This is a strange one to me.
Just recently, my Centos 6 has been "gentle crashing" where it doesn't crash, per se, but the screen goes black, and the login prompt pops up.
It only happens when I run Firefox and it only started happening this morning. However, it has happened a dozen times, and always when Firefox is running.
It does NOT happen when Firefox is not running. I haven't tested with Chrome yet; so that's my next test - but - the question I have is:
How do I log the events that are causing Centos to log me out of user1 and to present the login screen?
On Mon, 15 Apr 2013 19:29:05 +0000 (UTC) Rock wrote:
Just recently, my Centos 6 has been "gentle crashing" where it doesn't crash, per se, but the screen goes black, and the login prompt pops up.
It only happens when I run Firefox and it only started happening this morning. However, it has happened a dozen times, and always when Firefox is running.
Your desktop is crashing, not the entire machine.
Try renaming ~/.mozilla to something else, then run firefox and see if it still crashes.
On Mon, 15 Apr 2013 13:38:27 -0600, Frank Cox wrote:
Your desktop is crashing, not the entire machine.
Try renaming ~/.mozilla to something else, then run firefox and see if it still crashes.
Ah, that makes sense that the "os" isn't crashing, but that the "Desktop" is.
I tried with Chrome, and, guess what? No crash.
Then I uninstalled Firefox & the Tor Browser Bundle (TBB), both of which use Firefox, and I removed the ~/.mozilla directory. $ sudo yum remove firefox
I re-installed Firefox, and it crashed the desktop again: $ sudo yum install firefox
So, now I'm using Chrome (but I prefer Firefox). I must admit I had Firefox honed for protection, with noscript, ghostery, user-agent switcher, dns-flusher, etc.
So, maybe those addons were causing the problem (but I did not add anything recently).
QUESTION: Is there a Desktop crash log somewhere?
On Tue, 16 Apr 2013 00:21:33 +0000, Rock wrote:
Is there a Desktop crash log somewhere?
Given that the advice was that X was crashing due to Nvidia driver mismatch problems after my "yum update", I took a look at the ~/.xsession-errors log file before and after the crash.
And, you seem to be correct: The xsession is crashing when using Firefox (but not Chrome):
Window manager warning: Fatal IO error 11 (Resource temporarily unavailable) on display ':0'. nautilus: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. polkit-gnome-authentication-agent-1: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. nm-applet: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. rhythmbox: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gpk-update-icon: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. pan: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0. gnome-screensaver: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gdu-notification-daemon: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gnome-terminal: Fatal IO error 104 (Connection reset by peer) on X server :0.0. [14555:14555:0415/175602:ERROR:chrome_browser_main_x11.cc(62)] X IO error received (X server probably went away)
Rock wrote:
On Tue, 16 Apr 2013 00:21:33 +0000, Rock wrote:
Is there a Desktop crash log somewhere?
Given that the advice was that X was crashing due to Nvidia driver mismatch problems after my "yum update", I took a look at the ~/.xsession-errors log file before and after the crash.
And, you seem to be correct: The xsession is crashing when using Firefox (but not Chrome):
Window manager warning: Fatal IO error 11 (Resource temporarily unavailable) on display ':0'. nautilus: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. polkit-gnome-authentication-agent-1: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. nm-applet: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. rhythmbox: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gpk-update-icon: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. pan: Fatal IO error 11 (Resource temporarily unavailable) on X server :0.0. gnome-screensaver: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gdu-notification-daemon: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. gnome-terminal: Fatal IO error 104 (Connection reset by peer) on X server :0.0. [14555:14555:0415/175602:ERROR:chrome_browser_main_x11.cc(62)] X IO error received (X server probably went away)
Those messages are a result of Xorg crashing - i.e. various X11 apps complaining that Xorg has gone away ...
Any Xorg crash details will be in /var/log/Xorg.?.log - however, if Xorg has restarted since the crash, then the error should be in /var/log/Xorg.?.log.old
The error may also be in one of the log files in /var/log/gdm/
James Pearson
On Tue, 16 Apr 2013 09:55:27 +0100, James Pearson wrote:
Any Xorg crash details will be in /var/log/Xorg.?.log - however, if Xorg has restarted since the crash, then the error should be in /var/log/Xorg.?.log.old
The error may also be in one of the log files in /var/log/gdm/
Thanks James, This is helpful, especially for debugging.
I'm looking through the Xorg.?.log files in /var/log/ Xorg.{0,2,3,4,5,9}.log Xorg.{0,,2,3,4,5}.log.old
There is a lot there - but I'm not sure what I'm looking for.
Likewise, I'm looking through the files in /var/log/gdm for clues: :0-greeter.log :0-greeter.log.{1,2,3,4} :0.log :0.log.{1,2,3,4} :0-slave.log :0-slave.log.{1,2,3,4} etc.
Rock wrote:
On Tue, 16 Apr 2013 09:55:27 +0100, James Pearson wrote:
Thanks James, This is helpful, especially for debugging.
I'm looking through the Xorg.?.log files in /var/log/ Xorg.{0,2,3,4,5,9}.log Xorg.{0,,2,3,4,5}.log.old
There is a lot there - but I'm not sure what I'm looking for.
Likewise, I'm looking through the files in /var/log/gdm for clues: :0-greeter.log :0-greeter.log.{1,2,3,4} :0.log :0.log.{1,2,3,4} :0-slave.log :0-slave.log.{1,2,3,4} etc.
The numbers refer to the display X is running on - usually 0 - I would start of by looking at the end of /var/log/Xorg.0.log.old and/or /var/log/gdm/:0.log.1
James Pearson
On Mon, Apr 15, 2013 at 07:29:05PM +0000, Rock wrote:
This is a strange one to me.
Just recently, my Centos 6 has been "gentle crashing" where it doesn't crash, per se, but the screen goes black, and the login prompt pops up.
It only happens when I run Firefox and it only started happening this morning. However, it has happened a dozen times, and always when Firefox is running.
It does NOT happen when Firefox is not running. I haven't tested with Chrome yet; so that's my next test - but - the question I have is:
How do I log the events that are causing Centos to log me out of user1 and to present the login screen?
Do you have an NVIDIA graphics card?
if yes, are you using the driver package directly from NVIDIA (as opposed to onoe of the RPM versions from, e.g., epel, or wherever...
I've had what you describe happen to me under those circumstances. The nvidia drivers replace some system library, and now and then (not too often, in my experience, in fact pretty rarely) that system lib gets updated, threby trashing the Nvidia replacement.
When that happens, firing up firefox will kill X, dropping you back to the login.
if that sounds like what you see, just go rerun the nvidia installer, then you should be working again.
On Mon, 15 Apr 2013 19:07:46 -0400, Fred Smith wrote:
Do you have an NVIDIA graphics card?
Yes, I do. And, come to think of it, I "did" recently run a "yum update".
Maybe the El Repo Nvidia driver has a problem with Firefox?
I remember it was miserable installing the Nvidia driver a while ago...
Looking at my log file, here's how I had installed the Nvidia driver from the El Repo repository:
1. I identified my kernel version: $ uname -r ==> 2.6.32-279.5.1.el6.x86_64
2. I identified my graphics card $ /sbin/lspci -nn | grep VGA ==>01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GT216 [Quadro FX 880M] [10de:0a3c] (rev a2)
3. I identified the latest version of the graphics driver available: http://www.nvidia.com/object/linux-display-amd64-304.37-driver.html Note: I compared this latest version & support information with the latest available at El Repo in the next step.
4. I compared the Nvidia version with the latest available El Repo version: http://elrepo.org/tiki/Driver+Versions Note: I searched for "nvidia" to find the latest version to be "nvidia 295.40".
5. I enabled the El Repo repository: $ sudo rpm --import http://elrepo.org/RPM-GPG-KEY-elrepo.org $ sudo rpm -Uvh http://elrepo.org/elrepo-release-6-4.el6.elrepo.noarch.rpm
5. I installed the latest Nvidia driver from the El Repo repository: $ sudo yum --enablerepo elrepo install kmod-nvidia $ sudo yum --disablerepo=* --enablerepo=elrepo install nvidia-x11-drv-32bit
if yes, are you using the driver package directly from NVIDIA (as opposed to one of the RPM versions from, e.g., epel, or wherever...
I'm using the one from El Repo.
I've had what you describe happen to me under those circumstances. The nvidia drivers replace some system library, and now and then (not too often, in my experience, in fact pretty rarely) that system lib gets updated, threby trashing the Nvidia replacement.
Come to think of it, I 'did' recently run a "yum update"!
When that happens, firing up firefox will kill X, dropping you back to the login.
if that sounds like what you see, just go rerun the nvidia installer, then you should be working again.
So, you suggest I install the Nvidia drivers directly from Nvidia and not the Nvidia drivers from the El Repo repository?
Note: I'm not sure I know how to do that.
On Tue, 16 Apr 2013 00:30:15 +0000 (UTC) Rock RockSockDoc@gmail.com wrote:
So, you suggest I install the Nvidia drivers directly from Nvidia and not the Nvidia drivers from the El Repo repository?
Note: I'm not sure I know how to do that.
I too would suggest that you install the Nvidia drivers from the Nvidia site. That is what I do. You will need to install the kernel headers in order to install the drivers from Nvidia site.
Then make the Nvidia driver executable and run it. Follow the instruction, which are not much and you good to go. Also be aware every time you upgrade the kernel you are going to have to install the driver new again.
-- Regards Robert
Linux The adventure of a lifetime.
Linux User #296285 Get Counted http://linuxcounter.net/
On Mon, 15 Apr 2013 20:50:12 -0400, Robert wrote:
I too would suggest that you install the Nvidia drivers from the Nvidia site. That is what I do. You will need to install the kernel headers in order to install the drivers from Nvidia site.
Thanks for the advice to install the Nvidia drivers.
I'm just about to reboot, so, before I forget the details, here's my log file, sufficient for the next person to follow (including myself) in the future:
I had to re-install Firefox to prove that the desktop was crashing: $ sudo yum install firefox ==> firefox.x86_64 0:17.0.5-1.el6.centos
I cleared the ~/.xsession-errors file so that I could start fresh: $ mv ~/.xsession-errors /tmp/xsession_before
I rebooted so as to start a fresh xsession: $ sudo reboot
I ran Firefox until the Desktop crashes again: $ firefox (it didn't take long for the Desktop to crash)
I logged back into the Desktop and compared the xsession logs: $ cp ~/.xsession-errors /tmp/xsession_after $ diff /tmp/xsession_before /tmp/xsession_after Found lots of stuff, for example: ==> Window manager warning: Fatal IO error 11 (Resource temporarily unavailable) ==> nautilus: Fatal IO error 11 (Resource temporarily unavailable) on X server :0. etc.
So, it's now time to update the graphics card driver.
Identify your kernel: $ uname -r ==> 2.6.32-358.2.1.el6.x86_64
Identify your graphics card: $ /sbin/lspci -nn | grep VGA ==> 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GT216 [Quadro FX 880M] [10de:0a3c] (rev a2)
Identify the installed version of Nvidia drivers: $ $ sudo updatedb; locate libvdpau_nvidia.so ==> /usr/lib/vdpau/libvdpau_nvidia.so.304.37 ==> /usr/lib64/vdpau/libvdpau_nvidia.so.304.37
Go to Nvidia support: http://www.nvidia.com/page/support.html
Press on "Download Drivers" (that brings you to): http://www.nvidia.com/Download/.aspx?lang=en-us
Enter the following information: - Product Type: Quadro - Product Series: Quadro FX Series (Notebooks) <== important to get right! - Product: Quadro FX 880M - Operating System: Linux 64-bit - Download Type: Linux Long Lived Driver - Language: English (US) Press the "Search" button (that brings you to): ==> http://www.nvidia.com/object/linux-display-amd64-310.44-driver.html - Version: 310.44 - Release Date: 2013.04.02 - Operating System: Linux 64-bit - Language: English (U.S.) - File Size: 65.2 MB Press the "Download" button (save the file but don't install it).
Compare the Nvidia version with the version at the El Repo repository: http://elrepo.org/tiki/Driver+Versions Search for "nvidia" to find the latest version, which is: - Kmod Package: nvidia - ElRepo Kmod Version: 310.44
Since the El Repo version is up to date, let's use it:
Enable the El Repo repository (if not already installed): $ sudo rpm --import http://elrepo.org/RPM-GPG-KEY-elrepo.org $ sudo rpm -Uvh http://elrepo.org/elrepo-release-6-4.el6.elrepo.noarch.rpm Note: Skip this step if El Repo is already enabled.
Install that latest version from the El Repo repository (do not use ATrpms!): $ sudo yum --enablerepo elrepo install kmod-nvidia $ sudo yum --disablerepo=* --enablerepo=elrepo install nvidia-x11-drv-32bit ==> Updated: kmod-nvidia.x86_64 0:310.44-1.el6.elrepo ==> Dependency Updated: nvidia-x11-drv.x86_64 0:310.44-1.el6.elrepo ==> Dependency Updated: nvidia-x11-drv-32bit.x86_64 0:310.44-1.el6.elrepo
Or, if you want the absolute latest in the El Repo testing repository: $ sudo yum --enablerepo elrepo-testing install kmod-nvidia $ sudo yum --disablerepo=* --enablerepo=elrepo-testing install nvidia-x11-drv-32bit
Test if you have installed the desired driver: $ $ sudo updatedb; locate libvdpau_nvidia.so ==> /usr/lib/vdpau/libvdpau_nvidia.so.310.44 ==> /usr/lib64/vdpau/libvdpau_nvidia.so.310.44
Reboot.
Robert wrote:
On Tue, 16 Apr 2013 00:30:15 +0000 (UTC) Rock RockSockDoc@gmail.com wrote:
So, you suggest I install the Nvidia drivers directly from Nvidia and not the Nvidia drivers from the El Repo repository?
Note: I'm not sure I know how to do that.
I too would suggest that you install the Nvidia drivers from the Nvidia site.
I would definitely advise you *not* to do that. As someone pointed out in this thread, the nvidia-provided driver can mess with some system libs. The elrepo-provided one does not, and if it did you'ld know about it as rpm would detect a conflict. Having the driver in an rpm is a huge plus. ... ah, I see in a later email you went the elrepo way, well done!
Note that there are now 4 branches of the nvidia driver (3 legacy + latest), each supporting different devices. To know which branch you need, simply install nvidia-detect from elrepo and run it. http://elrepo.org/tiki/nvidia-detect
On Tue, 16 Apr 2013 09:19:23 +0200, Nicolas Thierry-Mieg wrote:
Note that there are now 4 branches of the nvidia driver (3 legacy + latest), each supporting different devices. To know which branch you need, simply install nvidia-detect from elrepo and run it. http://elrepo.org/tiki/nvidia-detect
Thanks!
The results from that program are nice because it is problematic to obtain the correct driver version from the Nvidia support page (due to multiple Quadro FX entries, only one of which is the correct one - so one can easily conclude the wrong driver from the Nvidia support pages).
I have, AFAIK, the Nvidia "Quadro FX 880M" graphics card: $ /sbin/lspci -nn | grep VGA ==> 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GT216 [Quadro FX 880M] [10de:0a3c] (rev a2)
$ sudo yum --enablerepo elrepo install nvidia-detect ==> Installed: nvidia-detect.x86_64 0:310.40-1.el6.elrepo
$ nvidia-detect Probing for supported NVIDIA devices... [10de:0a3c] NVIDIA Corporation GT216 [Quadro FX 880M] This device requires the current 310.40 NVIDIA driver (kmod-nvidia).