Deploy

Step 3 — Squash merge

• Once CI is green, squash merge: gh pr merge <number> --squash --delete-branch
• Confirm the merge succeeded

Step 4 — Watch ArgoCD detect the change

Do NOT use sleep. Instead, actively poll for the rollout:

1. First, wait for the manifests branch to update (CI exports manifests on push to main):

   gh run list --branch main --limit 1 --json status,conclusion
   

   Poll this with gh run watch until the post-merge CI run completes.

2. Then watch the affected pods for restart/rollout using kubectl:

   kubectl get pods -n <namespace> -l app=<app-name> -w
   

   Use a short timeout (90s). Look for new pod creation or container restart. If the pod doesn't restart, check if ArgoCD has synced:

   argocd app get <app-name> --grpc-web -o json | jq '{syncStatus: .status.sync.status, healthStatus: .status.health.status}'
   

Step 5 — Verify health

For each affected service:

1. Pod status: Confirm the pod is Running and containers are Ready:

   kubectl get pods -n <namespace> -l app=<app-name>
   

2. Logs: Check recent logs for errors (last 2 minutes since rollout):

   kubectl logs -n <namespace> -l app=<app-name> --tail=50 --since=2m
   

3. Probes (if the change added/modified probes): Confirm no restart loops by checking RESTARTS column and describe the pod to verify probe config:

   kubectl describe pod -n <namespace> -l app=<app-name> | grep -A5 'Liveness\|Readiness\|Startup'
   

4. Related services: If the change could affect dependent services (e.g. a DB change might affect apps using it), check those pods too.

Report a clear summary: which pods rolled, current status, and any warnings from logs.