Practice Exams:

Linux Foundation KCNA Dumps

EXAM: Kubernetes and Cloud Native Associate
Verified By Experts
Premium Bundle
$40.00
$59.97 $99.97

KCNA Premium Bundle


  • Premium File 333 Questions & Answers
  • Training Course 54 Lectures
  • Study Guide 410 Pages

Linux Foundation KCNA Practice Test 31

Question 300

A hosting provider taints maintenance workers and needs to distinguish tolerated jobs from guaranteed placement. Which implementation together with matching acceptance evidence best meets this case?

  1. Give each workload an appropriate Kubernetes ServiceAccount and narrow RBAC permissions rather than embedding an administrator kubeconfig in its image. Acceptance evidence: Run a synthetic read-only Pod with a restricted ServiceAccount and verify one permitted API read and one forbidden Secret list.
  2. Use an authorized kubectl debug ephemeral container when an application image lacks troubleshooting tools and policy permits ephemeral containers. Acceptance evidence: In a lab add a read-only debug container, inspect network conditions and verify audit logs record the ephemeral container action.
  3. Taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required. Acceptance evidence: Taint a lab worker dedicated=training:NoSchedule, then compare placement of Pods with and without the toleration.
  4. Use requiredDuringSchedulingIgnoredDuringExecution for mandatory constraints and preferredDuringSchedulingIgnoredDuringExecution for a soft placement preference. Acceptance evidence: Test the workload with and without a preferred zone and verify it still schedules when preferred nodes are busy.

Explanation

Correct Answer

C:

The situation is that a hosting provider taints maintenance workers and needs to distinguish tolerated jobs from guaranteed placement. Taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required. Acceptance evidence: Taint a lab worker dedicated=training:NoSchedule, then compare placement of Pods with and without the toleration. This pairing is justified because a NoSchedule taint repels Pods without a matching toleration; tolerance permits consideration but does not force placement on that node.

Incorrect Answers

A:

This action and check address the distinct objective “Differentiate ServiceAccounts from human kubeconfig identities” instead of the stated acceptance requirement. The matching action is to taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required. Its confirming test is to taint a lab worker dedicated=training:NoSchedule, then compare placement of Pods with and without the toleration.

B:

This action and check address the distinct objective “Use ephemeral debugging containers only with appropriate permission and safeguards” instead of the stated acceptance requirement. The matching action is to taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required. Its confirming test is to taint a lab worker dedicated=training:NoSchedule, then compare placement of Pods with and without the toleration.

D:

This action and check address the distinct objective “Distinguish preferred node affinity from required scheduling rules” instead of the stated acceptance requirement. The matching action is to taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required. Its confirming test is to taint a lab worker dedicated=training:NoSchedule, then compare placement of Pods with and without the toleration.

Question 301

A bank's Kubernetes workers require kernel updates without disrupting a high-availability checkout API. Which engineering approach is the most appropriate for this cloud-native case?

  1. Record non-identifying build or audit metadata as annotations and use labels for indexing, grouping and selector-based routing.
  2. Configure anti-affinity against other Pods with a specific label so replicas prefer or require separate failure domains as appropriate.
  3. Inspect probe configuration, Pod Ready conditions and restart events to identify whether traffic eligibility or container execution is failing.
  4. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance.

Explanation

Correct Answer

D:

Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. This resolves the issue because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit. To verify, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Incorrect Answers

A:

This option addresses the distinct objective “Use annotations for tool metadata rather than workload selection”, not the immediate requirement. Labels are designed for selectors and filtering, whereas annotations hold descriptive metadata not intended as a resource-matching control. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

B:

This option addresses the distinct objective “Use Pod anti-affinity to reduce correlated replica failure”, not the immediate requirement. Pod anti-affinity considers labels on existing Pods and topology keys such as hostname; it does not directly guarantee unlimited nodes. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

C:

This option addresses the distinct objective “Investigate readiness failures separately from liveness-related restarts”, not the immediate requirement. A failed readiness probe removes a Pod from ready Service endpoints, while failed liveness checks can restart a process; their operational consequences differ. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Question 302

A bank's Kubernetes workers require kernel updates without disrupting a high-availability checkout API. What infrastructure principle makes the correct treatment effective?

  1. NetworkPolicies select Pods and define allowed flows; when a Pod becomes isolated, traffic is allowed only as permitted by applicable rules and plugin behavior.
  2. The scheduler evaluates requests against node allocatable capacity; limits govern runtime ceilings and do not substitute for scheduling requests.
  3. Deployment rollout strategy gradually replaces Pods and preserves revision history, but success depends on availability and revision retention settings.
  4. Cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Explanation

Correct Answer

D:

Cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit. It supports this action: Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. To confirm, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Incorrect Answers

A:

That statement explains the distinct objective “Limit workload-to-workload access with explicit NetworkPolicy allow rules” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

B:

That statement explains the distinct objective “Use container resource requests to help the scheduler fit Pods on nodes” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

C:

That statement explains the distinct objective “Use rollout status and undo to validate and recover a Deployment release” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Question 303

A bank's Kubernetes workers require kernel updates without disrupting a high-availability checkout API. What verification result would most convincingly distinguish a working solution from a claim?

  1. Run a test web Pod without a token mount and verify service operation succeeds while in-cluster API calls are unavailable.
  2. Test the workload with and without a preferred zone and verify it still schedules when preferred nodes are busy.
  3. Delay a synthetic app response and compare EndpointSlice readiness with restart counts under controlled probe settings.
  4. Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Explanation

Correct Answer

D:

Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility. This is the relevant test because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

A:

This check would help evaluate the distinct objective “Enforce workload service account token boundaries and avoid needless mounts”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

B:

This check would help evaluate the distinct objective “Distinguish preferred node affinity from required scheduling rules”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

C:

This check would help evaluate the distinct objective “Understand readiness, liveness and startup probes as different signals”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Question 304

A bank's Kubernetes workers require kernel updates without disrupting a high-availability checkout API. The change review contains this proposal: “Power off a node with the last application replica without checking disruption budgets or replacement capacity.” What risk matters most?

  1. The operator can interrupt healthy networking while the original admission or image problem persists.
  2. The scheduler could place multiple replicas together, leaving a node without the intended local collector.
  3. The planned maintenance causes avoidable downtime because no safe application replica remains available.
  4. The Pod remains unschedulable because requested capacity and node feasibility did not change.

Explanation

Correct Answer

C:

The planned maintenance causes avoidable downtime because no safe application replica remains available. The shortcut creates this risk because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

A:

That outcome concerns the distinct objective “Use describe and Events to distinguish admission, scheduling and execution errors”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

B:

That outcome concerns the distinct objective “Select DaemonSet for node-local agents rather than Deployment replica counts”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

D:

That outcome concerns the distinct objective “Use container resource requests to help the scheduler fit Pods on nodes”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

Question 305

A bank's Kubernetes workers require kernel updates without disrupting a high-availability checkout API. Which implementation together with matching acceptance evidence best meets this case?

  1. Expose appropriately labeled /metrics endpoints or exporters and configure Prometheus scrape targets to collect time-series measurements. Acceptance evidence: Verify a synthetic exporter is listed UP in Prometheus target status and query its request counter after generating traffic.
  2. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Acceptance evidence: Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.
  3. Grant operators only required API verbs and resources using RBAC, then test with kubectl auth can-i and separate synthetic identities. Acceptance evidence: With restricted identities, check get pods succeeds while delete secrets and create clusterroles are denied.
  4. Use requiredDuringSchedulingIgnoredDuringExecution for mandatory constraints and preferredDuringSchedulingIgnoredDuringExecution for a soft placement preference. Acceptance evidence: Test the workload with and without a preferred zone and verify it still schedules when preferred nodes are busy.

Explanation

Correct Answer

B:

Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Acceptance evidence: Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility. This pairing is justified because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

A:

This action and check address the distinct objective “Understand Prometheus pull-based metrics collection and exporters” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

C:

This action and check address the distinct objective “Use least-privilege kubectl authentication and authorization during routine administration” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

D:

This action and check address the distinct objective “Distinguish preferred node affinity from required scheduling rules” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Question 306

A training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Which Kubernetes or cloud-native implementation decision best solves the stated problem?

  1. Inspect Deployment, ReplicaSet and Pod owner references when verifying which controller maintains replica count.
  2. Taint dedicated nodes and grant only intended workloads matching tolerations, combining node affinity when exclusive placement is required.
  3. Give each Pod an appropriate ServiceAccount, disable automountServiceAccountToken where API access is unnecessary and test token audience and role scope.
  4. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance.

Explanation

Correct Answer

D:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. This resolves the issue because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit. To verify, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Incorrect Answers

A:

This option addresses the distinct objective “Interpret ReplicaSet ownership and selectors in a Deployment”, not the immediate requirement. A Deployment creates or manages ReplicaSets, each ReplicaSet owns matching Pods, and selector consistency determines which Pods it counts. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

B:

This option addresses the distinct objective “Understand taints and tolerations as repulsion controls”, not the immediate requirement. A NoSchedule taint repels Pods without a matching toleration; tolerance permits consideration but does not force placement on that node. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

C:

This option addresses the distinct objective “Enforce workload service account token boundaries and avoid needless mounts”, not the immediate requirement. Mounted credentials can enable Kubernetes API calls from a compromised container; explicit token exposure and RBAC scope limit abuse. For the stated case, cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Question 307

A training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Which platform mechanism explains why the suitable approach works?

  1. CPU limits may throttle running processes, whereas exceeding memory constraints can cause OOM termination; Pod scheduling itself depends on requests.
  2. Open-source community processes provide transparent review and shared maintenance, but do not imply unreviewed external code should be merged into production.
  3. Cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.
  4. A quota constrains aggregate namespace resource usage, while a LimitRange validates or defaults eligible individual resource settings.

Explanation

Correct Answer

C:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit. It supports this action: Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. To confirm, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Incorrect Answers

A:

That statement explains the distinct objective “Distinguish memory limits from CPU throttling for scheduling and runtime symptoms” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

B:

That statement explains the distinct objective “Use open community governance and contribution processes for sustainable cloud-native projects” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

D:

That statement explains the distinct objective “Understand resource quotas and LimitRanges for namespace capacity governance” rather than the cause in this case. The deciding mechanism here is that cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Question 308

A training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Which controlled observation would best validate the intended improvement?

  1. Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.
  2. Set an intentionally excessive memory request in a lab and inspect Pod Pending Events before adjusting it.
  3. Run a synthetic read-only Pod with a restricted ServiceAccount and verify one permitted API read and one forbidden Secret list.
  4. Test a synthetic standard API manifest and OpenTelemetry trace export across two platforms, documenting any unsupported extensions.

Explanation

Correct Answer

A:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility. This is the relevant test because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

B:

This check would help evaluate the distinct objective “Use container resource requests to help the scheduler fit Pods on nodes”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

C:

This check would help evaluate the distinct objective “Differentiate ServiceAccounts from human kubeconfig identities”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

D:

This check would help evaluate the distinct objective “Understand open standards and vendor-neutral community collaboration”, but not the required result. For this case, observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Question 309

A training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. An engineer proposes: “Power off a node with the last application replica without checking disruption budgets or replacement capacity.” What adverse outcome would be most direct?

  1. Controllers may fail to recognize intended Pods or adopt unwanted matches, producing confusing replacement behavior.
  2. A migration fails because backend volume classes and cloud authentication semantics differ.
  3. The Pod remains unschedulable because requested capacity and node feasibility did not change.
  4. The planned maintenance causes avoidable downtime because no safe application replica remains available.

Explanation

Correct Answer

D:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. The planned maintenance causes avoidable downtime because no safe application replica remains available. The shortcut creates this risk because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

A:

That outcome concerns the distinct objective “Interpret ReplicaSet ownership and selectors in a Deployment”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

B:

That outcome concerns the distinct objective “Apply cloud-native portability and immutable infrastructure principles pragmatically”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

C:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. That outcome concerns the distinct objective “Use container resource requests to help the scheduler fit Pods on nodes”; it is not the most direct consequence of this shortcut. The relevant consequence is that the planned maintenance causes avoidable downtime because no safe application replica remains available.

Question 310

A training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. What pair of supported action and practical verification should be approved?

  1. Run a logging or node monitoring agent as a DaemonSet when every eligible worker needs its own instance. Acceptance evidence: Add a test worker node and inspect the corresponding DaemonSet agent Pod without scaling a Deployment.
  2. Use the scheduler's node-binding decision to explain initial Pod placement, then inspect the chosen node's kubelet for local container lifecycle status. Acceptance evidence: Inspect a new Pod's nodeName and Events, then compare scheduling events with kubelet-reported container status.
  3. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Acceptance evidence: Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.
  4. Require approved image registries, signature or digest verification and patch ownership when admitting production container images. Acceptance evidence: Deploy one signed approved image and one unverified synthetic image under a test admission policy, recording expected allow and deny results.

Explanation

Correct Answer

C:

The situation is that a training cluster operator must drain one node while several DaemonSet and StatefulSet Pods exist. Cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Acceptance evidence: Observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility. This pairing is justified because cordon marks a node unschedulable for new ordinary Pods; drain evicts managed workloads when policies and availability permit.

Incorrect Answers

A:

This action and check address the distinct objective “Select DaemonSet for node-local agents rather than Deployment replica counts” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

B:

This action and check address the distinct objective “Separate kube-scheduler decisions from kubelet execution” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

D:

This action and check address the distinct objective “Use image provenance and policy controls without trusting mutable tags” instead of the stated acceptance requirement. The matching action is to cordon a worker to prevent new scheduling and drain safely under PodDisruptionBudgets and DaemonSet rules before maintenance. Its confirming test is to observe a controlled drain with replica owners and a PDB, then verify workloads move and uncordon restores eligibility.

Kubernetes and Cloud Native Associate (KCNA) Practice Tests