CNPE Study Strategy¶

Follow the weights¶

Domain	Weight	Share of study time
Platform Architecture and Infrastructure	15%	Networking, storage, compute, cost, multi-tenancy
GitOps and Continuous Delivery	25%	Argo CD, Flux, Tekton, canary and blue/green
Platform APIs and Self-Service Capabilities	25%	CRDs, operators, Crossplane, self-service provisioning
Observability and Operations	20%	Prometheus, OpenTelemetry, DORA metrics, incidents
Security and Policy Enforcement	15%	mTLS, RBAC, audit, SBOM, policy engines, pipeline scanning

Practice finding, not remembering¶

The curriculum says candidates complete tasks "using unfamiliar tools by relying on help and documentation available in the exam." Train the skill the exam tests: given a tool you have never used, find a working example in its documentation, adapt it and verify it. Which documentation sites the exam permits is not stated on the LF page, so confirm in the candidate instructions.

Time¶

Two hours for multiple tasks, count not published. Practice triage: do the tasks you can finish fast first, flag the ones that need reading, and never leave a task half applied without noting it.

Verify every task¶

A performance-based exam grades the state of the cluster, not your intent. After every change run the check that proves it: kubectl get, a status condition, a sync status, a metric. Under GitOps remember that fixes made with kubectl can be reverted; change the source.

Distinctions to own¶

Argo CD prune versus selfHeal.
Source versus Kustomization in Flux.
Pipeline versus PipelineRun in Tekton.
Rollout versus Canary (Flagger): replace the Deployment versus wrap it.
CRD versus operator: schema versus controller.
XRD versus Composition versus XR.
PERMISSIVE versus STRICT.
Audit level versus audit stage.
Constraint versus ConstraintTemplate.

Do not over-study¶

You will not be tested on deep tool-specific knowledge unless it is in the competencies. Breadth across the listed tools, with strong Kubernetes fundamentals, beats depth in one.