Question 600A web application must update one Pod at a time while retaining capacity during a routine release. Following the proposed change—Configure rolling-update maxUnavailable and maxSurge according to available capacity and readiness requirements—which observation best demonstrates that the particular fault is resolved? kubectl rollout status shows the prior revision fully available and application probes succeed. The desired replica count is ready under the new image and old ReplicaSet replicas are reduced. During a controlled rollout, the configured availability floor is maintained while new Pods become Ready. New Pods use the updated Secret value and old processes terminate in a controlled rollout. ExplanationCorrect Answer C:During a controlled rollout, the configured availability floor is maintained while new Pods become Ready. This validates the stated intervention—configure rolling-update maxUnavailable and maxSurge according to available capacity and readiness requirements.—rather than an unrelated cluster change. The relevant mechanism is: The Deployment rollout strategy controls the allowed temporary excess and unavailable replicas.Incorrect Answers A:Deployment revisions track supported rollout history and a rollback restores a previous Pod template. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: During a controlled rollout, the configured availability floor is maintained while new Pods become Ready. B:Desired-state changes are asynchronous and a successful API update is not proof of healthy Pods. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: During a controlled rollout, the configured availability floor is maintained while new Pods become Ready. D:Environment variables sourced from Secrets are fixed for the lifetime of a container process. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: During a controlled rollout, the configured availability floor is maintained while new Pods become Ready.
Question 601An administrator upgrades worker kubelets before evaluating the supported kubeadm control-plane version sequence. What is the most direct consequence if the underlying problem is left unresolved? A bad upgrade may leave no reliable recovery point for cluster objects. Upgrade work may stop partway through or leave incompatible components. Unsupported version skew can create API and scheduling failures. Forcing eviction can violate the workload's minimum availability. ExplanationCorrect Answer C:Unsupported version skew can create API and scheduling failures. This directly follows from the failed condition. The preventive correction is: Follow Kubernetes version-skew rules and upgrade the control plane and workers in the supported order. The governing Kubernetes behavior is: Component version skew is constrained; a newer kubelet must not be assumed compatible with an older API server.Incorrect Answers A:Cluster state durability requires more than rerunning kubeadm; an etcd snapshot protects API objects. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Unsupported version skew can create API and scheduling failures. B:kubeadm upgrade plan surfaces supported upgrade paths and potential blockers, not just an install command. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Unsupported version skew can create API and scheduling failures. D:The eviction API respects PodDisruptionBudget limits for voluntary disruption. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Unsupported version skew can create API and scheduling failures.
Question 602A workload on a small node suffers severe CPU throttling despite reported CPU usage below node capacity. Following the proposed change—Compare configured CPU limits and container throttling metrics before changing node size—which observation best demonstrates that the particular fault is resolved? The specific bottlenecking Pod is identified and its latency improves after targeted resource changes. Adjusted requests or added capacity make the Pods schedulable without overcommit beyond policy. Controlled peak tests show stable replicas without OOMKilled and dashboards show reduced restarts. Throttle counters decline and application latency improves after a justified limit adjustment. ExplanationCorrect Answer D:Throttle counters decline and application latency improves after a justified limit adjustment. This validates the stated intervention—compare configured CPU limits and container throttling metrics before changing node size.—rather than an unrelated cluster change. The relevant mechanism is: CPU limits may cause throttling independent of overall node idle time.Incorrect Answers A:Node-wide utilization can hide a throttled or memory-constrained application container. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Throttle counters decline and application latency improves after a justified limit adjustment. B:Scheduling uses resource requests and constraints, not recent utilization percentages alone. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Throttle counters decline and application latency improves after a justified limit adjustment. C:OOMKilled is evidence of memory-limit or node memory pressure, not an ordinary successful exit. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Throttle counters decline and application latency improves after a justified limit adjustment.
Question 603A Service forwards port 80 to an application listening on container port 8080, but targetPort was set to 8081. Following the proposed change—Set the Service targetPort to the actual listener port or correctly named container port—which observation best demonstrates that the particular fault is resolved? A request to Service port 80 reaches the Pod's 8080 listener. The Service receives the expected load balancer ingress and end-to-end external probes succeed. DNS lookups return the intended Pod endpoint addresses and stateful clients connect to them. The Service resolves to valid managed endpoints and a test client reaches its backing process. ExplanationCorrect Answer A:A request to Service port 80 reaches the Pod's 8080 listener. This validates the stated intervention—set the Service targetPort to the actual listener port or correctly named container port.—rather than an unrelated cluster change. The relevant mechanism is: Service port and targetPort describe different ends of the virtual forwarding path.Incorrect Answers B:NodePort exposes a port on nodes, while LoadBalancer depends on provider or local controller provisioning. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A request to Service port 80 reaches the Pod's 8080 listener. C:A headless Service (clusterIP None) publishes endpoint records rather than a normal ClusterIP proxy target. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A request to Service port 80 reaches the Pod's 8080 listener. D:Selectorless Services require explicitly associated endpoints because no controller infers backing Pods. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A request to Service port 80 reaches the Pod's 8080 listener.
Question 604An API team assumes a Pod's IP remains stable across rescheduling and hardcodes it into client configuration. Following the proposed change—Use a Service DNS name or other stable discovery mechanism instead of a Pod IP—which observation best demonstrates that the particular fault is resolved? The client resolves the current Service record and reaches a ready backend. A test client connects over each supported IP family to healthy backends. External probes reach the published port through each intended node address. Clients continue connecting after the target Pod is deleted and replaced. ExplanationCorrect Answer D:Clients continue connecting after the target Pod is deleted and replaced. This validates the stated intervention—use a Service DNS name or other stable discovery mechanism instead of a Pod IP.—rather than an unrelated cluster change. The relevant mechanism is: Pods are replaceable and their IP addresses can change even when the application identity remains.Incorrect Answers A: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: Clients continue connecting after the target Pod is deleted and replaced. B: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: Clients continue connecting after the target Pod is deleted and replaced. C:NodePort connectivity depends on node networking and service-proxy dataplane behavior, not just a valid Service object. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Clients continue connecting after the target Pod is deleted and replaced.
Question 605A controller installed by an operator enters a reconciliation loop because its ServiceAccount cannot update the custom resource status subresource. Following the proposed change—Grant the controller the minimum required update or patch permission for the status subresource—which observation best demonstrates that the particular fault is resolved? Controller events show successful status reconciliation without allowing unwanted spec modification. The operator returns healthy and the managed Deployment appears with the expected owner references. Impersonated edits fail outside the permitted namespace while intended local access remains. Existing custom objects are readable and writable through the supported new version after migration. ExplanationCorrect Answer A:Controller events show successful status reconciliation without allowing unwanted spec modification. This validates the stated intervention—grant the controller the minimum required update or patch permission for the status subresource.—rather than an unrelated cluster change. The relevant mechanism is: Custom resource status updates are separately authorized from spec updates when status is enabled.Incorrect Answers B:A CRD defines the API schema while a running controller implements the desired behavior. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Controller events show successful status reconciliation without allowing unwanted spec modification. C:New custom resource types participate in authorization using their API group and scope. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Controller events show successful status reconciliation without allowing unwanted spec modification. D:CRD versioning separates served API versions from the storage version and may require conversion webhooks. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: Controller events show successful status reconciliation without allowing unwanted spec modification.
Question 606A Gateway route splits traffic between two backend Services but weight settings send most requests to an untested canary. Which Kubernetes administration action most directly addresses the observed failure? Change parentRefs sectionName or create the intended listener and confirm route attachment status. Configure the listener allowedRoutes namespaces policy or appropriate cross-namespace reference permissions. Adjust backendRefs weights and validate rollout percentages before production promotion. Correct the Service selector to match the intended ready Pod labels. ExplanationCorrect Answer C:Adjust backendRefs weights and validate rollout percentages before production promotion. This corrects the specific failure because gateway API backend weights influence proportional HTTPRoute forwarding among valid backends. Verify it by checking: Observed test traffic approximates approved weights and both backends report ready.Incorrect Answers A:Gateway API HTTPRoutes must attach to compatible allowed listeners through parent references. That is relevant to a different intervention. For the stated situation, the required action is: Adjust backendRefs weights and validate rollout percentages before production promotion. The governing reason is: Gateway API backend weights influence proportional HTTPRoute forwarding among valid backends. B:Gateway API constrains route attachment and cross-namespace references through explicit policies. That is relevant to a different intervention. For the stated situation, the required action is: Adjust backendRefs weights and validate rollout percentages before production promotion. The governing reason is: Gateway API backend weights influence proportional HTTPRoute forwarding among valid backends. 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: Adjust backendRefs weights and validate rollout percentages before production promotion. The governing reason is: Gateway API backend weights influence proportional HTTPRoute forwarding among valid backends.
Question 607After restoring an etcd snapshot, an engineer starts two inconsistent members with mismatched cluster identifiers. What is the most direct consequence if the underlying problem is left unresolved? A supposedly redundant control plane can provide less resilience than planned. Read-only-looking API availability can conceal inability to persist state. A partial restore can split or permanently block cluster state recovery. Control loops can stall while the process itself remains Running. ExplanationCorrect Answer C:A partial restore can split or permanently block cluster state recovery. This directly follows from the failed condition. The preventive correction is: Follow the supported etcd snapshot restore topology and consistent member configuration. The governing Kubernetes behavior is: Restored etcd members must agree on cluster identity and membership rather than mix prior data directories.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: A partial restore can split or permanently block cluster state recovery. B:Etcd requires a majority of voters to commit changes; an API listener alone does not imply healthy storage. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: A partial restore can split or permanently block cluster state recovery. D: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: A partial restore can split or permanently block cluster state recovery.
Question 608A team creates a NetworkPolicy resource but its chosen CNI has no policy enforcement feature enabled. Following the proposed change—Deploy or enable a NetworkPolicy-capable CNI dataplane and test actual packet enforcement—which observation best demonstrates that the particular fault is resolved? Unauthorized ingress and egress probes fail while necessary bidirectional flows pass. Approved frontend traffic succeeds while an unrelated test Pod remains denied. A disallowed Pod connection is denied on the network, not merely recorded as a policy object. The Pod resolves a Service name and reaches it while unrelated outbound access remains denied. ExplanationCorrect Answer C:A disallowed Pod connection is denied on the network, not merely recorded as a policy object. This validates the stated intervention—deploy or enable a NetworkPolicy-capable CNI dataplane and test actual packet enforcement.—rather than an unrelated cluster change. The relevant mechanism is: Kubernetes accepts policy objects even when a CNI does not implement enforcement.Incorrect Answers A: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: A disallowed Pod connection is denied on the network, not merely recorded as a policy object. B:NetworkPolicies are additive for selected Pods; allow rules define permitted flows after isolation. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A disallowed Pod connection is denied on the network, not merely recorded as a policy object. D:DNS name resolution is a separate network flow from the application connection. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A disallowed Pod connection is denied on the network, not merely recorded as a policy object.
Question 609A newly added control-plane node runs etcd but has not joined the expected quorum member set. Following the proposed change—Use supported etcd member verification and safely add or reconcile the intended member—which observation best demonstrates that the particular fault is resolved? Members form a healthy quorum and expected Kubernetes objects are visible before traffic is restored. etcd endpoint health and a safe API create or update operation both succeed. Worker kubelet connections and administrator requests succeed using current endpoints. The membership list includes the new peer and quorum health remains good after failure testing. ExplanationCorrect Answer D:The membership list includes the new peer and quorum health remains good after failure testing. This validates the stated intervention—use supported etcd member verification and safely add or reconcile the intended member.—rather than an unrelated cluster change. The relevant mechanism is: A running process is not necessarily a healthy voting member of the cluster's configured etcd topology.Incorrect Answers A:Restored etcd members must agree on cluster identity and membership rather than mix prior data directories. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The membership list includes the new peer and quorum health remains good after failure testing. B:Etcd requires a majority of voters to commit changes; an API listener alone does not imply healthy storage. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The membership list includes the new peer and quorum health remains good after failure testing. C:Control-plane advertised addresses must be reachable through the cluster's intended topology. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The membership list includes the new peer and quorum health remains good after failure testing.
Question 610A chart upgrade unexpectedly changes several resources when the operator only intends a Service annotation. Following the proposed change—Render and diff the proposed Helm release before applying a controlled upgrade—which observation best demonstrates that the particular fault is resolved? helm get values and rendered manifests show the change, which survives the next upgrade. The Helm release is absent and remaining CRDs or cluster-scoped resources are explicitly reviewed. The reviewed rendered difference contains only intended resource changes before promotion. The upgrade plan lists the expected target and preflight checks pass before apply. ExplanationCorrect Answer C:The reviewed rendered difference contains only intended resource changes before promotion. This validates the stated intervention—render and diff the proposed Helm release before applying a controlled upgrade.—rather than an unrelated cluster change. The relevant mechanism is: Template rendering plus a values diff exposes chart-side effects before they affect cluster resources.Incorrect Answers A:Helm's release history and desired manifests should remain the authoritative source for chart-managed resources. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The reviewed rendered difference contains only intended resource changes before promotion. B:Helm tracks release objects, while some custom resources and cluster-scoped artifacts can outlive a namespace. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The reviewed rendered difference contains only intended resource changes before promotion. D:kubeadm upgrade plan surfaces supported upgrade paths and potential blockers, not just an install command. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The reviewed rendered difference contains only intended resource changes before promotion.