On 01/19/11 16:19, Elmar Prüße wrote:
On 01/19/2011 05:38 PM, Bob Doolittle wrote:
A few things to check:
- Does gdmdynamic -l list the same set of displays as utsession -p?
Not really...
Server 1: 23 displays occur in both outputs, 4 are unique to gdmdynamic and 17
to utsession.
Server 2: 18 displays occur in both outputs, 5 are unique to gdmdynamic and 1
to utsession.
- Do you see an Xnewt process for every display in gdmdynamic -l?
Also a no...
Server 1: 13 displays listed in gdmdynamic have a xnewt process, 10 don't
Server 2: 20 displays have a process, 7 don't
Interestingly, if you substract display ":0" which is happily running Xorg on the local
display, there is a matching number (9 and 6) of gdm processes with exactly one child: "[kill]
<defunct>".
There are some session types that won't utilize an Xnewt process (mostly things displaying icons
like "insert smartcard" or a multihead secondary "waiting for primary to connect to
FOG"), so it's possible that the utsession -p count would be higher. Except for display 0,
however all of the gdmdynamic sessions should be running an Xnewt.
The X server appears to be dying and GDM is not restarting it. Can you locate a core
dump? Are your users in the habit of using the "Control-Alt-Bksp-Bksp" key
sequence to abort the X server? GDM has had a history of bugs related to handling that.
-Bob
_______________________________________________
SunRay-Users mailing list
SunRay-Users@filibeto.org
http://www.filibeto.org/mailman/listinfo/sunray-users
