For the Clínica Bíblica patient API, downtime wasn’t an option — patients hit the service around the clock. Here’s the pipeline that let us deploy continuously without anyone noticing.
Build once, promote everywhere
The CI job builds a single artifact and promotes it through staging to production. The thing you tested is the thing you ship.
Reload, don’t restart
Nginx reload swaps workers gracefully, draining in-flight requests instead of dropping them. Combined with health checks, the cutover is invisible.
Roll back in seconds
Keep the previous release symlinked. A bad deploy is a one-command revert, not a 2am incident.