Open dearkafka opened 1 month ago
Also rabbit is cluster
So for same processing code that I have, same time and resource consumption, for Running FastStream 0.4.3 with CPython 3.11.7 on Linux - it works. I have a hunch that > 5 does not work for me, but maybe I'm doing something wrong
@dearkafka can you, please, check it with 0.5.11
? I can't reproducer your exception now
Describe the bug
So I have a complex setup, I have a RabbitBroker, Header Exchange and very long process in the worker, like 10-20 minutes, ML related. So when training finished, its time to publish task from subscriber to same exchange but different header. there it fails with errors:
How to reproduce
Problem is, without this training logic - everything works. Code above works without fail. However I can not comprehend how this training logic that I have can disrupt connection to rabbit (?). Even in theory. because channel is suddenly closed. Should I add very high timeout to publish? This also something that has not happened in previous versions of faststream and than I tried Middlewares and it became like this.
and to be clear app does not exit
Environment
Running FastStream 0.5.10 with CPython 3.11.7 on Linux