Question 90A Pod requests 8 GiB of memory but each suitable node has only 4 GiB allocatable. Following the proposed change—Resize requests or provide sufficient eligible nodes based on actual allocatable capacity—which observation best demonstrates that the particular fault is resolved? Events no longer report Insufficient memory and the Pod schedules after capacity is corrected. The Pod schedules onto a matching GPU node and a nonmatching node is excluded. A matching eligible Pod exists in the specified topology and the new Pod schedules. Only intended workloads land on dedicated nodes while unapproved Pods remain excluded. ExplanationCorrect Answer A:Events no longer report Insufficient memory and the Pod schedules after capacity is corrected. This validates the stated intervention—resize requests or provide sufficient eligible nodes based on actual allocatable capacity.—rather than an unrelated cluster change. The relevant mechanism is: The scheduler evaluates Pod requests against node allocatable resources, not observed averages alone.Incorrect Answers B:The scheduler matches Pod node constraints to node labels and available resources. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Events no longer report Insufficient memory and the Pod schedules after capacity is corrected. C:Required pod affinity is a hard predicate and cannot match nonexistent target Pods. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Events no longer report Insufficient memory and the Pod schedules after capacity is corrected. D:Tolerations permit scheduling onto tainted nodes but do not by themselves select those nodes. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Events no longer report Insufficient memory and the Pod schedules after capacity is corrected.
Question 91A bare-metal worker can contact the API but cannot pull images because the container runtime is not configured correctly. Following the proposed change—Configure a compatible CRI runtime and its cgroup driver consistently with the kubelet—which observation best demonstrates that the particular fault is resolved? The kubelet starts without swap-related errors and the node becomes Ready. The endpoint resolves to an active listener and kubeadm discovery succeeds from a worker. crictl can query the runtime endpoint and a test Pod starts successfully. A route review and trial Pod-to-Service connection show unique destination networks. ExplanationCorrect Answer C:crictl can query the runtime endpoint and a test Pod starts successfully. This validates the stated intervention—configure a compatible CRI runtime and its cgroup driver consistently with the kubelet.—rather than an unrelated cluster change. The relevant mechanism is: The kubelet delegates container creation to a CRI-compatible runtime and requires compatible host settings.Incorrect Answers A:Node prerequisites must be consistent with kubelet behavior and runtime support. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: crictl can query the runtime endpoint and a test Pod starts successfully. B:Clients and join operations need a durable API endpoint independent of one control-plane host. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: crictl can query the runtime endpoint and a test Pod starts successfully. D:Conflicting cluster and external address spaces can misroute traffic even when the API server is healthy. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: crictl can query the runtime endpoint and a test Pod starts successfully.
Question 92A team wants to clone a PVC into a test namespace but assumes a PVC dataSource can reference claims across namespaces automatically. Following the proposed change—Use supported same-namespace clone capabilities or a separate approved snapshot/restore transfer workflow—which observation best demonstrates that the particular fault is resolved? The Pod's namespace contains the referenced claim and volume mounts successfully. The cloned claim contains the expected data under the supported source and namespace rules. The PV is Available or intentionally bound to the new approved PVC without losing data. The storage driver reports successful attachment to the replacement node and the Pod mounts correctly. ExplanationCorrect Answer B:The cloned claim contains the expected data under the supported source and namespace rules. This validates the stated intervention—use supported same-namespace clone capabilities or a separate approved snapshot/restore transfer workflow.—rather than an unrelated cluster change. The relevant mechanism is: PVC cloning and dataSource references have scope and driver capability constraints.Incorrect Answers A:PVCs are namespace-scoped and Pods cannot mount another namespace's claim by name. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The cloned claim contains the expected data under the supported source and namespace rules. C:A Released PV may preserve its previous claim reference until an administrator makes it available. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The cloned claim contains the expected data under the supported source and namespace rules. D:A Bound PVC proves allocation but not that attach, stage and publish operations have succeeded. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The cloned claim contains the expected data under the supported source and namespace rules.
Question 93A nightly maintenance action is managed through a manually recreated Pod and frequently misses its schedule. An engineer recommends: Create a CronJob with an approved schedule, concurrencyPolicy and missed-run behavior. Which Kubernetes behavior best explains why this correction fits? Jobs manage successful completion of finite work; Deployments maintain continuous desired replicas. Deployment revisions track supported rollout history and a rollback restores a previous Pod template. The Deployment rollout strategy controls the allowed temporary excess and unavailable replicas. CronJobs create Jobs according to schedule while controlling overlap and missed executions. ExplanationCorrect Answer D:CronJobs create Jobs according to schedule while controlling overlap and missed executions. This explains why the proposed action—create a CronJob with an approved schedule, concurrencyPolicy and missed-run behavior.—is appropriate. Direct confirmation is: Scheduled Jobs appear at expected times and overlapping runs follow the configured policy.Incorrect Answers A:Jobs manage successful completion of finite work; Deployments maintain continuous desired replicas. This does not establish the proposed intervention. The governing principle here is: CronJobs create Jobs according to schedule while controlling overlap and missed executions. Apply it by: Create a CronJob with an approved schedule, concurrencyPolicy and missed-run behavior. B:Deployment revisions track supported rollout history and a rollback restores a previous Pod template. This does not establish the proposed intervention. The governing principle here is: CronJobs create Jobs according to schedule while controlling overlap and missed executions. Apply it by: Create a CronJob with an approved schedule, concurrencyPolicy and missed-run behavior. C:The Deployment rollout strategy controls the allowed temporary excess and unavailable replicas. This does not establish the proposed intervention. The governing principle here is: CronJobs create Jobs according to schedule while controlling overlap and missed executions. Apply it by: Create a CronJob with an approved schedule, concurrencyPolicy and missed-run behavior.
Question 94A new HA control-plane node receives a certificate that omits the load-balanced API hostname from its SANs. Following the proposed change—Issue certificates that include the approved stable API endpoint and node identities—which observation best demonstrates that the particular fault is resolved? A TLS connection to the load-balanced hostname succeeds without insecure verification overrides. A simulated single-control-plane outage retains API availability and healthy etcd quorum. Traffic drains away from an unready instance and remains on ready control-plane backends. An entire host outage leaves a majority of etcd voters and API access available. ExplanationCorrect Answer A:A TLS connection to the load-balanced hostname succeeds without insecure verification overrides. This validates the stated intervention—issue certificates that include the approved stable API endpoint and node identities.—rather than an unrelated cluster change. The relevant mechanism is: TLS clients verify the endpoint hostname against the presented certificate SANs.Incorrect Answers B:Highly available control planes depend on load-balanced API access and durable quorum-backed state. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A TLS connection to the load-balanced hostname succeeds without insecure verification overrides. C:A successful TCP handshake does not prove a particular API server is ready to serve requests. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A TLS connection to the load-balanced hostname succeeds without insecure verification overrides. D:Quorum and replication do not protect against a common power or hypervisor failure domain. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A TLS connection to the load-balanced hostname succeeds without insecure verification overrides.
Question 95The API server is running but writes fail after etcd loses quorum during maintenance. What is the most direct consequence if the underlying problem is left unresolved? A supposedly redundant control plane can provide less resilience than planned. A partial restore can split or permanently block cluster state recovery. Read-only-looking API availability can conceal inability to persist state. Cluster control functions remain unavailable while the API server fails. ExplanationCorrect Answer C:Read-only-looking API availability can conceal inability to persist state. This directly follows from the failed condition. The preventive correction is: Inspect etcd membership and quorum health, then restore sufficient approved voting members. The governing Kubernetes behavior is: Etcd requires a majority of voters to commit changes; an API listener alone does not imply healthy storage.Incorrect Answers A:A running process is not necessarily a healthy voting member of the cluster's configured etcd topology. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Read-only-looking API availability can conceal inability to persist state. B:Restored etcd members must agree on cluster identity and membership rather than mix prior data directories. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Read-only-looking API availability can conceal inability to persist state. D: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: Read-only-looking API availability can conceal inability to persist state.
Question 96CPU usage stays high but a web Deployment with three replicas never scales after installing the metrics server. What is the most direct consequence if the underlying problem is left unresolved? Insufficient replicas can cause latency and request failures. Autoscaling can remain ineffective despite observed load spikes. Backlogs can grow while autoscaling appears healthy on irrelevant CPU metrics. Overprovisioning or unstable replica churn can persist without conscious policy. ExplanationCorrect Answer A:Insufficient replicas can cause latency and request failures. This directly follows from the failed condition. The preventive correction is: Create a CPU-based HPA with matching resource requests and confirm metrics availability. The governing Kubernetes behavior is: HPA CPU utilization calculations depend on resource requests and relevant metrics API data.Incorrect Answers B:CPU percentage targets use utilization relative to requested CPU, not just absolute node usage. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Insufficient replicas can cause latency and request failures. C:Different workload bottlenecks require appropriately supported HPA metrics. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Insufficient replicas can cause latency and request failures. D:HPA may delay downscaling to avoid flapping even after metrics fall. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Insufficient replicas can cause latency and request failures.
Question 97A new HA control-plane node receives a certificate that omits the load-balanced API hostname from its SANs. Which Kubernetes administration action most directly addresses the observed failure? Design and test multiple control-plane nodes behind a durable API endpoint with compatible etcd quorum. Issue certificates that include the approved stable API endpoint and node identities. Configure a readiness-aware control-plane health check compatible with the API service. Distribute control-plane and etcd members across independent failure domains. ExplanationCorrect Answer B:Issue certificates that include the approved stable API endpoint and node identities. This corrects the specific failure because tLS clients verify the endpoint hostname against the presented certificate SANs. Verify it by checking: A TLS connection to the load-balanced hostname succeeds without insecure verification overrides.Incorrect Answers A:Highly available control planes depend on load-balanced API access and durable quorum-backed state. That is relevant to a different intervention. For the stated situation, the required action is: Issue certificates that include the approved stable API endpoint and node identities. The governing reason is: TLS clients verify the endpoint hostname against the presented certificate SANs. C:A successful TCP handshake does not prove a particular API server is ready to serve requests. That is relevant to a different intervention. For the stated situation, the required action is: Issue certificates that include the approved stable API endpoint and node identities. The governing reason is: TLS clients verify the endpoint hostname against the presented certificate SANs. D:Quorum and replication do not protect against a common power or hypervisor failure domain. That is relevant to a different intervention. For the stated situation, the required action is: Issue certificates that include the approved stable API endpoint and node identities. The governing reason is: TLS clients verify the endpoint hostname against the presented certificate SANs.
Question 98An administrator drains a node but DaemonSet Pods prevent a naive drain command from completing. Following the proposed change—Use drain with the documented DaemonSet handling and respect eviction and disruption constraints—which observation best demonstrates that the particular fault is resolved? The cluster lists one active intended Node identity and its Lease and Pods reference the replacement correctly. Evictable Pods relocate while the DaemonSet is handled according to the maintenance plan. The relevant Pods schedule onto approved labeled nodes and node inventory matches policy. Node DiskPressure clears and new Pods are admitted without deleting active volume data. ExplanationCorrect Answer B:Evictable Pods relocate while the DaemonSet is handled according to the maintenance plan. This validates the stated intervention—use drain with the documented DaemonSet handling and respect eviction and disruption constraints.—rather than an unrelated cluster change. The relevant mechanism is: DaemonSet Pods are managed for node coverage and are not treated like ordinary replicated workloads during drain.Incorrect Answers A:Node names are identity keys in the API and stale entries can confuse lifecycle operations. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Evictable Pods relocate while the DaemonSet is handled according to the maintenance plan. C:Hard placement rules require actual matching node labels at scheduling time. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Evictable Pods relocate while the DaemonSet is handled according to the maintenance plan. D:Kubelet disk-pressure signals can cause eviction and scheduling taints independent of CPU utilization. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Evictable Pods relocate while the DaemonSet is handled according to the maintenance plan.
Question 99A GPU training Pod remains Pending while GPUs exist on nodes carrying a dedicated accelerator label. Following the proposed change—Set a nodeSelector or required node affinity matching the available accelerator nodes—which observation best demonstrates that the particular fault is resolved? A matching eligible Pod exists in the specified topology and the new Pod schedules. Scheduling tests reject nodes that lack the required sovereignty label. The Pod schedules onto a matching GPU node and a nonmatching node is excluded. Only intended workloads land on dedicated nodes while unapproved Pods remain excluded. ExplanationCorrect Answer C:The Pod schedules onto a matching GPU node and a nonmatching node is excluded. This validates the stated intervention—set a nodeSelector or required node affinity matching the available accelerator nodes.—rather than an unrelated cluster change. The relevant mechanism is: The scheduler matches Pod node constraints to node labels and available resources.Incorrect Answers A:Required pod affinity is a hard predicate and cannot match nonexistent target Pods. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The Pod schedules onto a matching GPU node and a nonmatching node is excluded. B:Preferred affinity influences scoring but is not a hard scheduling constraint. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The Pod schedules onto a matching GPU node and a nonmatching node is excluded. D:Tolerations permit scheduling onto tainted nodes but do not by themselves select those nodes. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The Pod schedules onto a matching GPU node and a nonmatching node is excluded.
Question 100A namespace applies a default-deny ingress NetworkPolicy and legitimate frontend-to-backend requests begin timing out. Following the proposed change—Add a narrowly scoped ingress allow policy selecting the backend and trusted frontend Pods on the required port—which observation best demonstrates that the particular fault is resolved? Approved frontend traffic succeeds while an unrelated test Pod remains denied. Only the intended web client can connect to the protected backend. The approved API endpoint is reachable and a disallowed destination remains blocked. Unauthorized ingress and egress probes fail while necessary bidirectional flows pass. ExplanationCorrect Answer A:Approved frontend traffic succeeds while an unrelated test Pod remains denied. This validates the stated intervention—add a narrowly scoped ingress allow policy selecting the backend and trusted frontend Pods on the required port.—rather than an unrelated cluster change. The relevant mechanism is: NetworkPolicies are additive for selected Pods; allow rules define permitted flows after isolation.Incorrect Answers B:A NetworkPolicy's top-level podSelector selects protected Pods; ingress peers identify allowed sources. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Approved frontend traffic succeeds while an unrelated test Pod remains denied. C:An egress-isolated Pod needs an explicit permitted path to each required destination. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Approved frontend traffic succeeds while an unrelated test Pod remains denied. D:Ingress and egress restrictions are evaluated independently and policy effects compose additively. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Approved frontend traffic succeeds while an unrelated test Pod remains denied.