Bug#935471: systemd: bogus "Process .. as been marked to be excluded from killing" warning from systemd-shutdown

[Marc Lehmann]

On Fri, Aug 23, 2019 at 08:46:12AM +0200, Michael Biebl <biebl at debian.org> wrote:
> 
> If you have ideas how to improve the code, a patch would be great.

Well, a rather obvious improvement would be to simply remove the code, as
it doesn't seem to work (I'm not doing anything fancy other than starting
a process from the initrd) and is not actually required for correctness -
it's just a bogus nag message that does more damage than it helps.

This would of course be different if it actually worked correctly. There
might be a way to detect whether something is running from the real /, but
I wouldn't know how.

Thinking about this issue actually shows a bigger problem here as well:
many packages (such as ntfs-3g) write to /run/sendsigs.omit.d, which works
with sysvinit but fails with systemd. The problem here is that fixing
ntfs-3g is non-trivial, as patching the process name form outside is...
rather difficult, and there is no way to communicate to mount that it has
to start processes with a special name. Maybe fuse/ntfs-3g would need a
special mount option for this? Or maybe systemd could just implement the
de-facto standard in this area (sendsigs.omit.d) which integrates well into
gnu/linux.

This problem doesn't affect me, as I run ntfs-3g manually and thus
can easily change the name, but it occured to me that probably many
packages which rely on sendsigs.omit.d are broken, and would incur
data loss. Looking at ubuntu, this is exactly the case there - the
ubuntu initramfs supports loop mouints out of the box, and using ntfs as
underlying fs there causes data loss at reboot/shutdown.

(This also affects debian to a lesser extent, but debians initramfs-tools
doesn't support loop mounts out of the bnox, and few people will have ntfs
as their root fs, so ntfs-3g's broken state is relatively unimportant).

> Ideally, you should submit this upstream at
> https://github.com/systemd/systemd/

I'd like to stay away from having to run non-free software just to
report a bug against systemd. I understand that this is probably not a
debian-only issue and affects systemd upstream, but I am honestly super
relieved that debian has a libre way of reporting bugs. And I clearly know
systemd much less well than the debian package maintainer. So, if this
(minor) bug report gets swept under the rug because you are too busy and
there seem to be a lot of more important issues, so be it.

-- 
                The choice of a       Deliantra, the free code+content MORPG
      -----==-     _GNU_              http://www.deliantra.net
      ----==-- _       generation
      ---==---(_)__  __ ____  __      Marc Lehmann
      --==---/ / _ \/ // /\ \/ /      schmorp at schmorp.de
      -=====/_/_//_/\_,_/ /_/\_\