McAfee Secured

CNCF CKA Bundle

Exam Code: CKA

Exam Name Certified Kubernetes Administrator

Certification Provider: CNCF

Corresponding Certification: CKA

CNCF CKA Bundle $44.99

Pass CNCF CKA with Real Questions From Testking

Download Web's Best CKA Exam Preparation Materials Today

  • Questions & Answers

    CKA Questions & Answers

    58 Questions & Answers

    Questions & Answers Testing Engine software allows you to practice questions and answers in real CKA exam environment.

  • CKA Video Course

    CKA Video Course

    138 Video Lectures

    CKA Video Course is developed by CNCF Professionals to help you pass CKA exam.

    Description

    This course will improve your knowledge and skills required to pass Certified Kubernetes Administrator exam.
  • Study Guide

    CKA Study Guide

    268 PDF Pages

    Study Guide will give you a practical experience regarding the subject and provide an academic background. CKA Study Guide is available in PDF format.

Certified Kubernetes Administrator (CKA) Practice Tests

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Total Cost: $194.97
Bundle Price: $149.98

Purchase Individually

  • Questions & Answers

    Questions & Answers

    58 Questions

    $124.99
  • CKA Video Course

    Video Course

    138 Video Lectures

    $39.99
  • Study Guide

    Study Guide

    268 PDF Pages

    $29.99

CNCF CKA Practice Test 6

Question 50

A policy permits application TCP egress but forgets DNS and Pods cannot resolve Service names. An engineer recommends: Permit DNS traffic to the intended cluster DNS endpoints under the egress policy. Which Kubernetes behavior best explains why this correction fits?

  1. An egress-isolated Pod needs an explicit permitted path to each required destination.
  2. DNS name resolution is a separate network flow from the application connection.
  3. NetworkPolicies are additive for selected Pods; allow rules define permitted flows after isolation.
  4. Ingress and egress restrictions are evaluated independently and policy effects compose additively.

Explanation

Correct Answer

B:

DNS name resolution is a separate network flow from the application connection. This explains why the proposed action—permit DNS traffic to the intended cluster DNS endpoints under the egress policy.—is appropriate. Direct confirmation is: The Pod resolves a Service name and reaches it while unrelated outbound access remains denied.

Incorrect Answers

A:

An egress-isolated Pod needs an explicit permitted path to each required destination. This does not establish the proposed intervention. The governing principle here is: DNS name resolution is a separate network flow from the application connection. Apply it by: Permit DNS traffic to the intended cluster DNS endpoints under the egress policy.

C:

NetworkPolicies are additive for selected Pods; allow rules define permitted flows after isolation. This does not establish the proposed intervention. The governing principle here is: DNS name resolution is a separate network flow from the application connection. Apply it by: Permit DNS traffic to the intended cluster DNS endpoints under the egress policy.

D:

Ingress and egress restrictions are evaluated independently and policy effects compose additively. This does not establish the proposed intervention. The governing principle here is: DNS name resolution is a separate network flow from the application connection. Apply it by: Permit DNS traffic to the intended cluster DNS endpoints under the egress policy.

Question 51

An incident responder has permission to create Pods but not to attach interactively to running containers. An engineer recommends: Grant the narrowly required pods/attach or pods/exec subresource verbs according to the approved response workflow. Which Kubernetes behavior best explains why this correction fits?

  1. RBAC treats interactive Pod subresources separately from operations on the parent Pod resource.
  2. Kubernetes service accounts do not automatically inherit administrator permissions from the Pod owner.
  3. A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces.
  4. Least-privilege RBAC narrows blast radius by limiting the allowed verbs, resources, and API groups.

Explanation

Correct Answer

A:

RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. This explains why the proposed action—grant the narrowly required pods/attach or pods/exec subresource verbs according to the approved response workflow.—is appropriate. Direct confirmation is: Impersonated requests reach the approved subresource and reject unrelated secret reads.

Incorrect Answers

B:

Kubernetes service accounts do not automatically inherit administrator permissions from the Pod owner. This does not establish the proposed intervention. The governing principle here is: RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. Apply it by: Grant the narrowly required pods/attach or pods/exec subresource verbs according to the approved response workflow.

C:

A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. This does not establish the proposed intervention. The governing principle here is: RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. Apply it by: Grant the narrowly required pods/attach or pods/exec subresource verbs according to the approved response workflow.

D:

Least-privilege RBAC narrows blast radius by limiting the allowed verbs, resources, and API groups. This does not establish the proposed intervention. The governing principle here is: RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. Apply it by: Grant the narrowly required pods/attach or pods/exec subresource verbs according to the approved response workflow.

Question 52

A horizontal autoscaler stays at its maximum replica count after load subsides and no one checks its stabilization policy. Which Kubernetes administration action most directly addresses the observed failure?

  1. Inspect HPA behavior and scale-down stabilization windows before tuning them for the workload.
  2. Create a CPU-based HPA with matching resource requests and confirm metrics availability.
  3. Use a suitable custom or external metrics source and scale target based on backlog with validated adapters.
  4. Apply the intended image and observe rollout status and updated ReplicaSets before declaring success.

Explanation

Correct Answer

A:

Inspect HPA behavior and scale-down stabilization windows before tuning them for the workload. This corrects the specific failure because hPA may delay downscaling to avoid flapping even after metrics fall. Verify it by checking: Replicas decrease after the approved stabilization interval without oscillating under test load.

Incorrect Answers

B:

HPA CPU utilization calculations depend on resource requests and relevant metrics API data. That is relevant to a different intervention. For the stated situation, the required action is: Inspect HPA behavior and scale-down stabilization windows before tuning them for the workload. The governing reason is: HPA may delay downscaling to avoid flapping even after metrics fall.

C:

Different workload bottlenecks require appropriately supported HPA metrics. That is relevant to a different intervention. For the stated situation, the required action is: Inspect HPA behavior and scale-down stabilization windows before tuning them for the workload. The governing reason is: HPA may delay downscaling to avoid flapping even after metrics fall.

D:

Desired-state changes are asynchronous and a successful API update is not proof of healthy Pods. That is relevant to a different intervention. For the stated situation, the required action is: Inspect HPA behavior and scale-down stabilization windows before tuning them for the workload. The governing reason is: HPA may delay downscaling to avoid flapping even after metrics fall.

Question 53

An Ingress uses ingressClassName internal but the only controller watches class public. An engineer recommends: Align ingressClassName with the intended controller or deploy the requested internal class. Which Kubernetes behavior best explains why this correction fits?

  1. Ingress rules point to Services by name and port within their namespace.
  2. Ingress API objects do not themselves implement HTTP routing; a controller supplies the dataplane.
  3. Ingress TLS termination depends on a usable referenced Secret and an active controller.
  4. Ingress controllers commonly filter resources by configured IngressClass.

Explanation

Correct Answer

D:

Ingress controllers commonly filter resources by configured IngressClass. This explains why the proposed action—align ingressClassName with the intended controller or deploy the requested internal class.—is appropriate. Direct confirmation is: The intended controller accepts the Ingress and the undesired controller ignores it.

Incorrect Answers

A:

Ingress rules point to Services by name and port within their namespace. This does not establish the proposed intervention. The governing principle here is: Ingress controllers commonly filter resources by configured IngressClass. Apply it by: Align ingressClassName with the intended controller or deploy the requested internal class.

B:

Ingress API objects do not themselves implement HTTP routing; a controller supplies the dataplane. This does not establish the proposed intervention. The governing principle here is: Ingress controllers commonly filter resources by configured IngressClass. Apply it by: Align ingressClassName with the intended controller or deploy the requested internal class.

C:

Ingress TLS termination depends on a usable referenced Secret and an active controller. This does not establish the proposed intervention. The governing principle here is: Ingress controllers commonly filter resources by configured IngressClass. Apply it by: Align ingressClassName with the intended controller or deploy the requested internal class.

Question 54

A database must be exposed only to workloads inside the cluster but its Service type was changed to LoadBalancer. Which Kubernetes administration action most directly addresses the observed failure?

  1. Use a LoadBalancer Service with the installed infrastructure integration when a managed external frontend is required.
  2. Return to ClusterIP and remove the unnecessary external frontend.
  3. Document and use headless DNS/endpoint discovery for per-Pod identities instead of VIP load balancing.
  4. Correct the Service selector to match the intended ready Pod labels.

Explanation

Correct Answer

B:

Return to ClusterIP and remove the unnecessary external frontend. This corrects the specific failure because clusterIP is intended for internal service discovery; LoadBalancer requests external integration where supported. Verify it by checking: No external load-balancer address exists and internal client traffic succeeds.

Incorrect Answers

A:

NodePort exposes a port on nodes, while LoadBalancer depends on provider or local controller provisioning. That is relevant to a different intervention. For the stated situation, the required action is: Return to ClusterIP and remove the unnecessary external frontend. The governing reason is: ClusterIP is intended for internal service discovery; LoadBalancer requests external integration where supported.

C:

A headless Service (clusterIP None) publishes endpoint records rather than a normal ClusterIP proxy target. That is relevant to a different intervention. For the stated situation, the required action is: Return to ClusterIP and remove the unnecessary external frontend. The governing reason is: ClusterIP is intended for internal service discovery; LoadBalancer requests external integration where supported.

D:

Selector-backed Services route through EndpointSlices populated from matching eligible Pods. That is relevant to a different intervention. For the stated situation, the required action is: Return to ClusterIP and remove the unnecessary external frontend. The governing reason is: ClusterIP is intended for internal service discovery; LoadBalancer requests external integration where supported.

Question 55

A node is overrun by image garbage collection activity during every deployment wave. An engineer recommends: Inspect image filesystem consumption and kubelet garbage collection thresholds before changing policies. Which Kubernetes behavior best explains why this correction fits?

  1. CPU limits may cause throttling independent of overall node idle time.
  2. Node memory usage includes multiple consumers and cannot be attributed to one Pod without evidence.
  3. Node-wide utilization can hide a throttled or memory-constrained application container.
  4. Image storage pressure and image garbage collection are distinct from application CPU demand.

Explanation

Correct Answer

D:

Image storage pressure and image garbage collection are distinct from application CPU demand. This explains why the proposed action—inspect image filesystem consumption and kubelet garbage collection thresholds before changing policies.—is appropriate. Direct confirmation is: Image filesystem usage stabilizes and Pod starts remain timely during controlled releases.

Incorrect Answers

A:

CPU limits may cause throttling independent of overall node idle time. This does not establish the proposed intervention. The governing principle here is: Image storage pressure and image garbage collection are distinct from application CPU demand. Apply it by: Inspect image filesystem consumption and kubelet garbage collection thresholds before changing policies.

B:

Node memory usage includes multiple consumers and cannot be attributed to one Pod without evidence. This does not establish the proposed intervention. The governing principle here is: Image storage pressure and image garbage collection are distinct from application CPU demand. Apply it by: Inspect image filesystem consumption and kubelet garbage collection thresholds before changing policies.

C:

Node-wide utilization can hide a throttled or memory-constrained application container. This does not establish the proposed intervention. The governing principle here is: Image storage pressure and image garbage collection are distinct from application CPU demand. Apply it by: Inspect image filesystem consumption and kubelet garbage collection thresholds before changing policies.

Question 56

A GPU training Pod remains Pending while GPUs exist on nodes carrying a dedicated accelerator label. Which Kubernetes administration action most directly addresses the observed failure?

  1. Correct the label selector or intentionally seed the matching workload after validating the placement requirement.
  2. Replace preferred affinity with a required node affinity or other enforceable placement policy.
  3. Add narrowly scoped matching tolerations and appropriate node placement constraints to the authorized Pods.
  4. Set a nodeSelector or required node affinity matching the available accelerator nodes.

Explanation

Correct Answer

D:

Set a nodeSelector or required node affinity matching the available accelerator nodes. This corrects the specific failure because the scheduler matches Pod node constraints to node labels and available resources. Verify it by checking: The Pod schedules onto a matching GPU node and a nonmatching node is excluded.

Incorrect Answers

A:

Required pod affinity is a hard predicate and cannot match nonexistent target Pods. That is relevant to a different intervention. For the stated situation, the required action is: Set a nodeSelector or required node affinity matching the available accelerator nodes. The governing reason is: The scheduler matches Pod node constraints to node labels and available resources.

B:

Preferred affinity influences scoring but is not a hard scheduling constraint. That is relevant to a different intervention. For the stated situation, the required action is: Set a nodeSelector or required node affinity matching the available accelerator nodes. The governing reason is: The scheduler matches Pod node constraints to node labels and available resources.

C:

Tolerations permit scheduling onto tainted nodes but do not by themselves select those nodes. That is relevant to a different intervention. For the stated situation, the required action is: Set a nodeSelector or required node affinity matching the available accelerator nodes. The governing reason is: The scheduler matches Pod node constraints to node labels and available resources.

Question 57

A payment Pod can reach internal Services but egress to an external approved API is denied after a policy update. An engineer recommends: Add an egress rule that permits the required destination range and port while preserving other restrictions. Which Kubernetes behavior best explains why this correction fits?

  1. Ingress and egress restrictions are evaluated independently and policy effects compose additively.
  2. DNS name resolution is a separate network flow from the application connection.
  3. Kubernetes accepts policy objects even when a CNI does not implement enforcement.
  4. An egress-isolated Pod needs an explicit permitted path to each required destination.

Explanation

Correct Answer

D:

An egress-isolated Pod needs an explicit permitted path to each required destination. This explains why the proposed action—add an egress rule that permits the required destination range and port while preserving other restrictions.—is appropriate. Direct confirmation is: The approved API endpoint is reachable and a disallowed destination remains blocked.

Incorrect Answers

A:

Ingress and egress restrictions are evaluated independently and policy effects compose additively. This does not establish the proposed intervention. The governing principle here is: An egress-isolated Pod needs an explicit permitted path to each required destination. Apply it by: Add an egress rule that permits the required destination range and port while preserving other restrictions.

B:

DNS name resolution is a separate network flow from the application connection. This does not establish the proposed intervention. The governing principle here is: An egress-isolated Pod needs an explicit permitted path to each required destination. Apply it by: Add an egress rule that permits the required destination range and port while preserving other restrictions.

C:

Kubernetes accepts policy objects even when a CNI does not implement enforcement. This does not establish the proposed intervention. The governing principle here is: An egress-isolated Pod needs an explicit permitted path to each required destination. Apply it by: Add an egress rule that permits the required destination range and port while preserving other restrictions.

Question 58

Two Pods on separate workers cannot communicate using Pod IPs even though Services in their own namespaces work locally. Following the proposed change—Inspect the CNI node agents, routes, and encapsulation between worker Pod CIDRs—which observation best demonstrates that the particular fault is resolved?

  1. A test client connects over each supported IP family to healthy backends.
  2. Clients continue connecting after the target Pod is deleted and replaced.
  3. The client resolves the current Service record and reaches a ready backend.
  4. A test client connects directly to a Pod IP on another node and network counters show the expected path.

Explanation

Correct Answer

D:

A test client connects directly to a Pod IP on another node and network counters show the expected path. This validates the stated intervention—inspect the CNI node agents, routes, and encapsulation between worker Pod CIDRs.—rather than an unrelated cluster change. The relevant mechanism is: Kubernetes expects Pod-to-Pod reachability without application-managed NAT across nodes.

Incorrect Answers

A:

Dual-stack Services and Pods may expose more than one IP family depending on cluster configuration. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A test client connects directly to a Pod IP on another node and network counters show the expected path.

B:

Pods are replaceable and their IP addresses can change even when the application identity remains. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A test client connects directly to a Pod IP on another node and network counters show the expected path.

C:

Kubernetes Service DNS gives workloads a stable logical endpoint independent of Pod replacement. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A test client connects directly to a Pod IP on another node and network counters show the expected path.

Question 59

A controller-manager static Pod crashes after an administrator edits its command flags in place. What is the most direct consequence if the underlying problem is left unresolved?

  1. Cluster control functions remain unavailable while the API server fails.
  2. Control loops can stall while the process itself remains Running.
  3. An administrator may delete the wrong object while the underlying static Pod stays broken.
  4. Resources may stop reconciling even when direct API reads still work.

Explanation

Correct Answer

D:

Resources may stop reconciling even when direct API reads still work. This directly follows from the failed condition. The preventive correction is: Restore valid flags in the kube-controller-manager manifest and check kubelet recreation logs. The governing Kubernetes behavior is: The kubelet restarts static control-plane Pods when their manifest changes.

Incorrect Answers

A:

kubeadm control-plane components often run as static Pods from local manifests. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Resources may stop reconciling even when direct API reads still work.

B:

Control-plane components need functional credentials and narrowly appropriate permissions to reconcile state. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Resources may stop reconciling even when direct API reads still work.

C:

Mirror Pods represent static Pods in the API but are not their source of truth on the node. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Resources may stop reconciling even when direct API reads still work.

Question 60

A multi-container Pod logs an error but a responder reads only the default container and misses the sidecar failures. An engineer recommends: Specify the container name when retrieving Pod logs and compare related containers' output. Which Kubernetes behavior best explains why this correction fits?

  1. Restarted containers can have a previous log stream that is distinct from the current one.
  2. Container logs belong to individual containers even within one Pod.
  3. kubectl logs primarily reads the container runtime's captured stdout and stderr streams.
  4. Containers that never start may have no runtime logs; scheduling and image events carry the relevant diagnosis.

Explanation

Correct Answer

B:

Container logs belong to individual containers even within one Pod. This explains why the proposed action—specify the container name when retrieving Pod logs and compare related containers' output.—is appropriate. Direct confirmation is: Logs from the failing sidecar reveal the error while the app container's normal output is distinguished.

Incorrect Answers

A:

Restarted containers can have a previous log stream that is distinct from the current one. This does not establish the proposed intervention. The governing principle here is: Container logs belong to individual containers even within one Pod. Apply it by: Specify the container name when retrieving Pod logs and compare related containers' output.

C:

kubectl logs primarily reads the container runtime's captured stdout and stderr streams. This does not establish the proposed intervention. The governing principle here is: Container logs belong to individual containers even within one Pod. Apply it by: Specify the container name when retrieving Pod logs and compare related containers' output.

D:

Containers that never start may have no runtime logs; scheduling and image events carry the relevant diagnosis. This does not establish the proposed intervention. The governing principle here is: Container logs belong to individual containers even within one Pod. Apply it by: Specify the container name when retrieving Pod logs and compare related containers' output.