before typing a single command.

The CKA is achievable for any engineer willing to put in deliberate practice. 6–8 weeks of consistent, hands-on lab work is typically enough. If you want a structured path, our Kubernetes Training program takes corporate teams through exactly this curriculum with dedicated lab time.

Good luck — and may your pods always reach Running state.