ArgoCD rollback should start from Git, not kubectl

Problem

An application breaks after deployment and the first instinct is to run `kubectl` manually.

Why it happens

Manual rollback feels faster, but it creates drift between Git and the cluster.

Fix / Approach

Revert the Git commit or restore the previous manifest version, then let ArgoCD sync the desired state.

Command / Config

git revert <commit>
git push origin main
argocd app sync <app-name>

When to use

• GitOps-managed applications
• ArgoCD-managed clusters

When not to use

• Emergency break-glass situations where immediate cluster-level action is required