CI/CD Integration: CGOA practice questions

CI/CD integration is about where a classic build pipeline ends and the GitOps agent takes over. In a push-based setup the CI system builds an artifact and then deploys it, which means the pipeline holds credentials for every environment it deploys to. In a GitOps setup the pipeline still builds and tests, but its last step is to record the new desired state in the configuration repository; an agent that pulls from that repository applies it.

The hand-off usually happens through an image reference. CI builds a container image, pushes it to a registry, and updates the tag or digest in the manifests. Flux can automate that step itself: an ImageRepository tells Flux which registry to scan for new tags, an ImagePolicy selects which tag to use (for example within a semver range), a marker comment in the manifest names the policy that controls a field, and an ImageUpdateAutomation commits the new tag back to Git. The change then flows through the same reconciliation path as any other commit.

Argo CD's best practices explain why the configuration usually lives in a separate repository from the application source. Separation keeps application code cleanly apart from application config. It separates access, because people who may commit code are not automatically allowed to change production. It produces a cleaner audit log without the noise of everyday development commits. And it avoids an infinite loop of builds that can occur when a pipeline commits manifest changes into the same repository that triggers it.

Promotion between environments follows from this. The same artifact, identified by tag or digest, moves from development to staging to production as validated configuration changes in the repository, typically by pull request, so each promotion is reviewed and recorded. Argo CD also recommends that manifests refer to a Git tag or commit SHA rather than a moving HEAD when they pull in remote bases, so a deployed revision does not change meaning underneath you.

Rollback in this model is a change to the desired state. Argo CD can roll back to any application configuration committed to Git, but its documentation notes that rollback cannot be performed on an application with automated sync enabled; with automation on, you revert in Git and let the agent reconcile. Reverting the commit keeps the history intact, so the audit trail records both the change and its undoing.

Feedback closes the loop. Argo CD's CLI is built for automation and CI integration, and Flux's notification controller can report reconciliation events back to chat tools and receive webhooks from GitHub, GitLab or Jenkins to trigger an early fetch.