Migration and deployment failures

Stop before making it worse

Do not repeatedly redeploy a failing release while changing database schema or volumes blindly. Capture the migration error and service logs first.

Migration failure

Identify the failed migration, confirm whether it made partial changes, and use a tested backup/restore procedure if the migration is not safely reversible. Application image rollback does not automatically reverse database changes.

Service health failure

Check environment variables, database/Redis reachability, storage mounts, app key, permissions, and the health endpoint response inside the network.

Rollback

Roll back stateless application images to a previous immutable tag while preserving the same persistent volumes. Never use a cleanup command that deletes production volumes as part of ordinary rollback.