Bug#988844: daemon_notifier_socket bind: Address already in use

[Andreas Metzler]

Control: tags -1 confirmed

On 2021-05-20 Chris Hofstaedtler <zeha at debian.org> wrote:
> Package: exim4-daemon-heavy
> Version: 4.94.2-1~bpo10+1
> Severity: important

> Dear Maintainer,

> thank you for maintaining exim, also in backports.

> I have recently upgraded from the version in buster to
> 4.94.2-1~bpo10+1 from buster-backports.

> Since then, on each service restart, I get this error in
> /var/log/exim4/paniclog:
>    2021-05-20 11:41:39.862 [1309224] daemon_notifier_socket bind: Address already in use

> Specialities about this setup:
>   /etc/default/exim4: QUEUERUNNER='separate'
[...]
> I guess upstream ticket 2616 could be used to mitigate this issue,
> if the startup scripts set the new option when QUEUERUNNER=separate:
>   https://bugs.exim.org/show_bug.cgi?id=2616

> Please consider applying and using the upstream patch on top of
> 4.94.2 for bullseye (and maybe buster-backports).

Hello Chris,

thanks for report and suggested fix.

Setting QUEUERUNNER='separate' seems to be enough to trigger the bug.
With the patch you linked to applied and globally disabled
notifier_socket I do not see the error anymore.

My current plan is to cherrypick the patch and set -oY in the initscript
for both daemon instances if QUEUERUNNER='separate' is set.

I will let -5 enter testing though, might upload to experimental for
earlier testing.

cu Andreas

-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'