Add graceful shutdown to the HTTP server #1

New Issue

2026-06-02T04:26:38Z

logaritmisk commented

2026-06-02 04:26:38 +00:00

server::serve currently calls axum::serve(listener, app).await with no shutdown signal, so it runs forever and is killed abruptly (crates/server/src/lib.rs).

Before the service is deployed for real (self-host is a first-class target, see docs/specs/2026-06-02-mvp-architecture.md §12), wire with_graceful_shutdown to SIGTERM/SIGINT so in-flight requests can drain and the DB pool closes cleanly.

Add a shutdown-signal future (SIGTERM + SIGINT / Ctrl-C).
Pass it to axum::serve(...).with_graceful_shutdown(...).
Optionally close the sqlx pool on shutdown.

Source: Foundation (Plan 0) final review — deferred, non-blocking.

`server::serve` currently calls `axum::serve(listener, app).await` with no shutdown signal, so it runs forever and is killed abruptly (`crates/server/src/lib.rs`). Before the service is deployed for real (self-host is a first-class target, see `docs/specs/2026-06-02-mvp-architecture.md` §12), wire `with_graceful_shutdown` to `SIGTERM`/`SIGINT` so in-flight requests can drain and the DB pool closes cleanly. - [ ] Add a shutdown-signal future (SIGTERM + SIGINT / Ctrl-C). - [ ] Pass it to `axum::serve(...).with_graceful_shutdown(...)`. - [ ] Optionally close the `sqlx` pool on shutdown. _Source: Foundation (Plan 0) final review — deferred, non-blocking._

logaritmisk closed this issue

2026-06-04 20:10:20 +00:00

logaritmisk referenced this issue from a commit

2026-06-05 13:21:46 +00:00

docs(plans): tier 4 hardening batch 1 (#1 #2 #21)

logaritmisk referenced this issue from a commit

2026-06-05 13:21:46 +00:00

feat(server): graceful shutdown on SIGINT/SIGTERM (#1)

logaritmisk referenced this issue from a commit

2026-06-05 13:21:46 +00:00

merge: tier 4 hardening batch 1 (#1 #2 #21)

Sign in to join this conversation.