Free CNPA practice test: 30 questions with answers

Declarative management describes the desired state, and a controller works to make the actual state match it. Imperative scripts and ad hoc terminal changes record steps rather than intent and drift over time.

The reconciliation loop repeatedly observes the actual state, compares it with the desired state, and acts to converge them. Admission controllers validate or mutate requests, and garbage collection removes unused objects.

A platform reduces cognitive load by providing paved paths and feedback loops are central to DevOps. Platforms do not remove shared responsibility, and bespoke stacks per team work against standardization.

A shared base with environment overlays or values keeps the common definition in one place and isolates the variations. Separate hand-maintained copies and manual memory lead to configuration drift.

GitOps controllers such as Argo CD and Flux pull the desired state from Git and reconcile the cluster to it. Registries store images, test runners verify code, and DNS resolves names, so none of them enforce cluster state.

Traces record the timing and path of one request through each service as linked spans, which makes latency hotspots visible. Node CPU graphs and image lists do not show per-request paths.

OpenTelemetry defines vendor-neutral APIs, SDKs, and a collector for traces, metrics, and logs. Helm packages applications, Argo CD performs GitOps delivery, and Envoy Gateway handles traffic routing.

Mutual TLS authenticates both peers and encrypts traffic, and a service mesh can enforce it transparently through sidecars. A shared token does not authenticate individual peers, and public exposure weakens security.

Policy engines evaluate resources during admission and reject those that violate rules, such as privileged containers. Dashboards and scanners report on state, and DNS load balancing does not enforce policy.

Artifact signing and verification ensure that only artifacts produced by trusted pipelines are deployed. Shared credentials, disabled logs, and direct deployment all increase the attack surface.

A canary release routes a small percentage of traffic to the new version and expands only if metrics stay healthy. Big-bang and delete-first approaches expose all users at once, and test-only deployment does not validate production behavior.

A GitOps controller reconciles the cluster back to the desired state in Git, so a manual edit can be overwritten and the source of truth is bypassed. Manual edits are not committed to Git automatically, and they do not spread across environments.

Promoting a single immutable digest means what was tested is what runs in production. Rebuilding per environment can introduce different dependencies or build results, which breaks that guarantee.

Small, frequent changes and practiced rollback or forward-fix procedures shorten the time to recover. Large batches and skipped tests make failures harder to isolate and more likely to reach users.

Rolling back to the last known good version through the delivery path restores service fast and keeps the change recorded. Manual edits, long rebuilds, and disabled monitoring each make recovery slower or hide the problem.

CRDs extend the Kubernetes API with new resource kinds, which a controller can then reconcile into real infrastructure. ConfigMaps store data rather than define types, labels describe nodes, and pod security standards govern pod settings.

An operator combines a custom resource with a controller that reconciles it, encoding operational knowledge for a managed application or resource. The other patterns describe different topologies or deployment strategies.

Declarative infrastructure as code is versioned, reviewable, and can be reconciled, which makes changes repeatable and auditable. Console clicks, one-off commands, and copied IDs are not reproducible.

Reconciliation is repeated on events and retries, so the same input should lead to the same final state without side effects piling up. Creating new resources each run causes duplicates, and idempotency matters for controllers that write too.

openAPIV3Schema describes the structure and types the API server validates against when objects are created or updated. Printer columns affect display, owner references manage garbage collection, and finalizers control deletion.

A developer portal with a service catalog centralizes discovery of services, owners, and documentation, which reduces friction. Packet capture, tape backup, and replication monitoring do not serve discovery.

A developer portal typically provides a service catalog for discovery and templates that scaffold new services on paved paths. Printer backups and rack inventories are not developer portal features.

Keeping a human in the loop and enforcing policy checks limits the impact of wrong suggestions and keeps accountability clear. Automatic production changes, disabled logs, and broad admin rights all increase risk.

Backstage catalog entities are described in YAML files that are registered with the catalog. The other formats are not used for these descriptors.

Discoverable documentation and templates speed onboarding and reduce routine ticket volume. A portal does not remove documentation needs or guarantee zero incidents.

Deployment frequency and lead time for changes are two of the four DORA throughput and stability measures. Ticket counts and code volume measure activity rather than delivery performance.

Onboarding time to first deployment reflects how well the platform removes friction for product teams. Meeting counts, commit totals, and dashboard counts measure activity rather than impact.

Change failure rate is the share of production changes that cause a failure requiring remediation, which measures delivery stability. The other options describe support volume, coding time, or patch status.

Lead time for changes measures the elapsed time from commit to successful production deployment, which reflects delivery speed. The other options measure unrelated activity.

Adoption rate tracks how many teams use the paved path, which indicates whether the platform delivers value. The other measures do not describe adoption.