|
|
#1 | |
|
Registered User
Join Date: Nov 2009
Posts: 5
|
The steps to reproduce is unknown. Any kind of operations can cause trouble. All I can say is that the system seems to stay some stable for a few hours about once in four reboots. (So my problem of stability has been improved much better than before.)
|
|
|
|
|
|
|
#2 | |
|
Registered User
Join Date: Nov 2009
Posts: 5
|
When I was using 280.13 and Gnome 2, I found a way to almost alway get to my desktop. It's to press alt+ctrl+f2 right after login to skip watching my desktop builded. Not only can I get to my desktop, but my system gets stability. Crashes occur only like once in a day.I'm using 290.10 and Gnome Shell on Fedora 16 now. Without that trick, I can at least get to my desktop with no crashes.
I don't know what causes crashes as before. After a while with nvidia driver in use, google chrome's flash starts to crash. After that crash starts, google chrome refuses to be run with segmentation fault. If you start X with other drivers than nvidia like vesa or nouveau, flash runs properly. And oddly enough, now you go back to use nvidia driver again, and you'll be able to run chrome whithout problems and see flash run properly for a while. Can I talk about an acient version 195.36.31? I tested a few combinations of configs of xorg.conf an monitors.xml. My note back then said: xorg.conf "1600x1200_65 +0+0", monitors.xml 1280x960 - no good xorg.conf "1400x1050_75 +0+0", monitors.xml 1280x960 - very stable xorg.conf "1400x1050_75 +0+0", monitors.xml 1024x768 - no good xorg.conf "1280x960_85 +0+0", monitors.xml 1280x960 - no good And the size of a window seemed to have something with its stability too. It still does. I'm under the impression it's dangerous when the width is wider than 1200. |
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Mar 2005
Posts: 1
|
I see the freezes (not crashes) with a GT520, 290.11 driver (also various other recent versions) and Fedora 16.
However the freezes are not quite total. The graphical interface is so slow as to make one think that it has totally frozen, however, I can ssh in from another computer and run top. This is a 2-core machine and top reports both gnome-shell and Xorg at over 90%. I can kill them (requires SIGKILL), but when the X server is automatically re-spawned, it uses 90% cpu again (even with gnome-shell not running). I can drop to multi-user (via "systemctl isolate multi-user.target" - need to kill the server with SIGKILL again) unload the nvidia kernel module then go back to graphical display and Xorg still uses over 90%. I have to actually re-boot to restore sane operation (which typically lasts a few hours). This is not gnome-shell specific as I have seen it with xfce as well. Seeing as killing everything, including unloading the kernel module doesn't fix things, about the only possibility left is that the kernel module is leaving the kernel itself in a bad state. |
|
|
|
|
|
#4 | |
|
Registered User
Join Date: Sep 2008
Posts: 25
|
Quote:
The freezes happen more frequently after 5 days uptime. By 9-10 you can guarantee at least one every half hour. Similarly I can SSH in. Unlike your issue though, restarting the WM (eg compiz --replace) tends to free things up. Is there a thread for this issue? |
|
|
|
|
![]() |
| Thread Tools | |
|
|