atproto relay implementation in zig
zlay.waow.tech
fix GC thread teardown race: join instead of detach, mark DB success from gc()
- keep gc_thread handle and join it during shutdown before dp.deinit() runs —
dp is stack-owned, detaching left a use-after-free window
- add markDbSuccess() call at end of gc() so the health signal isn't solely
dependent on uidForDid (event ingestion path)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
+5
-2
+2
src/event_log.zig
+2
src/event_log.zig
+3
-2
src/main.zig
+3
-2
src/main.zig
···
297
297
log.err("failed to start GC thread: {s}", .{@errorName(err)});
298
298
return err;
299
299
};
300
-
gc_thread.detach();
301
300
302
301
// wire HTTP fallback into broadcaster (all API endpoints served on WS port)
303
302
var http_context = api.HttpContext{
···
366
365
ws_listener.deinit(io);
367
366
server_future.cancel(io);
368
367
369
-
// GC thread is detached and checks shutdown_flag — no cancel needed
368
+
// join GC thread — it checks shutdown_flag and will exit its sleep loop.
369
+
// must complete before dp.deinit() runs (dp is stack-owned).
370
+
gc_thread.join();
370
371
371
372
// cancel broadcaster fiber (shutdown flag already set, it will drain remaining)
372
373
broadcast_future.cancel(io);