Key takeaways
- Exams fall at the end of weeks 2, 4, 8, 10 and 13; the hands-on core gets nine of the thirteen weeks.
- Each week has a milestone; missing two in a row is the signal to stretch the plan, not to cram.
- Use the two Killer.sh simulator sessions included with each hands-on registration in that exam’s last ten days.
- The five exams are $1,835 at list price or $1,645 as the Kubestronaut bundle, and 540 minutes of exam time.
- All five must be active at the same time, so a short, dense schedule also keeps the expiry dates close together.
On this page
The schedule at a glance
The thirteen weeks split unevenly on purpose. The two multiple-choice exams get two weeks each, because 60 questions on a well-defined curriculum respond quickly to daily practice. CKA gets four weeks, because it is the base for the other two hands-on exams. CKAD gets two, because it shares most of its core with CKA. CKS gets three, because each task adds a security tool to the cluster skills.
| Weeks | Exam | Format | Exam at end of |
|---|---|---|---|
| 1-2 | KCNA | Multiple-choice | Week 2 |
| 3-4 | KCSA | Multiple-choice | Week 4 |
| 5-8 | CKA | Hands-on | Week 8 |
| 9-10 | CKAD | Hands-on | Week 10 |
| 11-13 | CKS | Hands-on | Week 13 |
Weeks 1-4: KCNA and KCSA
The first month is retrieval practice on two broad curricula. Answer a daily set of questions, clear due reviews first, and read every explanation. Build a small local cluster in week one and use it to make the objects you read about real. GoldenKube has 288 KCNA and 378 KCSA questions, which is enough to run a readiness check on new questions before each booking. Keep these weeks light on hours if you can; the hands-on weeks that follow need more of them. The aim is to pass both exams with enough margin that neither needs a retake.
| Week | Work | Milestone |
|---|---|---|
| 1 | KCNA curriculum, control plane and workload objects; local cluster running | First-attempt accuracy known for each KCNA domain |
| 2 | Networking, storage, delivery and architecture; one timed mock | Every domain above the pass mark on new questions; sit KCNA |
| 3 | KCSA: the 4Cs, cluster components, security fundamentals | A one-line control written for every cluster component |
| 4 | Threat model, platform security, compliance; one timed mock | Every domain above the pass mark on new questions; sit KCSA |
Weeks 5-8: CKA
CKA is the centre of the schedule. Switch from questions to tasks: every session should end with something you built or fixed in a cluster. Keep the cluster on Kubernetes v1.37. Rebuild it with kubeadm at least once, because installation and upgrades are tasks, and practise troubleshooting broken nodes and components until the order of checks is automatic.
| Week | Work | Milestone |
|---|---|---|
| 5 | Imperative kubectl, workloads, rollouts, ConfigMaps and Secrets | Create and update a Deployment, Service and ConfigMap without opening the docs |
| 6 | Services, Ingress, Gateway API, NetworkPolicy, CoreDNS; storage | A default-deny policy and an allow rule proven with a test Pod |
| 7 | kubeadm install and upgrade, RBAC, Helm and Kustomize; troubleshooting drills | A cluster upgraded by one minor version without errors |
| 8 | Both simulator sessions, retry missed tasks timed | Simulator tasks finished with time to spare; sit CKA |
Weeks 9-10: CKAD
Two weeks are enough for CKAD straight after CKA, because the shared core is fresh. Spend them on what CKA does not emphasise: building images, multi-container Pod patterns, probes, SecurityContexts, ServiceAccounts, Jobs and CronJobs, and deployment strategies built from Kubernetes primitives. If CKA felt tight on time, spend the first days of week nine on speed with the shared core before moving to the CKAD-only topics.
| Week | Work | Milestone |
|---|---|---|
| 9 | Images, multi-container Pods, probes, Jobs and CronJobs, SecurityContexts | A sidecar and an init container working in one Pod |
| 10 | Canary and blue/green with primitives, API deprecations; both simulator sessions | Simulator tasks finished with time to spare; sit CKAD |
Weeks 11-13: CKS
CKS requires the passed CKA from week 8, and its environment runs Kubernetes v1.35. Three weeks cover the six domains if each session configures a control and then proves it works. Keep a backup of every file you change under /etc/kubernetes, because a broken API server costs every remaining task.
| Week | Work | Milestone |
|---|---|---|
| 11 | Cluster setup and hardening: NetworkPolicies, benchmark review, RBAC, ServiceAccounts; AppArmor and seccomp | A Pod running under an AppArmor profile you loaded |
| 12 | Pod Security Admission, Secrets encryption, sandboxed runtimes, supply chain, static analysis | An image from a disallowed registry rejected at admission |
| 13 | Runtime detection, audit logs, immutability; both simulator sessions | Simulator tasks finished with time to spare; sit CKS |
A day in the schedule
A consistent daily shape matters more than long sessions. On multiple-choice weeks, the day is a set of questions, the due reviews first and the explanations read. On hands-on weeks, it is one or two timed tasks, a written note of what went wrong, and a short return to yesterday’s mistake. One day a week is lighter: review the mistake list and rebuild the cluster if it has drifted. On hands-on days, the note of what went wrong is the most valuable output. Read the week’s notes before each simulator session; they show the errors you are most likely to repeat under time pressure.
- Clear due reviews or redo yesterday’s failed task.
- Do the day’s new work: a question set or a timed task.
- Write one line for every mistake: the concept or command, not the answer.
- Check the week’s milestone and move a booking early if it is slipping.
When to stretch the plan
Milestones exist so you can change the plan early. Missing one is normal; missing two in a row means the plan is too dense for your week, and the fix is to add a week, not to cram. Stretch the hands-on weeks first, because they reward repetition and cannot be compressed by reading. A plan that takes four months and passes every exam the first time is cheaper and faster than a 90-day plan with two retakes.
Hands-on labs for CKA, CKAD, CKS, LFCS, ICA and CNPE are in preview: the GoldenKube CLI runs them on your own machine, but it has no public release yet.
- New to Linux: add two weeks before CKA, and consider LFCS-style shell practice first.
- Already administer clusters: shorten CKA to three weeks and give the spare week to CKS.
- A failed exam: two weeks of targeted review, the included retake, then shift the rest.
Exam weeks and bookings
Five exam weeks in thirteen means booking matters. Each exam is a separate purchase with a 12-month window to take it, so buying everything on day one is fine as long as you finish within the year. Book each exam as soon as the previous milestone is met, for a slot at the end of its exam week, and move it if a milestone slips: rescheduling is allowed up to 24 hours before the start, while a missed booking forfeits the attempt.
For CKA, CKAD and CKS the FAQ gives two Killer.sh simulator attempts per registration, each open for 36 hours, except for single-attempt purchases. Start the first a few days before the exam week and the second two or three days before the exam, so you have time to redo the tasks you missed.
- Week 2 and week 4: book KCNA and KCSA for the last day of the week, after a timed mock above the pass mark.
- Week 6: book CKA for the end of week 8 and plan the two simulator sessions for week 8.
- Week 8: after CKA, book CKAD for the end of week 10.
- Week 10: after CKAD, book CKS for the end of week 13; your CKA pass is the prerequisite.
- Any failed exam: book the included retake about two weeks out and shift everything after it.
Frequently asked questions
Is 90 days realistic for Kubestronaut?
For someone who already uses containers and can study an hour or two most days, it is demanding but workable. Nobody official publishes a preparation time. If you are new to Linux or Kubernetes, stretch the hands-on weeks first, for example to six weeks for CKA, rather than squeezing the plan.
Why this order: KCNA, KCSA, CKA, CKAD, CKS?
The two multiple-choice exams build vocabulary at low cost; CKA builds the cluster skills CKAD and CKS assume; and CKS requires a passed CKA by rule. KCSA before CKS also means CKS later renews KCSA under the CARE program.
Do I have to finish within a time limit?
No deadline exists, but all five must be active together: "Yes, all 5 certifications need to be active at the same time." Each exam purchase must also be used within 12 months.
What if I fail one of the exams?
Each purchase includes one retake within 12 months of the purchase, unless it is a single-attempt purchase. Insert two weeks of targeted review, retake it, and shift the rest of the schedule by the same two weeks.
When should I book each exam?
Book once the previous week’s milestone is met, for a slot at the end of the exam week. Bookings can be changed up to 24 hours before the start, so booking early and moving it if needed is safe; a no-show is not.
Which Kubernetes version should I practise on?
The candidate instructions say CKA and CKAD run Kubernetes v1.37 and CKS runs v1.35. Keep your practice cluster on those minors for the matching weeks.
Sources
Every exam fact on this page comes from these pages, read on the date shown.
- Kubestronaut program page, CNCF, checked
- Kubestronaut FAQ, CNCF, checked
- Linux Foundation: Kubernetes and Cloud Native Associate (KCNA) certification page, checked
- Linux Foundation: Kubernetes and Cloud Native Security Associate (KCSA) certification page, checked
- Linux Foundation: Certified Kubernetes Administrator (CKA) certification page, checked
- Linux Foundation: Certified Kubernetes Application Developer (CKAD) certification page, checked
- Linux Foundation: Certified Kubernetes Security Specialist (CKS) certification page, checked
- FAQ: CKA, CKAD and CKS, Linux Foundation certification docs, checked
- Exam terms of service, Linux Foundation certification docs, checked