fix(bus): silently drop publishes on closed bus instead of raising

Worker bus instances (collector, ingester) close their private buses
in finally blocks on shutdown, but stream threads holding closure
references kept calling publish after close — one `RuntimeError:
publish on closed bus` per stream line, caught by publish_safely
and logged per call, flooding server logs.

Changes:
- `UnixSocketBus.publish()` now drops post-close calls. First drop
  WARNs loudly (bus is critical infra — silent drops would hide real
  problems); subsequent drops on the same instance log at DEBUG to
  prevent the flood. Sticky `_closed_publish_warned` flag, reset
  naturally per new bus instance.
- `make_thread_safe_publisher` short-circuits on a closed bus before
  marshalling a coroutine onto the loop. Avoids the wasted scheduling
  work in the hot shutdown path.

Degradation is safe: callers go through `publish_safely`, which
already treats exceptions as 'dropped notification, DB is source of
truth.' We just stop manufacturing the exception in the first place
for a known-benign condition.
This commit is contained in:
2026-04-23 18:00:47 -04:00
parent eb2308d9e1
commit 4418608a54
3 changed files with 132 additions and 1 deletions

View File

@@ -94,6 +94,12 @@ class UnixSocketBus(BaseBus):
self._lock = asyncio.Lock()
self._write_lock = asyncio.Lock()
self._closed = False
# Sticky flag: the first publish-on-closed-bus call logs at
# WARNING so operators see that a publish was dropped; subsequent
# calls on the same instance log at DEBUG only to prevent a
# log flood when stream threads drain after close. The bus is
# critical infra, so the first warning is non-negotiable.
self._closed_publish_warned = False
# ─── Lifecycle ──────────────────────────────────────────────────────────
@@ -146,7 +152,21 @@ class UnixSocketBus(BaseBus):
event_type: str = "",
) -> None:
if self._closed:
raise RuntimeError("publish on closed bus")
# Degrade gracefully: the DB is the source of truth, the bus
# is only the notification layer. Raising here made every
# caller via publish_safely flood the logs once per stream
# line during shutdown races. First drop warns loudly;
# subsequent drops on the same instance are DEBUG-only.
if not self._closed_publish_warned:
self._closed_publish_warned = True
log.warning(
"bus.client: publish on closed bus dropped topic=%s "
"(further drops on this instance logged at DEBUG)",
topic,
)
else:
log.debug("bus.client: publish on closed bus dropped topic=%s", topic)
return
if self._writer is None:
await self.connect()
body = Event(topic=topic, payload=payload, type=event_type).to_dict()