Bug: When the redis server is killed, faststream can't recover #2611
Description
Describe the bug
Today, my redis server was killed.
Almost immediately, it was back alive and the data were flowing properly from the sender.
But faststream stayed stuck, complaining the group was not existing.
How to reproduce
- create a subscriber with a stream and consumer group on redis
- launch the application
- kill redis + empty it (if not done by the kill)
- faststream is sending the following message:
2025-10-23T14:38:01.670669760Z 2025-10-23 16:38:01,670 ERROR - | - Message fetch error
2025-10-23T14:38:01.670708509Z Traceback (most recent call last):
2025-10-23T14:38:01.670717162Z File "/venv/lib/python3.12/site-packages/faststream/redis/subscriber/usecases/basic.py", line 94, in _consume
2025-10-23T14:38:01.670722550Z await self._get_msgs(*args)
2025-10-23T14:38:01.670727037Z File "/venv/lib/python3.12/site-packages/faststream/redis/subscriber/usecases/stream_subscriber.py", line 289, in _get_msgs
2025-10-23T14:38:01.670732145Z for stream_name, msgs in await read(self.last_id):
2025-10-23T14:38:01.670737974Z ^^^^^^^^^^^^^^^^^^^^^^^^
2025-10-23T14:38:01.670747609Z File "/venv/lib/python3.12/site-packages/redis/asyncio/client.py", line 616, in execute_command
2025-10-23T14:38:01.670758445Z return await conn.retry.call_with_retry(
2025-10-23T14:38:01.670787409Z ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-10-23T14:38:01.670793789Z File "/venv/lib/python3.12/site-packages/redis/asyncio/retry.py", line 59, in call_with_retry
2025-10-23T14:38:01.670798466Z return await do()
2025-10-23T14:38:01.670804846Z ^^^^^^^^^^
2025-10-23T14:38:01.670808882Z File "/venv/lib/python3.12/site-packages/redis/asyncio/client.py", line 590, in _send_command_parse_response
2025-10-23T14:38:01.670813128Z return await self.parse_response(conn, command_name, **options)
2025-10-23T14:38:01.670816964Z ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-10-23T14:38:01.670820890Z File "/venv/lib/python3.12/site-packages/redis/asyncio/client.py", line 637, in parse_response
2025-10-23T14:38:01.670825317Z response = await connection.read_response()
2025-10-23T14:38:01.670829794Z ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2025-10-23T14:38:01.670834331Z File "/venv/lib/python3.12/site-packages/redis/asyncio/connection.py", line 577, in read_response
2025-10-23T14:38:01.670838778Z raise response from None
2025-10-23T14:38:01.670843765Z redis.exceptions.ResponseError: NOGROUP No such key 'global_input-stream' or consumer group 'call_activity_managers' in XREADGROUP with GROUP optionExpected behavior
I expect Faststream to detect the missing group and recreate it to recover the connection, in order to keep working as usual.
Throwing an error would be appreciated as well.
Observed behavior
FastStream application is dead
Environment
Faststream 0.6.1 on alpine docker image
Additional context
Redis being an in-memory DB, I think the reader need to make sure it can recover when redis is back.