Package:
release.debian.org
Severity: normal
Tags: bookworm
User:
release.d...@packages.debian.org
Usertags: pu
X-Debbugs-Cc:
python-cha...@packages.debian.org, Michael Fladischer <
fl...@debian.org>
Control: affects -1 + src:python-channels-redis
[ Reason ]
The version of python-channels-redis in bookworm suffers from
https://bugs.debian.org/1027387 /
https://github.com/django/channels_redis/issues/332, which was
introduced in 4.0.0 and is a regression from bullseye. I ran into this
while working on debusine.
[ Impact ]
I believe that any application that (a) runs Celery, (b) uses
channels-redis, and (c) sends to a channel via async_to_sync in a Celery
task without explicitly closing redis pools will encounter RuntimeErrors
as per the upstream bug report above.
I was able to work around this in my application
(
https://salsa.debian.org/freexian-team/debusine/-/merge_requests/594),
but I lost a few hours to figuring this out and it would be good if
others didn't have to.
[ Tests ]
It's covered by the upstream test suite, which is run via autopkgtest.
I also confirmed that this change fixes the problems I ran into in
debusine, without applying any workarounds.
[ Risks ]
The alternative is letting applications work around this individually,
which I think on balance is probably worse - I initially tried switching
to RedisPubSubChannelLayer instead, which did more or less work but
required various other rather strange workarounds.
[ Checklist ]
[x] *all* changes are documented in the d/changelog
[x] I reviewed all changes and I approve them
[x] attach debdiff against the package in (old)stable
[x] the issue is verified as fixed in unstable
[ Changes ]
This is a straight cherry-pick of an upstream change to close redis
connection pools when closing the asyncio loop.
--
Colin Watson (he/him) [
cjwa...@debian.org]