Free Linux Foundation CNPA Practice Exam with Questions & Answers | Set: 3

The most effective approach for scalable Kubernetes infrastructure provisioning is Crossplane compositions. Option D is correct because compositions let platform teams define custom CRDs (Composite Resources) that abstract infrastructure details while embedding organizational policies and guardrails. Developers then consume these abstractions through simple Kubernetes-native APIs, enabling self-service at scale.

Option A (Helm with values.yaml) is useful for application deployment but not for scalable infrastructure provisioning across multiple clouds. Option B (imperative scripts) lacks scalability, repeatability, and governance. Option C (static YAML with kubectl apply) is manual and not suited for dynamic, multi-team environments.

Crossplane compositions allow platform teams to curate golden paths while giving developers autonomy. This reduces complexity, ensures compliance, and supports multi-cloud provisioning—all key aspects of platform engineering.

Argo CD is a GitOps-native continuous delivery tool specifically designed for Kubernetes. Option B is correct because Argo CD continuously monitors Git repositories for desired application state and reconciles Kubernetes clusters accordingly. It is declarative, Kubernetes-native, and aligned with GitOps principles, making it a key tool in platform engineering.

Option A (TravisCI) and Option C (CircleCI) are CI/CD systems but not Kubernetes-native or GitOps-driven. Option D (Jenkins) is a widely used CI/CD tool but operates primarily in a push-based model unless extended with plugins, and is not purpose-built for GitOps.

Argo CD provides automated deployments, drift detection, rollback, and auditability—features central to GitOps workflows. It simplifies multi-cluster management, enforces compliance, and reduces manual intervention, making it a leading choice in Kubernetes-based platform engineering.

The defining capability of an IDP is enabling self-service so developers can independently access infrastructure and platform resources. Option D is correct because self-service resource provisioning APIs allow developers to provision resources such as namespaces, databases, or environments without relying on manual intervention from the platform team. These APIs embed governance, compliance, and organizational guardrails while giving autonomy to development teams.

Option A (manual deployment services) defeats the purpose of self-service. Option B (knowledge centers) improve documentation but do not provide automation. Option C (logging/monitoring interfaces) are observability tools, not resource provisioning mechanisms.

Self-service APIs empower developers, reduce cognitive load, and minimize bottlenecks. They also align with the platform engineering principle of “treating the platform as a product,” where developers are customers, and the platform offers curated golden paths to simplify consumption of infrastructure and services.

The most effective stage to generate a Software Bill of Materials (SBOM) is during the build process. Option C is correct because the build phase is when dependencies are resolved and artifacts (e.g., container images, binaries) are created. Generating an SBOM at this point provides a complete, accurate inventory of all included libraries and components, which is critical for vulnerability scanning, license compliance, and supply chain security.

Option A (testing) is too late to capture all dependencies reliably. Option B (before committing code) cannot provide a full SBOM because builds often introduce additional dependencies. Option D (after deployment) delays insights until production, missing the opportunity to detect and remediate issues early.

Integrating SBOM generation into CI/CD pipelines enables shift-left security, ensuring vulnerabilities are detected early and allowing remediation before artifacts reach production. This aligns with CNCF supply chain security practices and platform engineering goals.

The Kubernetes reconciliation loop ensures that the actual state of a resource matches the desired state defined in its manifest. If the replica count of a Deployment is changed from 3 to 5, option B is correct: Kubernetes will automatically create two new Pods to satisfy the new desired replica count.

Option A is incorrect because Deployments are not deleted; they are updated in place. Option C contradicts Kubernetes’ declarative model—no manual intervention is required. Option D is wrong because Kubernetes does not restart existing Pods unless necessary; it simply adds additional Pods.

This reconciliation process is core to Kubernetes’ declarative infrastructure approach, where desired states are continuously monitored and enforced. It reduces human toil and ensures consistency, making it fundamental for platform engineering practices like GitOps.