[systemd-devel] Scope units get SIGKILL when stopped, not SIGTERM

[Lennart Poettering]

On Wed, 26.11.14 11:55, Michael Chapman (mike at very.puzzling.org) wrote:

Heya,

> When I stop a scope unit, it looks like all processes in it get a SIGKILL
> immediately, not a SIGTERM.
> 
> I believe this issue has been brought up before in
> http://lists.freedesktop.org/archives/systemd-devel/2014-October/024452.html,
> but there was no resolution then. That thread indicates that commit 743970d2
> was where the regression was introduced.
> 
> I noticed this problem specifically when using "reboot" from a shell (it
> happened to be over SSH, but I don't think that's significant). After
> reboot, the user's shell history did *not* contain any of the commands from
> that shell session, since it had been SIGKILLed.
> 
> Is there any solution for this problem on the horizon? Losing shell history
> is relatively minor in the grand scheme of things, but I could well imagine
> some other uses for scope units that expect an orderly shutdown.

The only solution that will really fix this for good is probably to
move things to the new unified cgroup controller logic that finally
gives us useful ways to get notifications for cgroups running empty.

Also see what I just replied here:

http://lists.freedesktop.org/archives/systemd-devel/2014-November/025734.html

Lennart

-- 
Lennart Poettering, Red Hat