GNU bug report logs - #76338
shepherd restarts service during shutdown and then waits forever

Previous Next

Full log

Message #55 received at 76338 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
Cc: Dariqq <dariqq <at> posteo.net>, 76338 <at> debbugs.gnu.org, 76368 <at> debbugs.gnu.org
Subject: Re: bug#76368: Cannot reboot when virtlogd and containerd is running?
Date: Sun, 23 Feb 2025 22:51:49 +0100

Hi Maxim,

Maxim Cournoyer <maxim.cournoyer <at> gmail.com> skribis:

> I was wondering if this shouldn't be part of an implicit default %shepherd-default-requires that
> could be overridden if needed... but I think even in systemd land you
> must explicitly specify wanted using e.g. 'WantedBy=multi-user.target'
> in your systemd unit definition, so I guess it's fine that it must be
> explicitly specified.

I almost changed the default value of ‘requirement’ to '(user-processes)
in <shepherd-service>, but then I thought this would need more testing
and discussion: surely there are places in Guix and in code/config out
there that relies on the current default value.

> It'd be nice though, that in the case where it's missing, if it didn't
> cause a hang upon shutdown.  I don't think anything should be able to
> hang the system when calling 'shutdown', no matter what (perhaps there
> count be some timeout... with a warning like:

Even better: I think shepherd just shouldn’t try to respawn anything
while it’s shutting down.

Not depending on ‘user-processes’ could still cause problems such as
“device or resource busy” when trying to unmount file systems, but at
least it would lead to that infinite loop.

Thanks for your feedback,
Ludo’.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.