danielgtaylor / danielgtaylor/python-betterproto

AsyncChannel produces unexpected exception on cancel

Abierto
#233 1 comentario 0 reacciones 0 asignados Ver en GitHub
bug
Lenguaje dominante
Python
Estrellas
1.8k
Forks
234
Métricas de merge de PR
Sin PR fusionados en 30 d

Descripción

Hey there,

I'm using the AsyncChannel to send messages to a GRPC server. When receiving a grpc error the client prints some non-catched errors of the sending task in the console (related to: https://github.com/danielgtaylor/python-betterproto/issues/188).

```
future: exception=ValueError('task_done() called too many times')>
Traceback (most recent call last):
File "/home/PycharmProjects/Worker/venv/lib/python3.8/site-packages/betterproto/grpc/util/async_channel.py", line 87, in __anext__
result = await self._queue.get()
File "/usr/lib/python3.8/asyncio/queues.py", line 163, in get
await getter
asyncio.exceptions.CancelledError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/PycharmProjects/Worker/venv/lib/python3.8/site-packages/betterproto/grpc/grpclib_client.py", line 166, in _send_messages
async for message in messages:
File "/home/PycharmProjects/Worker/venv/lib/python3.8/site-packages/betterproto/grpc/util/async_channel.py", line 93, in __anext__
self._queue.task_done()
File "/usr/lib/python3.8/asyncio/queues.py", line 206, in task_done
raise ValueError('task_done() called too many times')
ValueError: task_done() called too many times

```

When the server sends a grpc-error the error is handled in this except-block, which cancels the sending task:
https://github.com/danielgtaylor/python-betterproto/blob/master/src/betterproto/grpc/grpclib_client.py#L159-L161

This cancel-method results in throwing a CancelledError inside the current await of the sending task, which is the `get`-method on the queue of the AsyncChannel:
https://github.com/danielgtaylor/python-betterproto/blob/master/src/betterproto/grpc/util/async_channel.py#L87

In this case the `get`-method is raising a CancelledError even if the queue is empty. If this happens the `task_done`-method is called on an empty queue, and the above error message is shown.

A quick fix would be the following, however I haven't checked if this fix has any side effects:
```
try:
...
finally:
self._waiting_receivers -= 1
if not self._queue.empty():
self._queue.task_done()
```

Guía de contribución

Abrir la guía de contribución

Evaluación

Este issue todavía no se ha evaluado.

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.