To pass the Certified Kubernetes Administrator (CKA) exam first time, you need two things: hands-on fluency across the five curriculum domains, and the speed to finish 15–20 live tasks in two hours. Most people who fail know enough Kubernetes. They run out of time or lose marks on small slips, such as working on the wrong host.
This guide covers the current exam format, where the marks are, speed techniques, a four-week plan and what to do on the day.
The exam at a glance (October 2026)
| Item | Detail |
|---|---|
| Format | Online, proctored, performance-based tasks on live clusters |
| Length | 2 hours, 15–20 tasks |
| Kubernetes version | v1.35 (the Linux Foundation updates the environment roughly 4–8 weeks after each Kubernetes minor release) |
| Price | US$445 for the exam; bundles with a course or subscription cost more |
| Included | One free retake and two simulator sessions, each with 36 hours of access |
| Window | 12 months from purchase to take the exam |
| Validity | 2 years |
| Prerequisites | None |
At an exchange rate anywhere between ₹85 and ₹95 to the dollar, US$445 works out to about ₹38,000–42,000, plus any foreign-currency charges on your card. That is a meaningful sum, so it pays to make the first attempt count. The Linux Foundation regularly runs discount sales, which can cut the price noticeably if your timing is flexible.
Where the marks are
| Domain | Weight | What to practise |
|---|---|---|
| Troubleshooting | 30% | Broken nodes and control-plane components, failing pods, services with no endpoints, reading logs and events, resource usage with kubectl top |
| Cluster Architecture, Installation & Configuration | 25% | kubeadm install and upgrade, RBAC, highly available control plane, Helm and Kustomize, CRDs and operators, CNI/CSI/CRI basics |
| Services & Networking | 20% | Service types, network policies, Ingress, the Gateway API, CoreDNS |
| Workloads & Scheduling | 15% | Deployments, rolling updates and rollbacks, ConfigMaps and Secrets, autoscaling, node affinity, limits |
| Storage | 10% | Storage classes, dynamic provisioning, access modes, reclaim policies, PVs and PVCs |
Two points are easy to miss. First, troubleshooting alone is almost a third of the score, so it deserves a third of your practice time. Second, the curriculum changed on 18 February 2025 to add Helm, Kustomize, the Gateway API, CRDs and operators. Older courses and blog posts may not cover these, so check any material against the official competency list. etcd backup and restore is no longer named explicitly, but it falls under cluster lifecycle management and is still worth practising.
Speed techniques
- Generate YAML instead of typing it. Use
kubectl create deployment,kubectl runandkubectl exposewith--dry-run=client -o yaml > file.yaml, then edit only what the task needs. - Use the built-in alias.
kis already aliased to kubectl with Bash completion. Add short resource names:po,deploy,svc,ns,pv,pvc,netpol. - Use
kubectl explain(for examplek explain pod.spec.affinity --recursive) to check field names without leaving the terminal. - Know where the documentation examples live. You can open the official Kubernetes documentation during the exam. Bookmark mentally the pages for network policies, persistent volumes, RBAC, kubeadm upgrades and the Gateway API.
- Practise your editor. The Insert key is disabled in the exam terminal, so get comfortable with vi modes. Copy and paste use Ctrl+Shift+C and Ctrl+Shift+V.
How the exam environment works now
Older guides tell you to switch kubectl context before each task. The current environment works differently: you start on a base machine, and each task names a host. You connect with ssh <host>, do the work, then type exit to return. Nested SSH is not supported, and the base machine does not have kubectl installed. Use sudo -i when a task needs root.
The most common avoidable error is doing a perfect task on the wrong host. Make it a habit: read the host name, SSH in, confirm the prompt, then start.
A four-week plan for people already using Kubernetes
| Week | Focus | Goal by the end of the week |
|---|---|---|
| 1 | Build a two-node cluster with kubeadm on local VMs; workloads, scheduling, ConfigMaps, Secrets | Create and roll back a deployment from memory in under five minutes |
| 2 | Services, network policies, Ingress, Gateway API, CoreDNS; storage classes and PVCs | Write a working network policy without copying from documentation |
| 3 | kubeadm upgrade, etcd snapshot and restore, RBAC, Helm and Kustomize; daily troubleshooting drills | Fix a deliberately broken kubelet or static pod manifest in under ten minutes |
| 4 | Simulator session 1, review every wrong answer, timed drills, simulator session 2, then book the exam | Finish a simulator session with time to spare |
If you are new to Kubernetes, add four to six weeks of fundamentals first: pods, deployments, services and basic Linux administration. The simulator is deliberately harder than the real exam, so a low first score is normal. Use it to find gaps, not to judge readiness. For a general approach to studying around a full-time job, see our guide on preparing for any IT certification while working.
On exam day
- Check your room, ID and system requirements well before the slot; proctoring problems eat into your nerves even if they do not eat into your time.
- Skim all tasks first. Each one shows its weight, so do quick, high-weight tasks early.
- Flag anything that takes more than about eight minutes and move on.
- Verify each answer: the pod is Running, the service has endpoints, the PVC is Bound.
- Partial credit is possible, so finish as much of a hard task as you can rather than abandoning it.
Results arrive by email within about 24 hours. If you fail, the free retake is included, and the score report tells you which domains were weak.
Common mistakes that cost marks
- Forgetting the namespace. If a task says a resource lives in a given namespace, put
-non every command or set it in the YAML. - Editing a running pod when the task really wants the deployment changed, so the fix disappears on the next rollout.
- Leaving a broken node half-fixed: always restart the kubelet with
systemctl restart kubeletand checksystemctl status kubeletafter editing its configuration. - Writing output files to the wrong path. When a task asks you to save something to a file, copy the exact path from the question.
- Relying on memorised exam dumps. The tasks change, and the certification agreement prohibits sharing exam content.
FAQ
Should I take CKA, CKAD or CKS first?
CKA suits administrators and platform engineers; CKAD suits developers. CKS requires an active CKA, so it always comes after. Our CKA vs CKAD vs CKS comparison explains the difference.
What happens after two years?
The certification expires and must be renewed by passing the current exam again. Our renewal rules guide compares this with other vendors.
Last reviewed: 8 October 2026.
Sources
- Linux Foundation – Certified Kubernetes Administrator (CKA)
- Linux Foundation – Important instructions: CKA and CKAD
- Linux Foundation – CKA program changes (February 2025)
- CNCF – curriculum repository
- Kubernetes documentation

Leave a Reply