[systemd-devel] [Patch 0/2] logind: make sure that closed sessions will be cleaned

[Zbigniew Jędrzejewski-Szmek]

On Sun, Jan 12, 2014 at 02:07:32AM +0100, Djalal Harouni wrote:
> On Sat, Jan 11, 2014 at 10:26:13PM +0100, Zbigniew Jędrzejewski-Szmek wrote:
> > On Fri, Jan 03, 2014 at 02:19:19PM +0100, Djalal Harouni wrote:
> > > On logout pam_systemd should ensures the following:
> > > "If the last concurrent session of a user ends, the $XDG_RUNTIME_DIR
> > > directory and all its contents are removed, too." from manpage.
> > > 
> > > Using git HEAD, and a simple systemd-nspawn test will show that the
> > > above is not ensured and the sessions will stay!
> > I can't reproduce this (with todays git). In the examples below, I
> > understand that you're logging in through getty. Can you test with
> > current git and/or provide a complete recipe to reproduce this?
> Yes through getty, and I guess this issue will also be visible for
> ssh/remote sessions, or sessions where TerminateSession() is not called.
Thank you for the recipe. This helps.

Indeed, in a container (without your patches), sessions remain in
"closing" state. But with your patches, systemd --user instance is
started and killed immediately during login. Not good either :)
With just the first patch, session still remain as "closing".

Also, there seems to be a regression with Fedora installs with yum:
I installed a fresh one, and there was no /var/run -> /run symlink,
the first boot was mostly broken.
-> https://bugzilla.redhat.com/show_bug.cgi?id=1053983

Zbyszek