fix(app): cancel readiness loop on lifespan shutdown (review round 2)

_readiness_refresh_loop is started with tg.start_soon and loops forever with no
shutdown_event check. anyio waits for start_soon tasks on normal task-group exit
rather than cancelling them, so graceful shutdown hung until uvicorn's timeout.
Cancel the task group's scope after teardown() to stop the loop and any
stragglers, mirroring _maybe_login_flow_cleanup.

Also document the cache ttl_seconds startup-override and the inclusive is_stale
boundary.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

nextcloud_mcp_server/app.py

@@ -434,6 +434,8 @@ def _clear_vector_sync_state() -> None:
 # Fallback public URL when NEXTCLOUD_MCP_SERVER_URL is unset (dev/local).
 _DEFAULT_MCP_SERVER_URL = "http://localhost:8000"
 
+# Pre-loop default; _readiness_refresh_loop overrides ttl_seconds at startup to
+# 2x the configured refresh interval, so bumping this value alone has no effect.
 _readiness_cache = ReadinessCache(ttl_seconds=30.0)
 
 
@@ -2186,6 +2188,12 @@ def get_app(transport: str = "streamable-http", enabled_apps: list[str] | None =
                     # Request path must not spawn into a cancelling group.
                     _vector_sync_state.eviction_task_group = None
                     await teardown()
+            # teardown() signals the sync tasks to drain, but the readiness loop
+            # runs forever and has no shutdown_event to observe. anyio waits for
+            # start_soon tasks on normal exit rather than cancelling them, so
+            # without this the lifespan shutdown hangs until uvicorn's graceful
+            # timeout. Cancel the group to stop the loop (and any stragglers).
+            tg.cancel_scope.cancel()
 
     # Health check endpoints for Kubernetes probes
     def health_live(request):

nextcloud_mcp_server/observability/readiness.py

@@ -70,6 +70,7 @@ class ReadinessCache(BaseModel):
         if not self.statuses:
             return True
         current = time.monotonic() if now is None else now
+        # Inclusive boundary: exactly ttl_seconds old counts as stale.
         return any(
             current - status.checked_at >= self.ttl_seconds
             for status in self.statuses.values()