Question 310A security review finds a wildcard RBAC rule allowing all verbs on all API groups for a log collector. Following the proposed change—Replace wildcard rules with explicit get list watch permissions on the exact resources the collector reads—which observation best demonstrates that the particular fault is resolved? kubectl auth can-i list configmaps -n billing --as=<identity> changes to yes without granting delete rights. An impersonated service-account access test reads the permitted Secret and rejects unrelated Secrets. Impersonated requests reach the approved subresource and reject unrelated secret reads. kubectl auth can-i checks show read-only access on log inputs and denial of workload modification. ExplanationCorrect Answer D:kubectl auth can-i checks show read-only access on log inputs and denial of workload modification. This validates the stated intervention—replace wildcard rules with explicit get list watch permissions on the exact resources the collector reads.—rather than an unrelated cluster change. The relevant mechanism is: Least-privilege RBAC narrows blast radius by limiting the allowed verbs, resources, and API groups.Incorrect Answers A:Authentication establishes identity while authorization separately decides permitted API verbs and resources. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: kubectl auth can-i checks show read-only access on log inputs and denial of workload modification. B:Kubernetes service accounts do not automatically inherit administrator permissions from the Pod owner. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: kubectl auth can-i checks show read-only access on log inputs and denial of workload modification. C:RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: kubectl auth can-i checks show read-only access on log inputs and denial of workload modification.
Question 311Three stacked-etcd control-plane nodes are planned, but the team schedules all of them on one physical hypervisor. 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. Distribute control-plane and etcd members across independent failure domains. Inspect etcd endpoint status and membership before modifying the cluster's voter set. Issue certificates that include the approved stable API endpoint and node identities. ExplanationCorrect Answer B:Distribute control-plane and etcd members across independent failure domains. This corrects the specific failure because quorum and replication do not protect against a common power or hypervisor failure domain. Verify it by checking: An entire host outage leaves a majority of etcd voters and API access available.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: Distribute control-plane and etcd members across independent failure domains. The governing reason is: Quorum and replication do not protect against a common power or hypervisor failure domain. C:Etcd commits require a majority of voting members; removal and replacement must preserve quorum. That is relevant to a different intervention. For the stated situation, the required action is: Distribute control-plane and etcd members across independent failure domains. The governing reason is: Quorum and replication do not protect against a common power or hypervisor failure domain. D:TLS clients verify the endpoint hostname against the presented certificate SANs. That is relevant to a different intervention. For the stated situation, the required action is: Distribute control-plane and etcd members across independent failure domains. The governing reason is: Quorum and replication do not protect against a common power or hypervisor failure domain.
Question 312A webhook service's certificate expires and API object creation fails with an admission webhook TLS error. Following the proposed change—Renew the webhook serving certificate and align its caBundle and Service reference—which observation best demonstrates that the particular fault is resolved? EndpointSlices list ready Pods and a port-specific request succeeds after selector repair. Approved external probes succeed while prohibited destinations remain blocked. The expected number of ready endpoints is present and Service load tests distribute successfully. A safe object admission test passes with verified webhook TLS and expected policy behavior. ExplanationCorrect Answer D:A safe object admission test passes with verified webhook TLS and expected policy behavior. This validates the stated intervention—renew the webhook serving certificate and align its caBundle and Service reference.—rather than an unrelated cluster change. The relevant mechanism is: API admission webhooks require trusted TLS service connectivity during object admission.Incorrect Answers A:DNS resolution and a Service object do not guarantee a valid ready backend endpoint. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A safe object admission test passes with verified webhook TLS and expected policy behavior. B:Egress isolation may deny flows independently of Service discovery and Pod health. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A safe object admission test passes with verified webhook TLS and expected policy behavior. C:Services generally route only to endpoints considered ready, so Pod IP probes do not establish Service capacity. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: A safe object admission test passes with verified webhook TLS and expected policy behavior.
Question 313A CRD upgrade removes a served version while stored objects still depend on the old conversion path. What is the most direct consequence if the underlying problem is left unresolved? Resource deployment fails with no matches for kind even though YAML syntax is valid. Custom resources persist but never result in actual managed workloads. Old objects may become unreadable or controller reconciliation may fail. The custom object can remain perpetually degraded despite healthy dependencies. ExplanationCorrect Answer C:Old objects may become unreadable or controller reconciliation may fail. This directly follows from the failed condition. The preventive correction is: Plan served and storage-version conversion carefully and test existing object reads before changing the CRD. The governing Kubernetes behavior is: CRD versioning separates served API versions from the storage version and may require conversion webhooks.Incorrect Answers A:The API server must register an extension type before it can accept objects of that kind. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Old objects may become unreadable or controller reconciliation may fail. B:A CRD defines the API schema while a running controller implements the desired behavior. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Old objects may become unreadable or controller reconciliation may fail. D:Custom resource status updates are separately authorized from spec updates when status is enabled. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Old objects may become unreadable or controller reconciliation may fail.
Question 314A namespace maintainer can delete Deployments in every team namespace after a broad ClusterRoleBinding was applied. An engineer recommends: Replace the cluster-wide binding with a namespaced RoleBinding granting only needed resource verbs. Which Kubernetes behavior best explains why this correction fits? Aggregated ClusterRoles rely on matching labels on source ClusterRoles; ordinary ClusterRoles do not aggregate automatically. Authentication establishes identity while authorization separately decides permitted API verbs and resources. RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. ExplanationCorrect Answer D:A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. This explains why the proposed action—replace the cluster-wide binding with a namespaced RoleBinding granting only needed resource verbs.—is appropriate. Direct confirmation is: An impersonated authorization check denies deletion in another team's namespace but permits the required local update.Incorrect Answers A:Aggregated ClusterRoles rely on matching labels on source ClusterRoles; ordinary ClusterRoles do not aggregate automatically. This does not establish the proposed intervention. The governing principle here is: A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. Apply it by: Replace the cluster-wide binding with a namespaced RoleBinding granting only needed resource verbs. B:Authentication establishes identity while authorization separately decides permitted API verbs and resources. This does not establish the proposed intervention. The governing principle here is: A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. Apply it by: Replace the cluster-wide binding with a namespaced RoleBinding granting only needed resource verbs. C:RBAC treats interactive Pod subresources separately from operations on the parent Pod resource. This does not establish the proposed intervention. The governing principle here is: A RoleBinding restricts a Role or referenced ClusterRole to its own namespace; a ClusterRoleBinding grants across namespaces. Apply it by: Replace the cluster-wide binding with a namespaced RoleBinding granting only needed resource verbs.
Question 315A new CNI DaemonSet is running, yet cross-node Pods cannot communicate because the chosen plugin does not match node routing. What is the most direct consequence if the underlying problem is left unresolved? CNI Pods can appear healthy while application traffic is partitioned. Connectivity or security policy semantics may silently change during migration. A missing network plugin may be overlooked while storage health appears normal. The node may register but cannot create Pod sandboxes reliably. ExplanationCorrect Answer A:CNI Pods can appear healthy while application traffic is partitioned. This directly follows from the failed condition. The preventive correction is: Validate the CNI dataplane and Pod CIDR routing requirements against the infrastructure before promoting the plugin. The governing Kubernetes behavior is: CNI controls Pod network setup but cross-node reachability also depends on encapsulation or underlay routes.Incorrect Answers B:Network policy implementations and CNI-specific resources can differ despite common Pod-network requirements. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: CNI Pods can appear healthy while application traffic is partitioned. C:CNI handles networking while CSI handles storage; shared namespaces do not imply interchangeable capabilities. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: CNI Pods can appear healthy while application traffic is partitioned. D:The container runtime interface is the kubelet's integration boundary for Pod sandboxes and containers. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: CNI Pods can appear healthy while application traffic is partitioned.
Question 316A dynamic PVC stays Pending because its StorageClass provisioner name does not match any installed CSI driver. What is the most direct consequence if the underlying problem is left unresolved? Volume provisioning can succeed while Pods fail at mount time. Stateful Pods remain unschedulable without a provisioned volume. Sensitive application data may be stored on noncompliant volumes. Stateful workloads may run out of space despite a requested resize. ExplanationCorrect Answer B:Stateful Pods remain unschedulable without a provisioned volume. This directly follows from the failed condition. The preventive correction is: Select a StorageClass with an installed valid provisioner or install the required CSI driver. The governing Kubernetes behavior is: Dynamic provisioning depends on a controller that recognizes the StorageClass provisioner identifier.Incorrect Answers A:A controller may provision a volume while the node-side CSI plugin is still required to mount it. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Stateful Pods remain unschedulable without a provisioned volume. C:StorageClass parameters are driver-specific and determine backend provisioning behavior. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Stateful Pods remain unschedulable without a provisioned volume. D:PVC size expansion depends on both StorageClass permission and storage-driver capabilities. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Stateful Pods remain unschedulable without a provisioned volume.
Question 317A Gateway is deployed but the namespace's HTTPRoute has no permission to attach across namespace boundaries. Following the proposed change—Configure the listener allowedRoutes namespaces policy or appropriate cross-namespace reference permissions—which observation best demonstrates that the particular fault is resolved? HTTPRoute status reports Accepted and traffic matches its intended listener. The intended route becomes Accepted only from an approved namespace. Gateway conditions become Programmed/Accepted and the assigned address receives requests. Only Pods in approved labeled namespaces reach the target Service. ExplanationCorrect Answer B:The intended route becomes Accepted only from an approved namespace. This validates the stated intervention—configure the listener allowedRoutes namespaces policy or appropriate cross-namespace reference permissions.—rather than an unrelated cluster change. The relevant mechanism is: Gateway API constrains route attachment and cross-namespace references through explicit policies.Incorrect Answers A:Gateway API HTTPRoutes must attach to compatible allowed listeners through parent references. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The intended route becomes Accepted only from an approved namespace. C:Gateway API resources require an implementing controller to provision and program data-plane listeners. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The intended route becomes Accepted only from an approved namespace. D:An empty namespaceSelector matches all namespaces; namespace and Pod peers must be scoped deliberately. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The intended route becomes Accepted only from an approved namespace.
Question 318Kube-controller-manager logs report forbidden errors when updating Endpoints after its kubeconfig credentials or API permissions changed. Following the proposed change—Restore the intended component identity and required API authorization rather than granting cluster-admin broadly—which observation best demonstrates that the particular fault is resolved? The controller-manager stays Running and a Deployment reconciles desired replicas. The specific controller resumes writes and unauthorized privileges remain denied. The API server static Pod starts and authenticated kubectl requests succeed. The scheduler becomes healthy and newly created Pods receive node assignments. ExplanationCorrect Answer B:The specific controller resumes writes and unauthorized privileges remain denied. This validates the stated intervention—restore the intended component identity and required API authorization rather than granting cluster-admin broadly.—rather than an unrelated cluster change. The relevant mechanism is: Control-plane components need functional credentials and narrowly appropriate permissions to reconcile state.Incorrect Answers A:The kubelet restarts static control-plane Pods when their manifest changes. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The specific controller resumes writes and unauthorized privileges remain denied. C:kubeadm control-plane components often run as static Pods from local manifests. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The specific controller resumes writes and unauthorized privileges remain denied. D:The scheduler assigns unscheduled Pods to eligible nodes; other control-plane components cannot replace it. This relates to a separate check and does not verify the specified repair. The relevant acceptance evidence is: The specific controller resumes writes and unauthorized privileges remain denied.
Question 319A worker image uses a container engine without the CRI interface required by the kubelet. What is the most direct consequence if the underlying problem is left unresolved? Stateful workloads remain Pending or stuck during volume attach. The node may register but cannot create Pod sandboxes reliably. A missing network plugin may be overlooked while storage health appears normal. CNI Pods can appear healthy while application traffic is partitioned. ExplanationCorrect Answer B:The node may register but cannot create Pod sandboxes reliably. This directly follows from the failed condition. The preventive correction is: Install a supported CRI runtime and configure the kubelet runtime endpoint correctly. The governing Kubernetes behavior is: The container runtime interface is the kubelet's integration boundary for Pod sandboxes and containers.Incorrect Answers A:CSI driver registration and controller/node capabilities are required for external volume provisioning and mounting. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: The node may register but cannot create Pod sandboxes reliably. C:CNI handles networking while CSI handles storage; shared namespaces do not imply interchangeable capabilities. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: The node may register but cannot create Pod sandboxes reliably. D:CNI controls Pod network setup but cross-node reachability also depends on encapsulation or underlay routes. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: The node may register but cannot create Pod sandboxes reliably.
Question 320A team installed metrics-server but kubectl top nodes returns an unavailable metrics API error. What is the most direct consequence if the underlying problem is left unresolved? Tuning the wrong node resource may leave the slow service unchanged. Operators remain blind to live resource pressure during incidents. Incorrect attribution may cause wasteful scaling and leave the true issue uncorrected. Performance can degrade even when the host has unused CPU. ExplanationCorrect Answer B:Operators remain blind to live resource pressure during incidents. This directly follows from the failed condition. The preventive correction is: Verify the aggregated metrics APIService and metrics-server connectivity to kubelets. The governing Kubernetes behavior is: kubectl top relies on metrics.k8s.io aggregation and healthy scraper access to node statistics.Incorrect Answers A:Node-wide utilization can hide a throttled or memory-constrained application container. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Operators remain blind to live resource pressure during incidents. C:Node memory usage includes multiple consumers and cannot be attributed to one Pod without evidence. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Operators remain blind to live resource pressure during incidents. D:CPU limits may cause throttling independent of overall node idle time. That does not describe the principal exposure from the stated failure. The direct avoidable risk is: Operators remain blind to live resource pressure during incidents.