Passing the Certified Kubernetes Administrator exam comes down to allocating study time where the scoring weight sits: troubleshooting first, cluster administration second, everything else in decreasing order. The Certified Kubernetes Administrator exam costs $445, runs against a Kubernetes v1.35 environment, and rewards candidates who prepare by domain weight instead of memorizing trivia. This CKA prep plan maps every study block to the official curriculum so your hours go where the points actually are. For exam-day tactics that this plan deliberately does not repeat, our guide to passing the CKA on v1.35 is the companion read.
How the CKA exam works
The CKA is an online, proctored, performance-based test delivered through PSI’s Bridge platform. There are no multiple-choice questions: you solve 15 to 20 hands-on tasks from a command line against a live cluster, and results are typically emailed within 24 hours. A score of 66% or above must be earned to pass, and there are no multiple-choice questions anywhere on the exam. Registration on the Linux Foundation catalog includes two exam attempts plus 12 months of eligibility, which changes how you should think about risk: a first attempt that falls short is not a sunk cost.
Because every task is graded on what you execute, prep that only builds reading familiarity fails on exam day. The exam rewards navigation speed, kubectl fluency under pressure, and the judgment to triage a broken cluster. Build your study routine around performing repairs, not reviewing slides.
Where the points live
The official CNCF curriculum for v1.35 defines five weighted domains, and the distribution is uneven on purpose. Troubleshooting carries 30% of the exam, Cluster Architecture, Installation and Configuration carries 25%, Services and Networking carries 20%, Workloads and Scheduling carries 15%, and Storage carries 10%. Two domains together account for more than half of your score, which is why the study plan below front-loads them.
| Domain | Weight | Priority practice |
|---|---|---|
| Troubleshooting | 30% | Broken nodes, failing pods, service outages, log analysis |
| Cluster Architecture, Installation & Configuration | 25% | kubeadm, RBAC, etcd backups, upgrades, HA control planes |
| Services & Networking | 20% | NetworkPolicies, ingress, CoreDNS, service types |
| Workloads & Scheduling | 15% | Deployments, rolling updates, scheduling constraints |
| Storage | 10% | Persistent volumes, claims, storage classes |
Treat the table as a budget. If you have ten practice hours in a week, roughly three belong to troubleshooting drills, two and a half to cluster installation and configuration work, and the remainder split across the smaller domains. Candidates who invert this order and spend weeks on storage or workload theory consistently underperform relative to their effort.
A twelve-week study plan
The sequence below assumes you already run Linux comfortably and have deployed at least one cluster. Adjust the pace, not the order.
- Weeks 1-2, foundation: rebuild a cluster from scratch with kubeadm twice, without notes the second time. Practice etcd backup and restore until it is reflexive.
- Weeks 3-5, troubleshooting: break clusters deliberately. Kill a kubelet, corrupt a certificate, mismatch a service selector, then time yourself diagnosing each failure.
- Weeks 6-7, RBAC and upgrades: create roles and bindings to spec, then perform a full kubeadm upgrade on a three-node cluster.
- Weeks 8-9, networking: implement NetworkPolicies that satisfy written requirements, expose workloads through services and ingress, and debug CoreDNS resolution failures.
- Weeks 10-11, workloads and storage: rehearse rolling updates, rollbacks, scheduling constraints, and persistent volume claim workflows.
- Week 12, simulation: run a full timed Killer.sh session, review the graded report, and patch only your two weakest domains.
Use the simulator early
Too many candidates save the official simulator for the final days and discover format problems they can no longer fix. Registration includes two attempts at the Killer.sh exam simulator, and each attempt grants 36 hours of access starting from activation. The window is long enough to review every wrong answer calmly, so activate the first attempt around week 5 rather than week 12.
Before each simulator run, verify this checklist:
- You can context-switch between namespaces without retyping the full command
- You have an alias file ready and know it is permitted
- You can find any Kubernetes object definition in the allowed docs in under a minute
- You can SSH into a node and restart a failed component from memory
Validity and retake rules
The credential is valid for 2 years from the date you pass, and the renewal path is simply passing the exam again before that window closes. A failed first attempt does not end your registration: the bundled retake must be completed within 12 months of the original purchase, and candidates may retake as soon as they wish. If you are stacking credentials, weigh this against adjacent options like the comparison in our piece on Security+ versus Pentest+ ordering, since expiration windows compound when you hold multiple certifications.
Sources
- Linux Foundation, Certified Kubernetes Administrator (CKA) exam page
- Linux Foundation, CKA/CKAD/CKS certification FAQ
- EveryExamPrep, CKA exam overview and pricing