Exam Bundle

The production rollout starts with image pulls succeeding and no missing-credential failure during the first Pod startup. That observed result would support another specific correction: Add pre-deployment checks for registry access, required Secrets and namespace permissions before promotion. The other scenario concerns: A software team promotes manifests without checking whether container registry credentials exist in the target environment. The current Brookside Manufacturing problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

An installed chart creates ordinary Deployments and Services which the Kubernetes control plane subsequently reconciles. That observed result would support another specific correction: Describe Helm as a packaging and release tool that renders and manages Kubernetes resource definitions. The other scenario concerns: A team believes Helm charts run containers by themselves and expects an installed chart to replace Kubernetes controllers. The current Brookside Manufacturing problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

The pipeline verifies that new Pods become ready before they receive normal production Service traffic. That observed result would support another specific correction: Include appropriate readiness and startup checks as part of a tested declarative release specification. The other scenario concerns: An operator applies a new Deployment manifest but removes a container health check that guarded the previous release. The current Brookside Manufacturing problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

Use the HPA's supported desired behavior and define min/max replicas rather than continually overriding controller intent. That choice addresses a different foundational issue: An engineer scales an application manually during every traffic spike while an approved HPA already controls its replica target. The issue at Juniper Media is instead: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. It does not establish the required correction; the decisive verification would be: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

Use a repeatable deployment pipeline that publishes a versioned image and updates declarative Kubernetes manifests. That choice addresses a different foundational issue: An organization merges code frequently but deploys new versions by manually copying files into running containers. The issue at Juniper Media is instead: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. It does not establish the required correction; the decisive verification would be: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

Explain that Kustomize customizes declarative manifests before API submission, rather than acting as a cluster runtime. That choice addresses a different foundational issue: A team changes a Kustomize overlay expecting it to create a runtime configuration service on a worker. The issue at Juniper Media is instead: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. It does not establish the required correction; the decisive verification would be: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

The pipeline verifies that new Pods become ready before they receive normal production Service traffic. That observed result would support another specific correction: Include appropriate readiness and startup checks as part of a tested declarative release specification. The other scenario concerns: An operator applies a new Deployment manifest but removes a container health check that guarded the previous release. The current Juniper Media problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

The output manifests contain the environment-specific patch and the normal Kubernetes controllers run the final workload. That observed result would support another specific correction: Explain that Kustomize customizes declarative manifests before API submission, rather than acting as a cluster runtime. The other scenario concerns: A team changes a Kustomize overlay expecting it to create a runtime configuration service on a worker. The current Juniper Media problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

A simulated failed rollout restores a known good revision and its exact image digest is still retrievable. That observed result would support another specific correction: Retain suitable Deployment revision history and keep verified release images available for controlled rollback. The other scenario concerns: A delivery team deletes older ReplicaSets after every rollout and later discovers there is no supported revision to undo. The current Juniper Media problem is: A rolling update becomes unavailable because all old Pods are removed before any replacements pass readiness. The evidence required here is: A representative rollout keeps the agreed minimum number of ready replicas available throughout the update.

Establish automated build and test validation in CI, with a separate approved continuous delivery or deployment stage. That choice addresses a different foundational issue: A release team pushes untested images directly into production and calls the sequence continuous integration. The issue at Highland Education is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Validate the template rendering and configuration values before applying the resulting Kubernetes resources. That choice addresses a different foundational issue: A release build fails because a Helm values file references a nonexistent key and the team deploys unrendered templates anyway. The issue at Highland Education is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Add pre-deployment checks for registry access, required Secrets and namespace permissions before promotion. That choice addresses a different foundational issue: A software team promotes manifests without checking whether container registry credentials exist in the target environment. The issue at Highland Education is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

The HPA displays valid current metrics and adjusts replicas after an authorized load change. That observed result would support another specific correction: Supply supported resource or custom metrics with appropriate requests and a working metrics adapter for autoscaling. The other scenario concerns: A team tries to use an HPA without any usable metrics source and assumes it will infer load from container image size. The current Highland Education problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

The pipeline rejects a failing commit before publishing or deploying an unverified application image. That observed result would support another specific correction: Establish automated build and test validation in CI, with a separate approved continuous delivery or deployment stage. The other scenario concerns: A release team pushes untested images directly into production and calls the sequence continuous integration. The current Highland Education problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

The production rollout starts with image pulls succeeding and no missing-credential failure during the first Pod startup. That observed result would support another specific correction: Add pre-deployment checks for registry access, required Secrets and namespace permissions before promotion. The other scenario concerns: A software team promotes manifests without checking whether container registry credentials exist in the target environment. The current Highland Education problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Retain suitable Deployment revision history and keep verified release images available for controlled rollback. That choice addresses a different foundational issue: A delivery team deletes older ReplicaSets after every rollout and later discovers there is no supported revision to undo. The issue at Eastshore Energy is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Commit approved desired-state changes to the authoritative repository and let the GitOps reconciler apply them. That choice addresses a different foundational issue: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The issue at Eastshore Energy is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Validate the template rendering and configuration values before applying the resulting Kubernetes resources. That choice addresses a different foundational issue: A release build fails because a Helm values file references a nonexistent key and the team deploys unrendered templates anyway. The issue at Eastshore Energy is instead: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. It does not establish the required correction; the decisive verification would be: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Multiple apps render consistent manifests from one maintained package while retaining their approved environment settings. That observed result would support another specific correction: Use a reusable and versioned Helm chart or appropriate configuration template with explicit per-app values. The other scenario concerns: A project copies the same long Kubernetes manifest into twenty application repositories and struggles with consistent upgrades. The current Eastshore Energy problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

The output manifests contain the environment-specific patch and the normal Kubernetes controllers run the final workload. That observed result would support another specific correction: Explain that Kustomize customizes declarative manifests before API submission, rather than acting as a cluster runtime. The other scenario concerns: A team changes a Kustomize overlay expecting it to create a runtime configuration service on a worker. The current Eastshore Energy problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

The production rollout starts with image pulls succeeding and no missing-credential failure during the first Pod startup. That observed result would support another specific correction: Add pre-deployment checks for registry access, required Secrets and namespace permissions before promotion. The other scenario concerns: A software team promotes manifests without checking whether container registry credentials exist in the target environment. The current Eastshore Energy problem is: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The evidence required here is: The incident timeline identifies the exact application artifact and Kubernetes rollout revision running when the fault occurred.

Describe Helm as a packaging and release tool that renders and manages Kubernetes resource definitions. That choice addresses a different foundational issue: A team believes Helm charts run containers by themselves and expects an installed chart to replace Kubernetes controllers. The issue at Maplewood Schools is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Add pre-deployment checks for registry access, required Secrets and namespace permissions before promotion. That choice addresses a different foundational issue: A software team promotes manifests without checking whether container registry credentials exist in the target environment. The issue at Maplewood Schools is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Publish immutable versioned image references and retain release metadata linking a commit to its deployed digest. That choice addresses a different foundational issue: A team creates a Deployment with an identical application tag for every release and cannot correlate an incident to code. The issue at Maplewood Schools is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

The release changes Pod replica count without asserting new nodes were added unless a separate capacity mechanism runs. That observed result would support another specific correction: Distinguish workload replica scaling from underlying cluster/node autoscaling in the delivery design. The other scenario concerns: A release engineer increases Deployment replicas and describes the change as horizontal node autoscaling. The current Maplewood Schools problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

A simulated failed rollout restores a known good revision and its exact image digest is still retrievable. That observed result would support another specific correction: Retain suitable Deployment revision history and keep verified release images available for controlled rollback. The other scenario concerns: A delivery team deletes older ReplicaSets after every rollout and later discovers there is no supported revision to undo. The current Maplewood Schools problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Dev, staging and production show their intended parameter differences without untracked live drift. That observed result would support another specific correction: Use declarative environment overlays or configuration values under version control with a reproducible promotion process. The other scenario concerns: An application is configured in three environments through manual edits that leave different values undocumented. The current Maplewood Schools problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Validate the template rendering and configuration values before applying the resulting Kubernetes resources. That choice addresses a different foundational issue: A release build fails because a Helm values file references a nonexistent key and the team deploys unrendered templates anyway. The issue at Brookside Manufacturing is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Use declarative environment overlays or configuration values under version control with a reproducible promotion process. That choice addresses a different foundational issue: An application is configured in three environments through manual edits that leave different values undocumented. The issue at Brookside Manufacturing is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

Use the HPA's supported desired behavior and define min/max replicas rather than continually overriding controller intent. That choice addresses a different foundational issue: An engineer scales an application manually during every traffic spike while an approved HPA already controls its replica target. The issue at Brookside Manufacturing is instead: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. It does not establish the required correction; the decisive verification would be: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

The green version passes health and acceptance tests before clients are directed to it, with a working rollback path. That observed result would support another specific correction: Run and validate a separate green version before atomically switching the intended traffic selection. The other scenario concerns: An application requires blue-green deployment but the team updates the shared Service selector before testing the green environment. The current Brookside Manufacturing problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

A repository change follows review and audit controls while ordinary read access grants no direct cluster administrator privilege. That observed result would support another specific correction: Explain that GitOps uses version-controlled desired state with controlled reconciliation and separate authorization boundaries. The other scenario concerns: A platform team describes GitOps as granting every repository reader administrative kubectl access to the cluster. The current Brookside Manufacturing problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.

The HPA displays valid current metrics and adjusts replicas after an authorized load change. That observed result would support another specific correction: Supply supported resource or custom metrics with appropriate requests and a working metrics adapter for autoscaling. The other scenario concerns: A team tries to use an HPA without any usable metrics source and assumes it will infer load from container image size. The current Brookside Manufacturing problem is: A developer edits production Deployment resources from a laptop despite a GitOps controller continuously restoring the old manifest. The evidence required here is: The controller reports a synchronized state and no live change is repeatedly reversed by reconciliation.