Certified Kubernetes Administrator (CKA) - Concept Practice Exam 1

1. A container repeatedly exits and restarts. Its current log is nearly empty, but the previous instance emitted the failure. Which command retrieves that instance's log?
2. A container's last termination reason is OOMKilled, and its memory usage exceeded its configured memory limit. Which response directly addresses that resource failure?
3. A control plane uses three healthy voting etcd members. One member fails while the other two can communicate. What should the administrator expect from etcd quorum?
4. A Deployment has enough spare cluster capacity. During an update, the team requires all existing replicas to remain available while a replacement Pod becomes ready. Which rolling-update settings fit this requirement?
5. A Gateway controller and a compatible Gateway are installed. A team wants host- and path-based HTTP routing through that Gateway. Which resource normally attaches those HTTP routing rules?
6. A kubeadm-managed control plane reports errors consistent with expired certificates. Which command directly lists the expiration of certificates managed locally by kubeadm?
7. A manually managed PersistentVolume uses persistentVolumeReclaimPolicy: Retain. Its claim is deleted. What should the administrator expect?
8. A new cluster has a working container runtime, but newly created Pods cannot obtain the expected Pod network connectivity. Which integration is responsible for configuring Pod networking?
9. A node has the taint dedicated=batch:NoSchedule. A Pod has a matching toleration but no node selector or affinity. Which statement is correct?
10. A Pending Pod requests 4 CPU cores. Each eligible node has only 2 unallocated allocatable CPU cores, although current CPU usage is low. Why can scheduling still fail?
11. A platform team installed a controller as a Helm release. It wants to change a supported chart setting while keeping the release managed by Helm. Which approach is appropriate?
12. A Pod in namespace reports has ImagePullBackOff because its private registry requires authentication. An appropriate registry Secret exists only in default. What should be changed?
13. A policy should allow ingress only from Pods labeled app=frontend inside namespaces labeled team=shop. Which peer structure requires both conditions?
14. A release service account must read Deployments only in namespace staging. Which authorization setup grants the required access without granting it in other namespaces?
15. A Service exposes port 80 with targetPort 8080. Its selected, ready Pods listen only on port 8000, and direct connections to Pod port 8000 succeed. What should be corrected?
16. A Service selects app=payments, but the intended Pods are labeled app=payment. The Pods are healthy. What is the most direct configuration correction?
17. A stateful client needs DNS discovery of individual backend Pod addresses rather than a virtual Service IP. Which Service configuration provides headless discovery?
18. A supported CSI volume must be writable by only one Pod across the cluster. Which access mode directly expresses that requirement?
19. A team creates an Ingress object with valid HTTP rules, but its cluster has no Ingress controller. What is missing for those rules to route traffic?
20. A team installs a custom Database resource definition and expects creating a Database object to deploy and maintain a database. What additional component normally supplies that reconciliation behavior?
21. A topology-constrained storage system must provision a volume in a location compatible with the Pod that will use it. Which StorageClass binding mode delays provisioning until scheduling information is available?
22. A worker becomes NotReady after a local configuration change. The API server is reachable from other nodes. Which node-level evidence should the administrator inspect first?
23. An administrator is designing a kubeadm cluster with several control-plane nodes. How should clients reach a control plane that remains accessible when one API server fails?
24. An analytics Pod must run only on nodes labeled disk=ssd. Other nodes have sufficient resources but no such label. Which rule expresses a mandatory placement constraint?
25. An HPA uses average CPU utilization as a percentage of requested CPU. The application's containers have no CPU requests. What should be configured so this utilization target can be calculated normally?
26. Development and production share Kubernetes base manifests but require different replica counts. The team wants to avoid maintaining copied manifests. Which Kustomize structure fits?
27. Draining a node is blocked by a PodDisruptionBudget. Its application has two healthy replicas and minAvailable: 2. What is the appropriate explanation?
28. On a kubeadm control-plane node, the scheduler runs as a static Pod. Its manifest is configured through the kubelet's staticPodPath. Where should an administrator make a persistent change to that Pod's configuration?
29. Pods can connect to a Service's cluster IP, but Service DNS names fail across namespaces. Which investigation most directly targets the failing layer?
30. The cluster DNS domain is cluster.local. A Pod in namespace web must resolve Service inventory in namespace backend. Which fully qualified name identifies that Service?

Comments — 0