Real clusters. Real tasks. Real preparation.

A browser-based lab that gives every student their own real Kubernetes cluster, 17 hands-on tasks scored by programmatic validators, and an attempt history that shows exactly what you missed — and why.

free 3-question trial · €29 full access · 14 days · or €9.99 per attempt
17
graded tasks
1
real cluster
15 min
free trial

Three things, done well.

No video courses. No fake simulators. Use quick quizzes to warm up, then work the same kind of cluster, the same kind of task, and the same kind of grader you'll meet on exam day.

f.01 · runtime

Real Kubernetes, on demand.

Each attempt boots an isolated kind cluster wired through ttyd. SSH targets are real, kubectl is real, and the failure modes are real.

kind · ttyd · k8s 1.31 · 2 nodes
f.02 · scoring

Programmatic validators.

Every task ships its own check script. You get partial credit for partial work, an output trace per question, and a reviewable diff between your state and the target.

partial credit · per-task · idempotent
f.03 · history

Per-user attempt log.

Every finished exam — score, duration, missed tasks, the validator output, and the canonical solution — is saved to your console. Pattern-spot what you keep getting wrong.

persistent · solutions · diffable

From sign-up to scored, in four moves.

No installs. No local kubeconfig surgery. Open the lab, get a cluster, finish the tasks, read your report.

register

Create an account.

Email + password, or Google SSO. You land in your student console — profile, target score, sync clock.

provision

Open the lab.

One click spins a 2-node kind cluster and an embedded desktop. Provisioning runs once across all 17 tasks.

solve

Work in-context.

Each question gives you the SSH target, a docs link, and a question rail beside the live workspace. Submit when satisfied.

review

Read the report.

Programmatic validators run on finish. Pass/fail, per-task feedback, and the canonical solution post to your history.

A question, an instance, a grader. Side by side.

The lab is one window: a question rail on the left, an embedded Linux desktop on the right, and a progress strip you can navigate at any time. Below is what one task looks like.

› question 07 / 17 · node maintenance

Drain a node and reschedule its workload.

ssh root@ek8s-master

Set the node ek8s-node-1 as unavailable and reschedule all the pods running on it.

Use kubectl drain with --ignore-daemonsets. Daemonset-managed pods may remain on the node.

The control plane is reachable from ek8s-master. You'll be graded on whether the node is cordoned and all evictable pods are gone.

Mapped to the published CKA curriculum.

Domain weights mirror the official exam blueprint, with hands-on tasks covering the objectives you need to practice. See the full task list with descriptions.

d.01
Cluster Architecture & Installationkubeadm, etcd backup, control-plane upgrade
25%
d.02
Workloads & Schedulingdeployments, configmaps, scaling, scheduling
15%
d.03
Services & Networkingservices, ingress, DNS, network policies, CNI
20%
d.04
StoragePV, PVC, storage classes, volume modes
10%
d.05
Troubleshootingnode failures, control-plane logs, kubelet, networking
30%
d.06
Practice metaattempt history, flagged review, solution diffs

Pay once. No subscription.

Create a free account and try the first 3 questions for 15 minutes. Go full access for 14 days, or buy a single attempt — no recurring charges either way.

full access
cka practice lab
$ 29 one-time · 14 days
14 days of full lab access · no renewal · no auto-charge
  • Free account includes 3 questions · 15 minutes
  • Paid access unlocks 17 graded tasks · 120 minutes · unlimited retakes
  • Real on-demand kind cluster, every attempt
  • Programmatic, partial-credit validators
  • Persistent attempt log + canonical solutions
  • Unlimited retakes for 14 days from purchase
  • 7-day refund, no questions asked
Get access · €29 → Try for free stripe · visa · mastercard · amex
single attempt · pay as you go
€9.99 per attempt

One full 120-minute exam attempt with all 17 tasks, real cluster, and programmatic scoring. Buy more only when you need them.

Buy 1 attempt →
cohort · 5+ seats
Team licensing

Bulk seats, an admin console, aggregate progress reporting. Email [email protected] for a quote.

why so cheap
Built by people who sat the exam.

€29 covers the runtime, validators, and storage. We're not trying to sell you a course or a membership — just the practice surface we wanted before our own exam.

the certification itself (Linux Foundation) is sold separately at $445.
this lab is independent practice — not the exam.

Things people ask before signing up.

What does €29 actually buy me?

14 days of full lab access from the moment of purchase: all 17 graded tasks, unlimited retakes during the window, and the canonical solutions. Your attempt history stays on your account permanently — you just can't open new lab sessions after day 14. No recurring charge, no plan to upgrade. If it doesn't work for you, email [email protected] within 7 days for a full refund.

Is this affiliated with the Linux Foundation or CNCF?

No. CKA Practice Lab is an independent practice environment. We track the published curriculum closely, but we are not the certifier.

Do I need to install anything?

No. Everything runs in your browser — the cluster, the desktop, the editor, the validators. You'll need a modern browser and a stable connection.

What happens if I close the tab mid-exam?

Your runtime is released and the attempt is abandoned, but anything you'd already submitted (per-question) is preserved in your history. Open the lab again to start a fresh attempt.

How accurate are the validators?

Each task ships with a check script that inspects cluster state — not a string match on your commands. There are usually multiple valid solutions; the grader rewards any of them.

Can my team use this?

Yes. Admins can provision accounts and watch aggregate progress from the admin console. Talk to us about per-cohort licensing.

How does this compare to killer.sh?

Killer.sh gives you 2 sessions bundled with the exam purchase. This lab gives you unlimited retakes for 14 days (or per-attempt pricing), a persistent attempt history with solutions, and a free 3-question trial before you pay. Both use real cluster environments — we use kind clusters with the same kubectl, SSH, and failure modes you'll see on exam day.

How long should I study for the CKA?

Most people spend 4–8 weeks of focused study. If you already work with Kubernetes daily, 2–3 weeks of hands-on practice may be enough. The key is practice under timed conditions — reading docs alone won't build the muscle memory the exam requires. We recommend doing at least 3–4 full mock attempts before sitting the real exam.

Is the CKA exam hard?

The CKA has a first-attempt pass rate around 50%. It's entirely performance-based — no multiple choice. You get 2 hours to solve 15–20 tasks on real clusters via a remote desktop. The difficulty comes from time pressure, context switching between clusters, and troubleshooting tasks you haven't practiced. This lab is built specifically to close that gap.

Does this cover the 2025 CKA curriculum update?

Yes. Our tasks cover the post-February 2025 curriculum including Gateway API, Helm, and updated troubleshooting scenarios. See the full task list mapped to the CKA domains.

Should I take CKA or CKAD first?

CKA if you manage clusters (ops, platform engineering, SRE). CKAD if you only deploy workloads on clusters someone else runs. Most people start with CKA — it covers broader ground and the cluster-level knowledge makes CKAD easier afterward.

Going for more than one certification?

prepium.sh covers all three CNCF Kubernetes certifications on the same lab engine — real clusters, programmatic validators, and persistent attempt history.

Open a cluster. Start ticking tasks.

One-time €29 unlocks 14 days of full lab access — all 17 graded tasks, unlimited retakes, and a permanent attempt history. No subscription, no upsell.

  • One-time €29 · 14 days of access
  • 7-day refund, no questions asked
  • Real cluster, not a simulator
  • Programmatic, partial-credit scoring
  • Solutions revealed after submission
  • Persistent, per-user attempt log