Comparison
Graceful shutdownvsZero-downtime deployment
Graceful shutdown
on SIGTERM you stop accepting new requests, finish the ones in flight, and only then exit.
Ending a process in a way that does not drop work: deregister from the load balancer, stop accepting, complete in-flight requests, close connections, exit. Without it, every deploy sheds a burst of errors proportional to your request rate. The grace period must exceed your longest normal request, or the platform kills you mid-request anyway.
Full entry →Zero-downtime deployment
you ship in the middle of the afternoon and no request fails, because nothing is ever stopped while it is still serving.
Deploying without dropping requests, which requires readiness checks, graceful shutdown, compatible schema changes and a rollout that maintains capacity. It is less a feature than the sum of several other disciplines. Almost all deploy-time error spikes come from missing one of them, usually shutdown handling.
Full entry →