[systemd-devel] Centos 7 TigerVNC server crashing after unfriendly shutdowns

[Prof. Dr. Michael Schefczyk]

Dear Andrei,

I very much wish I could find out how to get tigervncserver to start without abandoning the failed configuration and moving to a higher display number. Since the fourth quarter of 2014, I have installed (including reinstalls) 12 servers working as hypervisors. The problem happened to six of them which I usually reinstalled afterwards. Journalctl-output did point either to stale file handles (like the post on https://www.centos.org/forums/viewtopic.php?f=48&t=50115) or even more cryptic issues like below. In none of the cases, I could resolve the problem directly. I would very much welcome any guidance regarding how to proceed.

Regards,

Michael

-----Ursprüngliche Nachricht-----
Von: Andrei Borzenkov [mailto:arvidjaar at gmail.com] 
Gesendet: Freitag, 7. August 2015 05:39
An: Prof. Dr. Michael Schefczyk <michael at schefczyk.net>
Cc: systemd-devel at lists.freedesktop.org
Betreff: Re: [systemd-devel] Centos 7 TigerVNC server crashing after unfriendly shutdowns

В Thu, 6 Aug 2015 20:08:55 +0000
"Prof. Dr. Michael Schefczyk" <michael at schefczyk.net> пишет:

> Dear All,
> 
> I like to implement virtualization via the KVM hypervisor on Centos 7 and - knowing that some people find that funny - use a server with GUI as the basis. To access the GUI within the LAN, I like to use TigerVNC server and set the user to root, since everything happens within the LAN and since all meaningful tasks using that route require root access anyway.
> 
> Unfortunately, TigerVNC does sometimes not survive unfriendly 
> shutdowns which will occur, e.g., due to fencing. For more details, 
> you may want to see this post 
> https://www.centos.org/forums/viewtopic.php?f=48&t=50115
> 
> That renders the machine reachable only via other means (e.g. via KVM or SSH). Simple solutions, such as removing the relevant file under /tmp/.X11-unix or uninstalling and reinstalling tigervnc with a fresh configuration will not help. Sometimes it is possible to create a new vncserver service using a higher display number, but that obvouisly is not a proper way to resolve this. Much less desirable would be to install the whole server again hoping that it will not happen again qickly.
> 
> Please point me to a way to resolve this problem effectively.
> 

You need find out what is necessary to make it start; then it can be discussed if this can be integrated in service startup.  

> A typical output of journalctl -xn is the enclosed below.
> 
> Regards
> 
> Michael
> 
> -- Logs begin at Do 2015-08-06 21:02:43 CEST, end at Do 2015-08-06 
> 21:04:52 CEST. -- Aug 06 21:04:10 server1.schefczyk.local systemd[1]: 
> vncserver@:1.service: control process exited, code=exited status=98 Aug 06 21:04:10 server1.schefczyk.local systemd[1]: Failed to start Remote desktop service (VNC).
> -- Subject: Unit vncserver@:1.service has failed
> -- Defined-By: systemd
> -- Support: 
> http://lists.freedesktop.org/mailman/listinfo/systemd-devel