fix(kanban): treat dashboard event-stream cancellation as normal shutdown (#20790)

plugins/kanban/dashboard/plugin_api.py

@@ -1521,6 +1521,13 @@ def _fetch_new(cursor_val: int) -> tuple[int, list[dict]]:
             await asyncio.sleep(_EVENT_POLL_SECONDS)
     except WebSocketDisconnect:
         return
+    except asyncio.CancelledError:
+        # Normal shutdown path: dashboard process exit (Ctrl-C) cancels the
+        # websocket task while it is sleeping in the poll loop.
+        # CancelledError is a BaseException in 3.8+ so the bare Exception
+        # handler below would not catch it; without this clause Uvicorn
+        # surfaces the cancellation as an application traceback. Quiet it.
+        return
     except Exception as exc:  # defensive: never crash the dashboard worker
         log.warning("Kanban event stream error: %s", exc)
         try:

scripts/release.py

@@ -79,6 +79,7 @@
     "51599529+stephen0110@users.noreply.github.com": "stephen0110",
     "265632032+sonic-netizen@users.noreply.github.com": "sonic-netizen",
     "82531659+mwnickerson@users.noreply.github.com": "mwnickerson",
+    "sandrohub013@gmail.com": "SandroHub013",
     "154585401+LeonSGP43@users.noreply.github.com": "LeonSGP43",
     "zjtan1@gmail.com": "zeejaytan",
     "asslaenn5@gmail.com": "Aslaaen",
@@ -455,6 +456,7 @@
     "51599529+stephen0110@users.noreply.github.com": "stephen0110",
     "265632032+sonic-netizen@users.noreply.github.com": "sonic-netizen",
     "82531659+mwnickerson@users.noreply.github.com": "mwnickerson",
+    "sandrohub013@gmail.com": "SandroHub013",
     "h3057183414@gmail.com": "CoreyNoDream",
     "franksong2702@gmail.com": "franksong2702",
     "673088860@qq.com": "ambition0802",

tests/plugins/test_kanban_dashboard_plugin.py

@@ -553,6 +553,67 @@ def test_ws_events_rejects_when_token_required(tmp_path, monkeypatch):
         assert ws is not None  # handshake succeeded
 
 
+def test_ws_events_swallows_cancellation_on_shutdown(tmp_path, monkeypatch):
+    """``asyncio.CancelledError`` while sleeping in the poll loop is the
+    normal uvicorn-shutdown path (``BaseException``, so the bare
+    ``except Exception:`` does NOT catch it). Without the explicit
+    clause the cancellation surfaces as an application traceback.
+
+    Regression test for #20790 (fix in #20938). Drives the coroutine
+    directly (rather than through FastAPI TestClient) so we can observe
+    the cancellation outcome deterministically.
+    """
+    import asyncio
+    import types
+    import sys as _sys
+
+    home = tmp_path / ".hermes"
+    home.mkdir()
+    monkeypatch.setenv("HERMES_HOME", str(home))
+    monkeypatch.setattr(Path, "home", lambda: tmp_path)
+    kb.init_db()
+
+    # Short-circuit the token check — this test is about the cancellation
+    # path, not auth.
+    import plugins.kanban.dashboard.plugin_api as pa
+    monkeypatch.setattr(pa, "_check_ws_token", lambda t: True)
+
+    class _FakeWS:
+        def __init__(self):
+            self.query_params = {"token": "x", "since": "0"}
+            self.accepted = False
+            self.closed = False
+
+        async def accept(self):
+            self.accepted = True
+
+        async def send_json(self, data):
+            pass
+
+        async def close(self, code=None):
+            self.closed = True
+
+    async def _run():
+        ws = _FakeWS()
+        task = asyncio.create_task(pa.stream_events(ws))
+        # Give the handler a tick to accept + start polling.
+        await asyncio.sleep(0.05)
+        assert ws.accepted is True
+        task.cancel()
+        # stream_events should swallow CancelledError and return cleanly.
+        # If it doesn't, this await re-raises the CancelledError.
+        result = await task
+        return result, ws
+
+    result, ws = asyncio.run(_run())
+    assert result is None, (
+        f"stream_events should return cleanly after cancellation, got {result!r}"
+    )
+    # The bug symptom was a traceback; we don't assert on stderr because
+    # capturing asyncio's internal "exception was never retrieved" logging
+    # is flaky. The assertion that matters is: no CancelledError escaped.
+
+
 # ---------------------------------------------------------------------------
 # Bulk actions
 # ---------------------------------------------------------------------------