rollback

name	rollback
description	Roll back a failed deployment to the previous revision
disable-model-invocation	true
argument-hint	[service] [env]

/rollback [service] [env]

Roll back a failed deployment to its previous working revision.

Arguments

• service — Service to roll back (e.g., api-gateway, customers-service) or all
• env — Target environment: dev or prod (default: dev)

Steps

1. Set the namespace:

   • dev → petclinic-dev
   • prod → petclinic-prod

2. Show current rollout status and history:

   kubectl rollout status deployment/{service} -n petclinic-{env}
   kubectl rollout history deployment/{service} -n petclinic-{env}
   

3. Show what changed in the current revision vs previous:

   kubectl rollout history deployment/{service} -n petclinic-{env} --revision={current}
   kubectl rollout history deployment/{service} -n petclinic-{env} --revision={previous}
   

4. For prod: require explicit confirmation. Show: "Rolling back {service} in PRODUCTION to revision {N}. This will change the running image. Confirm?"

5. Execute the rollback:

   kubectl rollout undo deployment/{service} -n petclinic-{env}
   

6. Monitor rollout:

   kubectl rollout status deployment/{service} -n petclinic-{env} --timeout=120s
   

7. Verify the rollback:

   kubectl get pods -l app.kubernetes.io/name={service} -n petclinic-{env}
   

8. If rolling back all:

   • Roll back in reverse order (application services first, then discovery, then config)
   • Verify each service before proceeding to the next
   • If any rollback fails, stop and report

9. Show summary:

   ## Rollback: {service} ({env})
   
   Previous revision: {N} (image: {old-image})
   Rolled back to: {N-1} (image: {previous-image})
   Status: {success/failed}
   Pod status: {Running/etc}
   

Important

• Always get explicit confirmation for prod rollbacks
• When rolling back all, go in reverse deployment order
• If rollback fails, do NOT retry — investigate first
• Suggest running /smoke-test {env} after rollback to verify health