Hi NVIDIA,
If this was addressed elsewhere, I couldn’t find it.
I updated the OS, and now I only get text-based login.
I’ve purged the existing drivers, blacklisted ‘nouveau’, set ‘nomodeset’ in ‘grub’, disabled ‘wayland’ and several other things a few times, but my login comes up text only.
I’ll try to upload '‘nvidia-bug-report.sh’, `nvidia-smi {,–query}’ and more (in the reply)
Another thought: As this box is remote to me now, I assume if ‘nvidia-smi’ shows ‘ No running processes found’ and ‘GPU-Util == 0’ that it’s not in use. Will the ‘GPU-Util’ show any use even if it’s just a login screen? Thx
These are the kernel modules and everything seems ok here, further evidenced by your proper nvidia-smi output. The problem is on X11 level: the Failed to load module “nvidia” (module does not exist, 0) log fragment is from X11 logs and refers to an X11 module, not the kernel module with the same name.
Please post the output of ls -l /usr/lib/xorg/modules/drivers/
Also, do you use any custom X11 config? Please post the contents of all files in /etc/X11/xorg.conf.d/ folder (if any).
I tried to look in /usr/lib/nvidia/ for a possible driver, but it only contains a text file alternate-install-available that tells me to use the GUI ;-)
So as it’s apparently missing, what does xserver-xorg-video-nvidia-535 provide?? Ubuntu packaging is… weird, to say least. Please post the output of dpkg -L xserver-xorg-video-nvidia-535.
I see -rw-r--r-- 1 root root 6400168 Sep 28 01:20 /usr/lib/x86_64-linux-gnu/nvidia/xorg/nvidia_drv.so
Should I try it in /usr/lib/xorg/modules/drivers/?
Yes, that is a brute-force solution that will probably work (you can try either copying it or soft-linking with ln -s /usr/lib/x86_64-linux-gnu/nvidia/xorg/nvidia_drv.so in /usr/lib/xorg/modules/drivers/ folder).
My guess is that some package configuration script failed to create such link automatically and “the proper” way to resolve this, would be to reinstall this package, but I am not familiar with Ubuntu Nviidia packaging, so I have no idea which one exactly. Perhaps other forum members will be able to advise more.
root@unstable:/usr/lib/xorg/modules/drivers# ls -l nvidia_drv.so
lrwxrwxrwx 1 root root 51 Dec 29 00:32 nvidia_drv.so -> /usr/lib/x86_64-linux-gnu/nvidia/xorg/nvidia_drv.so
Now I rebooted, but nvidia-smi still looks the same (0% GPU, no running processes). As this box is in the office, and I won’t be able to get there for at least 12 hours, what would be the best way to see if it worked? Another nvidia-bug-report.log.gz ?
yes, please: let’s see what is happening now that the X11 module is there.
Also to make it easier to browse through, please attach the latest /var/log/Xorg.0.log.
It’s stale. Like X is not even trying to start now for some reason.
I suppose some other, non-Nvidia related misconfiguration, but hard for me to say exactly what kind (I ditched Ubuntu exactly when attempting to upgrade to 24.04 a year ago as it was a complete joke of an OS IMHO, been a happy Debian user since).
What graphic environment have you chosen during installation? If it was Gnome or KDE, then you are probably using Wayland not X11, which would explain why X11 is not starting ;-] In such case I’m also unable to help more as I have exactly zero experience with Wayland. Again, hopefully other forum users will be able to advise more in such case.
I’m considering that
Most of our production runs on Debian and lately more Devuan
I think the Xorg.0.log was only logging when I tried to load nouveau driver hoping it would work.
This was an upgrade to existing Gnome environment, so it didn;t ask me.
Re Wayland
I had WaylandEnable=false uncommented in /etc/gdm3/custom.conf, thinking it disabled it.