The Certified Kubernetes Administrator (CKA) exam is one of the most respected, and most feared, credentials in cloud-native IT. It is fully hands-on, it runs against a live cluster, and the syllabus has quietly shifted over the last year. If you want to know how to pass the CKA exam in 2026, the honest answer is that you cannot cram it like a multiple-choice test. You have to be fast and fluent at the command line. This 12-week study plan is built to get you there, and it is aligned to the current Kubernetes v1.35 exam environment and the newer Gateway API content that older guides simply do not cover.
Roughly 814 open US roles list the CKA specifically, and certified administrators typically earn between 130,000 and 180,000 US dollars a year. That demand is exactly why the exam is worth the effort, and why a structured plan beats random YouTube binging. Let us break down what the exam actually is, what changed, and how to prepare week by week.
What the CKA Exam Actually Tests in 2026
The CKA is a performance-based exam. There are no multiple-choice questions. You are dropped into a real terminal with access to several Kubernetes clusters, and you have two hours to complete a set of practical tasks (usually around 17) using
kubectl and standard Linux tools. You are graded on the end state of the cluster, not on how you got there.
Here are the core facts you need to lock in before you start studying:
- Format: Live, hands-on, performance-based. No multiple choice.
- Duration: 2 hours.
- Tasks: Approximately 17 real-world problems.
- Passing score: 66%.
- Cost: 445 US dollars, which includes one free retake.
- Validity: 3 years from the date you pass.
- Kubernetes version: The exam environment tracks the current stable release and is updated to the latest minor version (v1.35) within a few weeks of release.
- Open book, sort of: You may keep one browser tab open on the official kubernetes.io/docs during the exam.
Exam Tip: The CKA is graded on results, so learn to work fast and verify your work. A task that "looks" done but has a typo in a label selector scores zero. Always confirm with
andkubectl getbefore moving on.kubectl describe
The Domains and Their Weights
The exam is split across five domains. Troubleshooting is the single largest, and it is also where most candidates lose marks under time pressure. Study in proportion to these weights.
| Domain | Weight | What it covers |
|---|---|---|
| Troubleshooting | 30% | Cluster and node failures, failing workloads, logging, monitoring, network debugging |
| Cluster Architecture, Installation and Configuration | 25% | RBAC, kubeadm cluster setup and upgrades, etcd backup and restore, Helm and Kustomize |
| Services and Networking | 20% | Pod and service connectivity, NetworkPolicies, Ingress, the Gateway API, CoreDNS |
| Workloads and Scheduling | 15% | Deployments, rolling updates, ConfigMaps and Secrets, scaling, scheduling, self-healing |
| Storage | 10% | Storage classes, persistent volumes, persistent volume claims, volume modes |
Troubleshooting and Cluster Architecture together make up 55% of the exam. If you can debug a broken cluster and rebuild control-plane components under pressure, you are already most of the way to a pass.
What Changed: v1.35, the Gateway API and Native Sidecars
If you are studying from a guide written before 2025, you are studying the wrong exam. The scope was revised and now leans harder into modern, real-world administration.
The biggest single change is the arrival of the Gateway API on the exam. The Gateway API is the more flexible, role-oriented successor to Ingress, and you are now expected to understand Gateway resources, GatewayClasses and routing rules, not just classic Ingress objects. Older prep material that stops at Ingress will leave you exposed.
Other notable shifts in the current syllabus include:
- Native sidecars are now generally available and can appear in workload tasks.
- Greater emphasis on Helm and Kustomize for managing manifests, rather than hand-writing every YAML file.
- A heavier focus on troubleshooting as the dominant domain, reflecting how the job is actually done.
- The exam environment now runs Kubernetes v1.35, so practise with a current cluster, not an old one.
Exam Tip: Do not rely on 2023 or 2024 CKA guides. Confirm that anything you study covers the Gateway API and native sidecars. If it does not mention them, it predates the current syllabus.
Before You Start: Prerequisites and Setup
The CKA assumes you already have real Kubernetes exposure. The Cloud Native Computing Foundation recommends 6 to 12 months of hands-on experience before you sit the exam. You do not strictly need it, but if you are brand new to containers, add a few weeks of Docker and Linux fundamentals to the front of this plan.
Set up your practice environment in week one and never let it go cold:
- Build a multi-node cluster with
, or usekubeadm
orkind
for quick iteration.minikube - Configure
aliases and shell autocompletion. Speed at the keyboard is a scored skill.kubectl - Bookmark the exact kubernetes.io/docs pages you will lean on during the exam.
- Get comfortable editing YAML in
inside a terminal, because that is the only environment you get.vim
The 12-Week CKA Study Plan
This plan assumes around 8 to 10 hours of study per week, weighted heavily toward hands-on practice. The golden rule of CKA prep is simple: roughly 80% of your time should be spent typing commands in a live cluster, not watching videos.
Weeks 1 to 2: Foundations and Cluster Architecture
Stand up your own cluster with
kubeadm from scratch. Learn the control-plane components (API server, scheduler, controller manager, etcd) and what each one does when it breaks. Practise kubectl basics until they are muscle memory: create, get, describe, edit, delete, and label.
- Build and reset a cluster at least three times.
- Practise an etcd backup and restore end to end.
- Set up RBAC roles, role bindings and service accounts.
Weeks 3 to 4: Workloads and Scheduling
Move into Deployments, ReplicaSets, DaemonSets and StatefulSets. Learn rolling updates and rollbacks cold. Get fluent with ConfigMaps, Secrets, resource requests and limits, node selectors, taints, tolerations and affinity.
- Perform a rolling update and a rollback under a five-minute timer.
- Schedule a pod to a specific node three different ways.
- Practise creating and consuming ConfigMaps and Secrets from memory.
Weeks 5 to 6: Services and Networking (including the Gateway API)
This is where the 2026 syllabus bites. Master ClusterIP, NodePort and LoadBalancer services, then CoreDNS and pod-to-pod connectivity. Spend real time on NetworkPolicies, classic Ingress and, crucially, the Gateway API.
- Write NetworkPolicies that allow and deny traffic, then verify them.
- Deploy an Ingress controller and expose an app.
- Create a Gateway, a GatewayClass and an HTTPRoute, and test the routing.
Exam Tip: Networking questions are worth 20% and are among the most time-consuming. If a task is not resolving, check the service selector, the endpoint list and DNS before you assume the NetworkPolicy is wrong.
Weeks 7 to 8: Storage and Configuration Management
Cover storage classes, persistent volumes, persistent volume claims, access modes and reclaim policies. Then add Helm and Kustomize, which now feature more heavily in configuration tasks.
- Provision a persistent volume and bind it to a claim used by a pod.
- Install and upgrade an application with Helm.
- Patch a base manifest with a Kustomize overlay.
Weeks 9 to 10: Troubleshooting (the 30% domain)
Now attack the largest and highest-value domain. Deliberately break things and fix them: kill the kubelet, corrupt a manifest, block DNS, exhaust node resources. Learn to read logs and events quickly with
kubectl logs, kubectl describe, kubectl get events and journalctl.
- Diagnose a NotReady node and bring it back.
- Fix a CrashLoopBackOff pod from first principles.
- Debug a service that returns no endpoints.
Weeks 11 to 12: Mock Exams and Speed
The final stretch is about time management. The CKA is not just about knowing the answer, it is about doing it fast enough. Sit full timed mock exams and review every miss.
- Complete at least two full timed practice exams.
- Use both of your included Killer.sh simulator sessions. The simulator is deliberately harder than the real exam, so if you can pass it, you are ready.
- Drill your
aliases and imperative command shortcuts to shave seconds off every task.kubectl
Exam Tip: Book the real exam only once you are consistently scoring 65% or higher on the Killer.sh simulator. Use your second simulator attempt three to four days before exam day as your final rehearsal.
Study Resources and Practice Strategy
You do not need to buy ten courses. You need one solid foundation and a lot of reps. Combine a structured course with relentless hands-on labs and timed question practice.
- Official docs: kubernetes.io/docs is your allowed reference in the exam, so learn to navigate it fast.
- Hands-on labs: Build, break and rebuild clusters constantly. This is non-negotiable.
- Timed practice questions: The best way to build exam-day speed is to answer realistic, scenario-based questions against the clock. You can drill practical CKA-style scenarios and track your weak domains with CertCrush practice tools.
- The included simulator: Your exam fee includes two Killer.sh sessions. Do not waste them early. Save them for weeks 11 and 12.
Common Reasons Candidates Fail the CKA
Most CKA failures are not about knowledge gaps. They come down to a handful of avoidable mistakes:
- Running out of time. Slow typing and hunting through docs kills scores. Aliases and imperative commands fix this.
- Studying theory over practice. Watching videos feels productive but does not build the reflexes the exam demands.
- Ignoring the Gateway API. Candidates using old guides get blindsided by modern networking tasks.
- Skipping verification. Not checking your work with
andkubectl get
leaves silent errors that score zero.describe - Neglecting troubleshooting. It is 30% of the exam, yet it is the domain people practise least.
For a broader look at why this pattern repeats across certifications, see our guide on why most people fail certification exams.
Is the CKA Worth It in 2026?
Short answer: yes, for the right person. Kubernetes is the default way large organisations run containers, and skilled administrators are scarce. The certification is a strong signal to employers, it gets you past HR filters, and the preparation itself makes you genuinely better at the job. With salaries commonly in the 130,000 to 180,000 US dollar range and hundreds of roles asking for it by name, the return on a 445 dollar exam is easy to justify.
If you are still mapping out your broader path, our roundup of the best IT certifications for 2026 puts the CKA in context alongside cloud and security credentials. And if you are coming from an AWS background, pairing Kubernetes skills with a cloud cert like the AWS Solutions Architect Associate is one of the strongest combinations on the market.
Ready to Start Practising?
Passing the CKA in 2026 comes down to one thing: hands-on repetition against a current, v1.35 syllabus. A 12-week plan gives you the structure, but you have to put in the reps and practise under real time pressure.
Build your speed and pinpoint your weak domains with realistic, scenario-based practice. Create your free CertCrush account to start drilling exam-style questions, track your progress across every CKA domain, and walk into exam day fast, confident and ready to pass on the first attempt.