Add WS heartbeat and harden broadcast against dead sockets (#76)

Messages could silently stop arriving until a room switch happened to
paper over it with a REST refetch. Root cause: a connection that dies
without a clean close (laptop sleep, an idle connection recycled by a
proxy/NAT) looks identical to a healthy-but-quiet one -- readyState
stays OPEN, onclose never fires, so the already-correct reconnect and
rejoin logic never runs.

The backend now pings every 30s and the frontend force-closes the
socket if nothing (any frame, not just a pong) arrives within 65s,
routing straight through the existing reconnect/rejoin path instead of
needing separate recovery logic.

Also found and fixed a more severe adjacent bug while tracing this:
ConnectionManager.broadcast()/send_to_user() had no per-socket error
handling, so one dead or racing socket could raise inside the app's
single, never-restarted Redis pubsub listener task and silently kill
live delivery for every room and every user on that process until
restart. Now caught and logged per-socket instead of propagating.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

backend/app/ws/broadcaster.py

@@ -1,10 +1,13 @@
 import json
+import logging
 import uuid
 
 from redis.asyncio import Redis
 
 from app.ws.connection_manager import ConnectionManager
 
+logger = logging.getLogger(__name__)
+
 ROOM_CHANNEL_PREFIX = "room:"
 USER_CHANNEL_PREFIX = "user:"
 
@@ -43,14 +46,26 @@ class Broadcaster:
             async for message in pubsub.listen():
                 if message["type"] != "pmessage":
                     continue
-                channel = message["channel"]
-                payload = json.loads(message["data"])
-                if channel.startswith(ROOM_CHANNEL_PREFIX):
-                    room_id = uuid.UUID(channel.removeprefix(ROOM_CHANNEL_PREFIX))
-                    await self._manager.broadcast(room_id, payload)
-                elif channel.startswith(USER_CHANNEL_PREFIX):
-                    user_id = uuid.UUID(channel.removeprefix(USER_CHANNEL_PREFIX))
-                    await self._manager.send_to_user(user_id, payload)
+                # #76: this is the one pubsub listener for the whole process
+                # (see main.py's lifespan) -- ConnectionManager.broadcast/
+                # send_to_user already guard against one dead socket taking
+                # the rest of a single fan-out down, but this belt-and-
+                # suspenders catch is for anything else unexpected (a
+                # malformed payload, e.g.) doing the same. An uncaught
+                # exception here previously meant *this whole async for*
+                # loop died silently -- no more live delivery to any room on
+                # this instance, for anyone, until the process restarted.
+                try:
+                    channel = message["channel"]
+                    payload = json.loads(message["data"])
+                    if channel.startswith(ROOM_CHANNEL_PREFIX):
+                        room_id = uuid.UUID(channel.removeprefix(ROOM_CHANNEL_PREFIX))
+                        await self._manager.broadcast(room_id, payload)
+                    elif channel.startswith(USER_CHANNEL_PREFIX):
+                        user_id = uuid.UUID(channel.removeprefix(USER_CHANNEL_PREFIX))
+                        await self._manager.send_to_user(user_id, payload)
+                except Exception:
+                    logger.error("Error handling pubsub message on channel %s", message.get("channel"), exc_info=True)
         finally:
             await pubsub.punsubscribe(f"{ROOM_CHANNEL_PREFIX}*", f"{USER_CHANNEL_PREFIX}*")
             await pubsub.aclose()

backend/app/ws/chat.py

@@ -1,3 +1,4 @@
+import asyncio
 import uuid
 
 from fastapi import APIRouter, Depends, WebSocket, WebSocketDisconnect
@@ -31,6 +32,18 @@ router = APIRouter(tags=["ws"])
 
 WS_UNAUTHENTICATED = 4401
 
+# #76: comfortably under the ~60s idle-connection timeout common to reverse
+# proxies/load balancers sitting in front of this endpoint, and short enough
+# that the frontend's own zombie-detection watchdog (useChatSocket.ts) can
+# notice a dead connection and reconnect well within a session -- a socket
+# can go silently dead (no close frame at all, e.g. a laptop sleeping or a
+# NAT dropping an idle mapping) without this, and no live message reaches it
+# again until *something* else finally triggers a close, which used to be
+# "switch rooms and back" purely by accident (that remounts the message
+# list, which refetches history over plain REST -- unrelated to the socket
+# itself actually recovering).
+WS_PING_INTERVAL_SECONDS = 30
+
 
 class ClientEnvelope(BaseModel):
     type: str
@@ -124,7 +137,18 @@ async def chat_endpoint(websocket: WebSocket, db: AsyncSession = Depends(get_db)
 
     try:
         while True:
-            raw = await websocket.receive_json()
+            try:
+                raw = await asyncio.wait_for(
+                    websocket.receive_json(), timeout=WS_PING_INTERVAL_SECONDS
+                )
+            except asyncio.TimeoutError:
+                # Nothing received in a while -- ping rather than assume the
+                # connection's still good. A genuinely dead socket fails this
+                # send (raising, caught below same as any other disconnect);
+                # a merely quiet one just gets proded and the loop continues
+                # waiting.
+                await websocket.send_json({"type": "ping"})
+                continue
             try:
                 try:
                     envelope = ClientEnvelope.model_validate(raw)
@@ -146,6 +170,14 @@ async def chat_endpoint(websocket: WebSocket, db: AsyncSession = Depends(get_db)
                     joined_rooms.add(envelope.room_id)
                     await websocket.send_json({"type": "joined", "room_id": str(envelope.room_id)})
 
+                elif envelope.type == "pong":
+                    # #76: the client's reply to our "ping" above -- nothing
+                    # to do with it. Its only purpose is resetting the
+                    # frontend's own liveness watchdog; the timeout on the
+                    # receive_json() above already gets reset by receiving
+                    # any frame at all, this one included.
+                    pass
+
                 elif envelope.type == "leave":
                     if envelope.room_id is None:
                         await websocket.send_json({"type": "error", "detail": "room_id required"})

backend/app/ws/connection_manager.py

@@ -1,8 +1,11 @@
+import logging
 import uuid
 from collections import defaultdict
 
 from fastapi import WebSocket
 
+logger = logging.getLogger(__name__)
+
 
 class ConnectionManager:
     """Local, single-process WebSocket socket registry.
@@ -43,8 +46,22 @@ class ConnectionManager:
 
     async def broadcast(self, room_id: uuid.UUID, payload: dict) -> None:
         for websocket in list(self._rooms.get(room_id, ())):
-            await websocket.send_json(payload)
+            # #76: one dead/racing socket (e.g. its own disconnect cleanup
+            # hasn't finished removing it from _rooms yet) must not take the
+            # rest of this broadcast down with it -- this runs inside
+            # Broadcaster.listen()'s single, never-restarted per-process
+            # pubsub loop, so an unhandled exception here previously meant
+            # one bad socket silently killed *live delivery for every room
+            # on this instance*, not just the one socket, until the process
+            # was restarted.
+            try:
+                await websocket.send_json(payload)
+            except Exception:
+                logger.warning("Dropping dead socket during broadcast to room %s", room_id, exc_info=True)
 
     async def send_to_user(self, user_id: uuid.UUID, payload: dict) -> None:
         for websocket in list(self._users.get(user_id, ())):
-            await websocket.send_json(payload)
+            try:
+                await websocket.send_json(payload)
+            except Exception:
+                logger.warning("Dropping dead socket sending to user %s", user_id, exc_info=True)

backend/tests/test_ws_chat.py

@@ -109,6 +109,24 @@ def test_history_tiebreaks_identical_timestamps_by_id(ws_client):
     assert [m["content"] for m in history] == [m["content"] for m in expected_order]
 
 
+def test_ws_sends_heartbeat_ping(ws_client, monkeypatch):
+    # #76: with nothing else sent on the connection, a "ping" must arrive
+    # proactively once the idle timeout elapses -- shrunk here so the test
+    # doesn't actually wait the real 30s. Replying "pong" (the frontend's
+    # own behavior) must be accepted silently and leave the connection
+    # fully usable afterward, not treated as an unknown message type.
+    monkeypatch.setattr("app.ws.chat.WS_PING_INTERVAL_SECONDS", 0.05)
+    username = _unique("alice")
+    _register(ws_client, username=username)
+    room = ws_client.post("/api/rooms", json={"name": _unique("general")}).json()
+
+    with ws_client.websocket_connect("/ws/chat") as ws:
+        assert ws.receive_json() == {"type": "ping"}
+        ws.send_json({"type": "pong"})
+        ws.send_json({"type": "join", "room_id": room["id"]})
+        assert ws.receive_json() == {"type": "joined", "room_id": room["id"]}
+
+
 def test_ws_message_without_join_errors(ws_client):
     _register(ws_client, username=_unique("alice"))
     room = ws_client.post("/api/rooms", json={"name": _unique("general")}).json()