Nvidia problem: Failed to assign any connected display devices to X screen 0

Latest response

Hi there,

After having spent hours of trial and error I am still not able to solve this problem. I tried several solutions I found on the internet --> nothing worked for me. Now I really hope somebody here can help me.
I have a HPZ820 workstation which used to work well last week. Over the weekend I shut it down to save some energy.
When I tried to turn it on yesterday it wasn't able to boot to the graphical environment. There was only a blinking cursor at the upper left corner. Then I tried to boot into runlevel 3 which worked well. From there i tried to start X which failed. The error log is attached in the file Xorg.txt.
If I delete the xorg.conf I am able to start X. But everytime I write a new xorg.conf (which is done by nvidia with nvidia-xorg) X can't be started and the mentioned log file is written.
And that is my problem.
I have really no clue what is the reason for that since I didn't change anything besides a few updates from RH (The updates should be listed in the attached file screenshot.png). Before the weekend-shutdown everything worked well with the Nvidia driver. (It is the prop. Nvidia driver which is recommended for CUDA use.)

Do you have an idea what my problem could be?

Your help is really appreciated. Thanks a lot in advance.

Best regards, Stefan Arzbacher

Attachments

Responses

Hi there,

my problem is fixed. I feel like the dumbest person on earth now ;)
It turned out, that the cause for the problem was a loosely attached monitor cable.
Somebody must have touched the cable over weekend, which deattached it partially.
After I reattached it properly everything was fine again.

Strange but true.

Best regards, Stefan

That's pretty weird. Thanks for letting us know what resolved the issue for you, Stefan.

Hi Stefan,

Things like that happen to many of us... I think that one happened to me last year.

For those times it's not a simple monitor connector that's loose...
On another note, I've done something on my satellite servers for my workstations with NVIDIA or ATI graphics cards so that after a known kernel update, the graphics driver will reload automatically without administrator intervention.

I created a /root/prep_graphics_card_after_kernel.sh script that
- changes init 5 to init 3 in /etc/inittab
- Changes /etc/rc3.d/s999nvidia_update.sh to /etc/rc3.d/S999nvidia_update
-- or /etc/rc3.d/s999ati_update.sh to /etc/rc3.d/S999ati_update.sh
- Verifies the proper nvidia driver is downloaded (on my satellite server), or downloads it if it is not there already. The ati driver typically doesn't need to be redownloaded in my environment.

The workstations then get patched including a kernel update (this bit in this post is only when the kernel needs to be updated and nvidia would then not be available due to the new kernel).

  • I have an ATI and NVIDIA configuration file group that I've subscribed workstations properly to. I have system groups defined for each collection. Those configuration groups have the needed files such as /root/prep_graphics_card_after_kernel.sh and /etc/rc3.d/S999nvidia_update.sh (or /etc/rc3.d/S999ati_update.sh)

The /root/prep_graphics_card_after_kernel.sh is ran before a kernel update. It changes /etc/inittab to run level 3 and activates the S999 script to uppercase.

After the system reboots, since /etc/inittab will now go to run level 3, the below script runs and...
-- The the nvidia binary runs at init 3 with (not the full output of the script)
(The "-q" below means take all defaults, ask no questions)

/the/path/NVIDIA-[version] --accept-license -q' 

-- The script above changes /etc/inittab back to run level 5 and if it runs clean, it executes the below to get the system back to init 5.

/sbin/init 5 

Out of the dozens of workstations, typically only one might require manual intervention due to some one-time anomoly.

The /etc/rc3.d/S999ati_update.sh works a tad differently, it merely re-runs the needed script to re-invoke the proper /etc/X11/xorg.conf file for either dual or single 'head' monitor setup, then goes to init 5 and resets /etc/inittab to init 5 as well.

Regards,
Remmele

Hey Remmele - I'm sure you're as surprised as I that a loose cable produced the results Stefan experienced. It did make me wonder: does the Nvidia driver attempt to probe the monitor for dimensions/rates/etc??? And did it get a return that it didn't like and default to the "flake-out" setting? Very bizarre either way. Brought back memories of kmod-nvidia and akmod-nvidia fun ;-)

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.