Why this topic matters in interviews
ArgoCD interviews test whether you understand more than clicking Sync. A strong engineer explains Git as desired state, cluster reconciliation, rollback, drift, repository structure, environment promotion and incident recovery.
15 interview questions to prepare
GitOps is an operating model where Git stores the desired state and automation reconciles live infrastructure or applications to match that state.
ArgoCD is a Kubernetes GitOps controller that continuously compares Git desired state with live cluster state and syncs applications.
Synced means live manifests match Git. Healthy means the Kubernetes resources are operational according to health checks.
It compares desired manifests from Git with live resources in the cluster. Differences become OutOfSync.
Auto-sync lets ArgoCD automatically apply Git changes to the cluster. In production it should be combined with approvals and policy controls.
Preferred rollback is often Git revert so history remains source-of-truth. ArgoCD history rollback may be used carefully depending on process.
A parent ArgoCD Application manages child Applications, allowing teams to bootstrap many applications or environments from one root app.
Projects define boundaries for applications, repositories, clusters, namespaces and permissions.
ApplicationSet generates multiple Applications from templates using generators such as list, cluster, Git or matrix generators.
Check application events, manifest generation errors, repository access, Helm/Kustomize errors, RBAC, namespace permissions and Kubernetes events.
Manual changes, defaulted fields, ignored differences, generated values, admission controllers or server-side mutations can cause drift.
Use SSO, RBAC, least privilege, private repo credentials, network controls, secret management and restricted Projects.
Use separate environment overlays/branches/directories and promote versions through reviewed pull requests.
Pruning deletes live resources no longer defined in Git. It is powerful but risky and should be controlled in production.
Say: Git stores desired state; ArgoCD continuously compares and reconciles clusters; changes happen through reviewed commits; drift and rollback become auditable.