deployment-procedures

Follow a five-phase process: prepare by verifying tests and environment variables, back up the current state, deploy while monitoring, verify health endpoints and key user flows, then confirm or roll back. Never deploy untested code or skip the backup step.

Rolling deployments replace instances one by one and suit standard releases. Blue-green switches traffic between two environments for easy rollback of high-risk changes. Canary shifts traffic gradually to validate changes with real users.

Roll back immediately when the service is down or critical errors appear. Consider rollback if performance degrades more than 50 percent. For minor issues, a quick fix forward is acceptable. Roll back first and debug after the service is stable.

On Vercel or Netlify, redeploy the previous commit. On Railway or Render, use the dashboard rollback. With Docker, redeploy the previous image tag. On Kubernetes, run kubectl rollout undo. For VPS with PM2, restore the backup and restart.

Verify the health endpoint responds, scan error logs for new exceptions, test key user flows, and confirm response times are acceptable. Monitor actively for the first 5 minutes, confirm stability at 15 minutes, and review metrics the next day.

Friday deployments leave little time to monitor and respond to issues before the weekend, when team availability is low. Deploy early in the week so you have time to observe behavior and roll back if problems appear.