The client had a Node.js app that was deployed the old-fashioned way manual SSH, manual restarts. Every deploy caused about 10 seconds of downtime (503 errors) while the app reloaded. We containerized the whole thing and rebuilt the deployment process from the ground up. What we did: Wrote optimized Dockerfiles for the app and its supporting services, and set it all up with Docker Compose. Used Docker Swarm to run 2 replicas of the app and do rolling releases, so one instance stays up while the other updates. That's what killed the downtime. Put Nginx in front as a reverse proxy, with HTTPS via Let's Encrypt/Certbot. Set it up to only accept HTTPS traffic behind Cloudflare, without creating a redirect loop. Added health checks and automatic container restarts, so a bad container doesn't stay bad. Set up persistent volumes so app data and logs survive deployments and container replacements. Added centralized logging (Graylog) and log rotation. Wired deployments to the client's Git repo, so a push kicks off the build and rollout automatically no more manual SSH for routine releases. Locked down the network: only the ports that need to be public are public, everything else stays internal. Set up database backups before any risky deployment, and documented recovery steps. The result: deployments went from "risky and manual" to "push and forget," with no downtime and a proper safety net if something goes wrong.