TCPServer crash

[apawlucz-cisco]

When not using the websocket.close() as part of the endpoint definition - it causes the TCPServer to crash, which causes drop of all the connection establish with this TCSPServer instance. E.g. multiple onging websocket connections.

This issue is very similar to: #327

The very same app can be used for repro as mentioned here: #327 (comment)

Before running the app - make sure the close line in the main_hypercorn.py is COMMENTED:

await websocket.send_bytes(data)
# Commented on purpose - no explicit close
# await websocket.close()

The issue only happens if:

1. There is a permanent Websocket connection created (app button "Connect to /permanent").
2. The 1. connection is created using HTTP2 (sometimes it's neccessary to use hard reload on the app and try several times so the connection is established using HTTP2).
3. While the permanent connection is established - then try to create another websocket connection (button "Connect to /ws").

If this happens this traceback can be seen:

    |   File "[...]/venvs/3.11/core/lib/python3.11/site-packages/hypercorn/protocol/h2.py", line 189, in handle
    |     await self._handle_events(events)
    |   File "[...]/venvs/3.11/core/lib/python3.11/site-packages/hypercorn/protocol/h2.py", line 252, in _handle_events
    |     await self.streams[event.stream_id].handle(
    |           ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
    | KeyError: 61
    +------------------------------------

This causes the TCPServer to crash and closes both websocket connections.

[FutureTraceback]

I'm very glad you have a reproducible environment, because I couldn't wrap my head around why my server (which has both REST and Websocket endpoints) occasionally crashes in production when switching from uvicorn to hypercorn (0.18.0):

ERROR: 2025-11-27 09:45:45,105 - SpawnProcess-1(14)/MainThread - /usr/local/lib/python3.11/asyncio/base_events.py:1771 - asyncio: Task exception was never retrieved
future: <Task finished name='Task-64' coro=<worker_serve.<locals>._server_callback() done, defined at /usr/local/lib/python3.11/site-packages/hypercorn/asyncio/run.py:105> exception=ExceptionGroup('unhandled errors in a TaskGroup', [KeyError(1)])>
  + Exception Group Traceback (most recent call last):
  |   File "/usr/local/lib/python3.11/site-packages/hypercorn/asyncio/run.py", line 109, in _server_callback
  |     await TCPServer(app, loop, config, context, lifespan_state, reader, writer)
  |   File "/usr/local/lib/python3.11/site-packages/hypercorn/asyncio/tcp_server.py", line 57, in run
  |     async with TaskGroup(self.loop) as task_group:
  |   File "/usr/local/lib/python3.11/site-packages/hypercorn/asyncio/task_group.py", line 75, in __aexit__
  |     await self._task_group.__aexit__(exc_type, exc_value, tb)
  |   File "/usr/local/lib/python3.11/asyncio/taskgroups.py", line 147, in __aexit__
  |     raise me from None
  | ExceptionGroup: unhandled errors in a TaskGroup (1 sub-exception)
  +-+---------------- 1 ----------------
    | Traceback (most recent call last):
    |   File "/usr/local/lib/python3.11/site-packages/hypercorn/asyncio/tcp_server.py", line 73, in run
    |     await self._read_data()
    |   File "/usr/local/lib/python3.11/site-packages/hypercorn/asyncio/tcp_server.py", line 108, in _read_data
    |     await self.protocol.handle(RawData(data))
    |   File "/usr/local/lib/python3.11/site-packages/hypercorn/protocol/__init__.py", line 66, in handle
    |     return await self.protocol.handle(event)
    |            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    |   File "/usr/local/lib/python3.11/site-packages/hypercorn/protocol/h2.py", line 190, in handle
    |     await self._handle_events(events)
    |   File "/usr/local/lib/python3.11/site-packages/hypercorn/protocol/h2.py", line 259, in _handle_events
    |     await self.streams[event.stream_id].handle(
    |           ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
    | KeyError: 1
    +------------------------------------