Comment # 7 on bug 952478 from
(In reply to Dr. Werner Fink from comment #3)
> You can not restart systemd-logind.service without loosing the login
> sessions!
> 
> Compare with bug #934901
> 
> Indeed systemd-logind can not do this.  There is no serialize/deserialize
> feature built into systemd-logind.
> 
> 
> Therefore zypper should *not* enforce this!

I wasn't suggesting a solution, just describing our workaround for the
symptoms, to help locating the problem.

(Since we usually experience this problem on servers, we can live with losing
the login session, usually there aren't any long-lived ones. I can see how this
would be a problem for GUIs, though.)


(In reply to Dr. Werner Fink from comment #6)
> It is simply that the running systemd-logind holds a lot of required data
> like file descriptors of sockets, devices, and some data more and if anyone
> enforce a restart then those descriptors are lost.
> 
> Therefore don't do this.  I know it is nasty.  Also I know that systemd its
> self can do a daemon-reexec and IMHO systemd-logind should have this feature
> as well.  That implies that if upstream goes to implement such a feature the
> correct way would not to stop/start but to re-execute.

For anybody arriving here via search engine: Upstream bug ID related to this
issue is probably https://github.com/systemd/systemd/issues/1163.


You are receiving this mail because: