Certified Kubernetes Administrator (CKA)
The CKA puts you on the command line of several real Kubernetes clusters and gives you tasks. There are no options to choose between and nothing to recognise.
Two hours, fifteen to twenty tasks, and the official documentation open beside you throughout.
Exam facts
- Exam code
- CKA
- Level
- Associate
- Field
- Cloud
- Duration
- 120 minutes
- Questions
- 15 to 20 hands-on tasks on the command line of several Kubernetes clusters (performance-based, no multiple choice)
- Passing score
- 66 %
- Languages
- English, Japanese, Chinese
- Price
- 445 USD (read 12 September 2026)
- Validity
- 2 years, renewable
- Exam delivery
- Online proctored through PSI; booked with the Linux Foundation (Training & Certification)
Prerequisites: None. Expected are a confident Linux command line, kubectl, YAML and container basics; the Linux Foundation rates the exam as intermediate.
Renewal: Pass the exam again before it expires; there is no renewal through continuing-education credits
Source: exam page at CNCF · Price: Linux Foundation list price including one retake and two sessions in the exam simulator; exam eligibility runs 12 months from purchase; bundles with courses cost more
How to prepare
1. The vendor's free learning path
CNCF publishes the exam objectives and a learning path free of charge — the authoritative source for scope and weighting. Open the learning path (opens in a new tab)
2. Online courses
Courses and practice questions from the directory that target exactly this exam — details, price and the provider link (affiliate) are on the course page.
Book the exam
Delivered by Online proctored through PSI; booked with the Linux Foundation (Training & Certification). Schedule with CNCF (opens in a new tab)
Affiliate disclosure: the course links above is affiliate links — buying through them may earn us a commission at no extra cost to you. The vendor's learning path and booking links carry no commission.
Open documentation changes what is being measured
- Nothing lookup-able is tested, because you can look it up. Memorising flags is wasted preparation.
- Speed is the constraint. Candidates who search for everything run out of time regardless of knowledge.
- Troubleshooting carries the largest share of the curriculum, which matches what operating a cluster actually involves.
- You work across several clusters and switching context correctly is itself a way to lose marks.
- Partial credit exists per task, so an imperfect attempt beats an abandoned one.
- Learn to navigate the documentation fast, because that is a tested skill here rather than a crutch.
- Practise on a real cluster, not slides. The simulator sessions included with the fee are the closest thing to the exam conditions.
- Get fluent with the command-line basics that save seconds: context switching, output formats, generating manifests rather than typing them.
- Rehearse under the clock, since the failure mode is almost never knowledge and almost always pace.
- Leave nothing blank, because per-task marking rewards the attempt.
The curriculum by weight
| Area | Share | What the tasks tend to be |
|---|---|---|
| Troubleshooting | 30 % | A node, a pod or a network path is broken and you find out why |
| Cluster architecture, installation and configuration | 25 % | Control plane, upgrades, backups, access |
| Services and networking | 20 % | Getting traffic to the right place and keeping the wrong traffic out |
| Workloads and scheduling | 15 % | Deployments, scaling, placement, configuration |
| Storage | 10 % | Volumes and claims, and what happens to the data |
Troubleshooting at nearly a third is the tell. This is an operator's exam, and the operator's day is spent on things that were working yesterday. The installation and configuration share is the other place candidates from a developer background lose time, because they have used clusters somebody else built and have never had to upgrade a control plane or restore a backup with a clock running. The exam is pinned to Kubernetes 1.35, delivered online by PSI with a proctor watching, and passed at 66 percent of the available marks.
What the fee buys
The list price includes one retake, twelve months in which to schedule, and two sessions in the exam simulator. That is a different product from a bare exam voucher: the simulator is the closest available rehearsal of the real conditions, and the retake removes most of the financial fear from a first attempt. Bundles with courses cost more and are not required.
Two years, then again
The certificate lasts two years and there is no credit-based renewal: you sit the exam again against whatever version is current. The Linux Foundation treats currency as the point, which is defensible for a platform that moves as fast as this one and expensive for the holder. Decide at the outset whether this is a credential to maintain, in which case budget for a sitting every second year, or one to spend on a job move and then let lapse with a clear conscience. The tasks are presented in English, Japanese or Simplified Chinese, and the clusters themselves speak the same language everywhere.
What a question looks like
Written by us in the exam's style. It is not a real question from any question bank, and we do not publish those.
A pod on a specified cluster is stuck and not running. Diagnose why, fix the cause, and leave the workload running as originally intended, without changing any other resource in that namespace.
There is no wording to parse and no options to eliminate. The difficulty is that the cause could be scheduling, storage, configuration or a node condition, and each is investigated differently. The exam measures whether you have a diagnostic habit rather than a memorised answer, and the final clause quietly forbids the shortcut of rebuilding everything.
What it costs to get and to keep
| Item | Amount | Note |
|---|---|---|
| Exam fee | 445 USD | includes one retake, twelve months of eligibility to sit it, and two sessions in the practice simulator, which is unusually generous and worth reading as part of the price (read 12 September 2026) |
| Retake | not published | no separate charge for the first one; it is bundled into the fee above (read 12 September 2026) |
| Renewal every two years | not published | sit the exam again. There is no continuing-education route of any kind, which makes this the shortest validity with the least flexible renewal on the site (read 12 September 2026) |
How much preparation, from where you are
- You operate clusters at work
- The gap is usually pace rather than knowledge, plus whichever parts of the curriculum your platform team handles for you. Rehearse against the clock rather than reading more.
- You deploy onto clusters somebody else runs
- A bigger jump than it feels. Workloads will be familiar; cluster architecture, installation and the troubleshooting domain assume an administrator's view you have never needed.
- You administer Linux and are moving toward containers
- Strong footing. The diagnostic instinct transfers directly, and what you are learning is where Kubernetes hides the information you already know how to interpret.
What passing this does not prove
- Whether you can design a platform, as opposed to keeping one running.
- Anything about cost, which is where most Kubernetes projects actually hurt.
- The developer-facing half of the ecosystem, which has its own certificate.
- Whether you would have chosen Kubernetes in the first place.
Against the alternatives
- AZ-400 — Microsoft Certified: DevOps Engineer Expert
- Microsoft's delivery exam is written rather than performed, and answers a different question: one evidences that you can set out a delivery process, the other that you can keep a cluster alive against the clock.
- SAA-C03 — AWS Certified Solutions Architect – Associate
- Broader and better known to recruiters, and it certifies design judgement rather than hands-on operation. Many people end up holding both, and they signal different things.
- EX200 — Red Hat Certified System Administrator (RHCSA)
- The other performance-based exam covered here, on Linux rather than Kubernetes. If you like this format, that one is the natural companion and the foundation underneath it.
CKA — quick answers
Is open documentation not cheating?
It is the opposite: it removes the part of the exam that could be gamed. Because you can look anything up, nothing that can be looked up is worth testing, so what remains is whether you can diagnose a problem and act on it quickly. That is a much better model of the job than recall.
What does the fee actually include?
One retake, twelve months of eligibility to schedule, and two sessions in a practice simulator. That bundle is unusual and materially changes the comparison with cheaper exams, particularly since the simulator is the closest available rehearsal of the real conditions.
Why only two years?
Kubernetes moves quickly and the exam is pinned to a specific version, so the vendor treats currency as the point. The consequence is stricter than most: there is no credit-based renewal, so keeping the certificate means sitting it again. Settle at the outset whether this is a credential to maintain or one to spend.
Which should I take first if I want the developer exam too?
This one, in most cases. The administrator curriculum covers the cluster context that makes the developer-facing material easier to reason about, and the reverse is less true. Both are performance-based, so the format practice carries over either way.
How does 66 percent work when tasks carry partial credit?
Each task is worth a stated number of points and is marked on what you achieved, so a half-finished task earns some of them. The pass mark applies to the total. That is why the advice to leave nothing blank is not motivational: an attempt that gets the object created but misconfigured can still score, and an abandoned task never does.
What comes next
- Microsoft · AZ-400Microsoft Certified: DevOps Engineer Expert
- AWS · SAA-C03AWS Certified Solutions Architect – Associate
- Google Cloud · ACEGoogle Cloud Associate Cloud Engineer
- CompTIA · XK0-006CompTIA Linux+
- Cisco · 200-901Cisco CCNA Automation (formerly DevNet Associate)
- LPI · 201-450 + 202-450LPIC-2 — Linux Engineer
- Red Hat · EX200Red Hat Certified System Administrator (RHCSA)
Concepts this exam draws on
Glossary entries with the reason each one matters for CKA.
- Containers
The CKA is container orchestration end to end; pods, images and runtimes are assumed.
- Cloud Hosting
Kubernetes clusters live on cloud infrastructure; storage and networking questions assume it.
- Load Balancing
Services, ingress and cluster networking are a quarter of the CKA.
- Kubernetes
The exam is two hours of live tasks on the system this concept describes.
Last reviewed 12 September 2026 · Getting Digital
