Practice round

{{ summaryValue }}

{{ summaryLine }}

CKA v1.37 concepts

Unofficial, independently authored practice for the Linux Foundation / CNCF Certified Kubernetes Administrator curriculum (CKA_Curriculum_v1.37.pdf), verified 17 October 2026. This bank supports only that version. The official exam requires hands-on command-line tasks; this multiple-choice tool does not run a cluster or assess those skills. Check the official target before your exam.

Leave blank to use the available minor 1.37. Pin 1.37 or v1.37 to require it; other versions, including patch-specific versions, are unavailable.
Use the date from a practice link or study record, or leave blank to use the available target. The local admitted curriculum snapshot is 2026-10-17. This is not your exam date or a provider effective date.

{{ examTargetValidation }}

{{ conflict.key }}: {{ conflict.values.join(' | ') }}

{{ params.curriculum_date || params.kubernetes_version ? 'The available target meets these requirements.' : 'No target requirements: use the available target.' }}

Both actions clear any saved attempt. Start a new round after choosing your requirements.
Available target
Foundation asks direct recall or identification. Applied asks you to interpret constraints, scope, lifecycle or a failure. These editorial levels are not calibrated to official exam difficulty.
Keep the seed to replay the same questions and option order.

54 original items cover all 27 objectives at concept level. Questions are sampled uniformly without replacement; rounds do not reproduce official domain weights. Answers and the replay seed are saved in this page URL, so shared links reveal the attempt.

{{ examSetupValidation }}

{{ progressPercent }}%
{{ answeredCount }} of {{ questionCount }} answered

{{ progressStatusLine }}

Question {{ examCursor + 1 }} of {{ questionCount }}

{{ currentQuestion.stem }}

{{ currentQuestion.accessible_clue }}

Answer saved. Corrections will appear after this round.

Review your practice

{{ resultLead }}

{{ resolvedTargetLabel }}

{{ readinessReviewTitle }}

{{ readinessReviewLine }}

This unofficial CKA v1.37 round measures only answers to these original questions. It does not test successful cluster changes, troubleshooting under exam conditions or hands-on speed, and it cannot predict passing. Practise the objectives in a disposable cluster and consult the current official curriculum.

  • {{ domain.label }} — {{ domain.context }}: {{ domain.study }}
{{ card.label }}
{{ card.value }}
{{ card.note }}

Missed answer review

  1. {{ item.prompt }}

    Your answer: {{ item.selected_answer_label }}

    Correct answer: {{ item.correct_answer_label }}

    {{ item.explanation }}

    Source: {{ sourceTitle(item.source) }} · {{ item.source.version }}

Domain performance

Correct and missed counts for the five CKA domains. Zero sampled answers means untested, not failed. Official domain percentages describe the blueprint; this bank and its sampled rounds are not blueprint-weighted.

Loading domain-performance chart…

{{ examChartStatus }}
  • {{ domain.label }} (blueprint {{ domain.weight_percent }}%): {{ domain.tested ? `${domain.correct_count} correct, ${domain.incorrect_count} missed / ${domain.answered_count} answered` : 'Untested in this round' }}

Answer ledger

Question ID Prompt Domain Your answer Correct answer Outcome Explanation Source Copy
{{ row.question_id }} {{ row.prompt }} {{ row.category_label }} {{ row.selected_answer_label }} {{ row.correct_answer_label }} {{ row.outcome_label }} {{ row.explanation }}
Kubernetes and Cloud Native Associate Exam Prep Practise KCNA concepts with original questions and blueprint-weighted rounds, then use explanations to review mistakes across four domains.
Certified Kubernetes Application Developer Concepts Exam Prep Practice CKAD v1.37 concepts with unofficial questions and explanations, then use missed answers to focus your hands-on Kubernetes study.
Kubernetes NetworkPolicy Generator Build Kubernetes NetworkPolicy YAML from pod labels and allow rules with selector checks, rollout guidance, and security review warnings.
Kubernetes Pod Security Context Checker Check Kubernetes workload YAML for inherited security contexts, Pod Security gaps, weighted findings, and a reviewable hardening patch.

A Kubernetes resource can be created successfully without producing a working application. A Deployment may exist while its Pods remain unscheduled, a Service may have no ready backends, or a storage claim may wait for a suitable volume. Kubernetes administration involves finding the unmet requirement, making a targeted change, and checking the resulting behavior rather than stopping at a successful command.

A cluster combines a control plane with nodes that run application containers. The control plane accepts requests, maintains the desired configuration, and coordinates work. The scheduler chooses suitable nodes for Pods, while each node's kubelet supervises its containers. A Pod groups containers that run together and share networking. When an application needs several copies or a controlled update, a Deployment manages its replicated Pods.

The distinction between desired and observed state is central to troubleshooting. Requesting three application replicas expresses an intention; it does not prove that three healthy replicas are serving requests. Available resources, placement restrictions, application health, and storage availability can prevent the requested state from being reached. An administrator needs both the configuration and evidence from status, events, and logs to identify the cause.

Service
A network access point for an application. For a selector-based Service, matching Pod labels identify the intended backends. Application health and the backend port still matter.
PersistentVolumeClaim and PersistentVolume
A PersistentVolumeClaim (PVC) requests storage for a workload. A PersistentVolume (PV) represents storage that can satisfy a claim. Provisioning, binding, access modes, and reclamation govern different parts of that storage lifecycle.
Namespace and access control
A namespace gives many resources a defined scope within a cluster. Role-based access control (RBAC) determines which operations an identity may perform. Being able to reach the Kubernetes application programming interface (API) does not by itself grant permission to change resources.

The Certified Kubernetes Administrator (CKA) examination, created by the Linux Foundation and the Cloud Native Computing Foundation (CNCF), assesses administration through hands-on command-line tasks. Candidates have two hours to complete the official examination. Its scope includes storage, troubleshooting, workloads and scheduling, cluster architecture and installation, and services and networking.

These areas interact during ordinary administration. A storage decision can restrict where a Pod runs, and a running container can still be unsuitable for application traffic. Changing several unrelated settings at once makes the cause harder to isolate. Useful preparation therefore includes diagnosing a specific fault, applying an appropriate correction, and verifying the intended outcome in a disposable cluster.

Multiple-choice concept practice can expose a mistaken definition or an incorrect interpretation of a command. It cannot demonstrate successful cluster changes, troubleshooting speed, or the ability to work under examination conditions. Reading a correct explanation and carrying out the corresponding task are different achievements. Version-matched documentation and practical exercises remain necessary alongside knowledge review.

How to Use This Tool:

This is unofficial English-language concept practice, not a cluster simulator or a Linux Foundation or CNCF examination. The starting setup uses All CKA domains, All levels, After each answer, and 10 questions. Rounds are untimed.

  1. Check the study target. Compare the current official CKA version with your study plan before relying on this material. Leave both optional requirements blank to use Kubernetes 1.37 and the local curriculum snapshot of October 17, 2026. To require that exact target, enter 1.37 or v1.37 and 2026-10-17, or select Pin available target. The snapshot date is not your examination date or a provider effective date.
    Other Kubernetes versions, including patch-specific values such as 1.37.0, and other snapshot dates are unavailable. Invalid or conflicting requirements block the round. Edit the requirement or choose Pin available target or Use available target. Both shortcuts discard any saved attempt; the latter clears the requirements rather than selecting another version. An outdated or unreadable saved attempt must be restarted, and its answers cannot be migrated.
  2. Choose a practice set and difficulty. Select all domains, one domain, Concept foundations — all domains, or Command and failure reasoning — all domains. The separate Question difficulty filter offers direct recall and identification through Foundation, or constraint, scope, lifecycle, and failure reasoning through Applied. A concept question can be Applied, and a command question can be Foundation. These are editorial distinctions, not official difficulty ratings.
  3. Choose the round length and replay seed. With all domains and all levels selected, the question counts are 5, 10, 20, 27, or 54. Other selections have smaller menus; Storage with Foundation contains just one question. Changing the set or difficulty can reset a count that is no longer allowed. Under Advanced, keep the default Replay seed of cka-135 or enter your own nonblank text within the 128-character input limit. Emoji can reach that limit sooner than plain letters. Keep the same setup as well as the seed when reproducing a round.
  4. Choose when corrections appear. After each answer reveals the outcome, explanation, and source after each selection. At the end saves the answer without revealing its correction until the round is finished. Both modes use the same one-answer format and lock a choice as soon as it is selected.
  5. Start concept practice and answer each question. Select one of the four choices, then use Next question. After the last answer, select View results to open the completed review. Reset to setup clears the attempt rather than pausing it.
  6. Review the missed answers before choosing another round. Use Retry missed for the questions answered incorrectly, Replay same seed to repeat the same round, or Try a new seed for another draw from the current eligible questions. Change setup clears the attempt and returns to the selections. Connect the corrections in the Answer ledger to the suggested hands-on study, rather than treating a replay score as a new examination result.

Interpreting Results:

The completed Practice round score is the number of correct answers out of the questions in that round. During practice, the answered count and progress bar measure completion, not correctness. Reaching 100% progress means every question has an answer, even when some or all answers are wrong.

Domain performance shows correct and missed counts for the five domains. A domain marked Untested in this round has no sampled answers and supplies no evidence of either success or failure. The official blueprint percentages shown beside the domains are contextual information, not weights applied to the questions or score.

The Readiness review is study guidance, not a readiness grade. There is no passing cutoff, official scaled score, pass prediction, or certificate. Even a perfect result concerns only the original questions answered in that round. The questions are independently authored, not official examination items. Reference links support the explanations without making the practice provider-endorsed.

Use Missed answer review to identify the specific misconception, then read its explanation and technical source. The Answer ledger also retains correct answers, which is useful when a correct selection was a guess. Test the corresponding objective in a disposable cluster. A higher score on familiar questions may reflect recall of the answers, so compare rounds only with their set, difficulty, length, and prior exposure in mind.

A chart-loading problem does not remove the written domain counts. Read those counts and use Retry chart when the chart is unavailable.

Technical Details:

Kubernetes resources express different kinds of constraints. Namespaces limit some object relationships, storage binding can depend on Pod placement, and health probes distinguish traffic readiness from restart decisions. Correct administration requires applying the rule that governs the failing relationship, rather than choosing any command that changes the cluster.

Namespace boundaries, storage binding and health checks

Resource or condition Governing rule Practical consequence
Pod, PVC, and PV A Pod uses a PVC in its own namespace. The PV is cluster-scoped, and a PV-to-claim binding is one-to-one. A matching capacity request does not make one bound PV available to every claim.
RoleBinding referencing a ClusterRole The binding grants the relevant namespaced permissions within the RoleBinding's namespace. A ClusterRole reference does not automatically grant those permissions across all namespaces.
Topology-constrained storage WaitForFirstConsumer delays volume binding or provisioning until a consuming Pod exists, allowing its scheduling constraints to be considered. With a supporting storage driver, this avoids choosing storage without considering where the Pod can run.
Readiness probe Reports whether a container is ready to serve requests and affects the Pod's eligibility for normal Service traffic. Readiness failure alone does not request a container restart.
Liveness probe Failure at the configured threshold causes the container to be killed, with restart behavior governed by its restart policy. A check that mistakes temporary slowness for unrecoverable failure can create unnecessary restarts.
Startup probe Holds back liveness and readiness probes until startup succeeds. It allows a separate initialization budget while retaining later health checks. Exhausting that budget can still cause the container to be killed.

Resource acceptance and operational success also differ when a controller is missing or unhealthy. A responding API server does not establish that scheduling and reconciliation are progressing. For application networking, compare the Service selector with Pod labels, inspect the discovered backend addresses and ports in EndpointSlices, and check application readiness. Each observation narrows a different possible cause.

Curriculum mapping and sampling

The practice target is the CKA v1.37 curriculum, with a local snapshot dated October 17, 2026. The bank contains 54 original questions mapped as two questions for each of the 27 curriculum objectives. The mapping describes intended concept coverage, not exhaustive testing of every task or practical skill within an objective.

Domain Official weight Original questions Examples of practice content
Storage 10% 6 Provisioning, claims, access modes, and reclaim policies.
Troubleshooting 30% 10 Scheduling failures, node and control-plane health, metrics, logs, and Service faults.
Workloads & Scheduling 15% 10 Rollouts, configuration, autoscaling, probes, affinity, and tolerations.
Cluster Architecture, Installation & Configuration 25% 16 RBAC, runtimes, kubeadm, highly available control planes, Helm 3, Kustomize, extension interfaces, and operators.
Services & Networking 20% 12 Pod connectivity, NetworkPolicy, Services, Gateway API, Ingress, and CoreDNS.

Question selection is uniform without replacement within the chosen set and difficulty. No question repeats within one round, but a short all-domain round need not include every domain. There are no official domain quotas. Even a full 54-question round gives Troubleshooting 10 equally scored questions rather than applying its official 30% weight.

Across the whole bank, 22 questions are classified as Foundation and 32 as Applied. These difficulty groups are separate from the two 27-question content sets for concept foundations and command or failure reasoning. Selecting a smaller group changes which questions can be sampled; it does not make the scoring stricter or more lenient.

With the same question bank, selection rules, eligible questions, feedback mode, count, and seed, both question order and option order are reproducible. A new seed rearranges and samples from a finite bank rather than creating new questions, so different rounds can overlap. A missed-only retry has a narrower scope and should not be compared with a broad round as though both tested the same material.

Counting correct, missed and unanswered questions

Each question has one keyed answer among four choices. A correct selection earns one point, and an incorrect selection earns zero. There is no partial credit or negative marking. Let C be the correct count, A the answered count, and N the total questions selected for the round.

The point score equals the correct count, with the round's question count as the possible points.

Points=C , Possible points=N

Accuracy divides correct answers by all questions in the round, not just the questions answered so far.

Accuracy (%)=100× CN

Progress uses answered questions instead. It reaches 100% when all questions have been answered, regardless of correctness.

Progress (%)=100× AN

The missed count is the answered count minus the correct count. Unanswered questions remain in the accuracy denominator but are not counted as incorrect. Scores range from zero to N points, and percentage values range from 0% to 100%. Counts are exact integers, and the main score is displayed as correct answers out of total questions. Raw accuracy and displayed progress both use the unrounded fraction. Completion requires every question to be answered, not a particular accuracy percentage.

Privacy Notes:

Answers are scored in the browser, but the attempt is also saved in the page URL. That saved information includes the replay seed and responses and is not encrypted. A compatible saved URL can restore the questions, answers, and position in the round.

  • Anyone receiving an attempt link can see the saved responses. The URL can remain in browser history and reach the site when the page is loaded or reloaded. Do not put confidential information in a seed or treat an attempt link as private.
  • Downloaded or copied answer ledgers also contain responses and study context. Check their contents before sharing them.
  • Page resources, additional chart and document features, and opened references can require network access. Browser scoring does not establish fully offline operation or a guarantee that attempt information never leaves the device.

Worked Examples:

Two storage misconceptions in a six-question round

Select Storage, All levels, and 6 questions to include the whole storage set. Suppose the four provisioning and claim questions are answered correctly, but the access-mode answer incorrectly treats ReadWriteOnce as a single-Pod restriction and the reclaim-policy answer incorrectly treats a released Retain volume as immediately empty and reusable. The result is 4/6, with 4 correct and 2 missed in Storage; the other four domains are untested. Retry missed gives a two-question correction round. ReadWriteOnce can allow several Pods on the same node; single-Pod exclusivity uses the separate ReadWriteOncePod mode on supported Container Storage Interface (CSI) volumes. Retain leaves the previous data for manual reclamation. Those are the two behaviors to investigate in a disposable storage exercise.

Correcting a Service's backend port

In the Troubleshooting set, a question describes ready Pods behind a Service on port 80. The Service's targetPort is 8080, but the application listens on 9090. With After each answer selected, choosing the correction that changes targetPort to 9090 shows correct feedback and earns one point. Changing only the exposure type to NodePort would leave the backend mismatch intact. The practice answer does not change a cluster: a corresponding lab check must inspect the actual Service mapping and test a request after the correction.

References: