Bug#791964: gdm3: upgrade causes X session to be terminated

[Michael Biebl]

[keeping pkg-systemd into the loop here]

Am 10.07.2015 um 14:37 schrieb Michael Biebl:
> I suggest we follow up on this bug report.
> 
> I'm not yet quite sure, if this is a gdm or systemd bug and whether the
> best course of action is to simply re-introduce the workaround patch and
> drop that in stretch+1 or if we drop the reload from gdm3.postinst or if
> this indeed another corner case which is not handled properly in systemd.

Joss, bigon: looking at gdm3.postinst, I wonder why we do run
"invoke-rc.d gdm3 reload".
Isn't that potentially dangerous, say we have a major new gdm3 upgrade
with incompatible configuration and we tell the old, running gdm3 to
reload this new configuration?

Wouldn't we rather want a full-restart once all user sessions are
closed? I'm not sure if gdm3 (still) supports that.

As for the systemd specific part, I made an interesting observation,
which might be related to the problem.
Say you have gdm-3.14-7 runnig which has BusName *not* set.
"systemctl show gdm3.service -p BusName -p Type"
will return Type=simple (as expected).
Merely changing the gdm3.service file on disk, and running "systemctl
daemon-reload" *without* actually restarting gdm3, will now yield
Type=dbus
BusName=org.gnome.DisplayManager

I find this behaviour unexpected and strange, i.e. changing the Type of
a running job.


-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-systemd-maintainers/attachments/20150710/1928da7e/attachment-0002.sig>