Mess with nvidia drivers and yum update

Support
1

I let the system deal with nvidia driver updates by itself. I did not configure anything in that sense. Nevertheless, I wanted to have nvidia-settings, or more precisely the GUI that comes with it. After digging around I ended up (re-)installing the drivers following the AlmaLinux guide (A03R9, Variant I, Binary Driver). What I mean here is that I just run those commands and, I guess, hopped for the best.

Now dnf/yum does not let me do update and after using the flags --allowerasing, --skip-broken and --nobest, I get the following from yum:

Skipping packages with broken dependencies:
 dkms                                    noarch                   3.1.8-1.el9                            epel                                 85 k
 kmod-nvidia-open-dkms                   noarch                   3:570.133.20-1.el9                     cuda-rhel9-x86_64                    12 M
 nvidia-kmod-common                      noarch                   3:570.133.20-1.el9                     cuda-rhel9-x86_64                    63 M

From looking at errors from yum and dnf nvidia-plugin, it seems to be related to incompatible kernel versions, but I have no idea what to do now. It doesn’t matter to me if I keep the nvidia repo or go back to whatever almalinux was doing by default.

What should I do? I am running AlmaLinux 9.5.

2

The dnf provides *bin/nvidia-settings says that the program is in package nvidia-settings.

I do usually go to “clean slate” for NVidia reinstall:

systemctl set-default multi-user.target
dnf rm \*nvidia\*
dnf rm $(dnf rq --installonly --latest=-1)
dnf module reset nvidia-driver
systemctl reboot

When I get back into the system:

dnf install epel-release
dnf config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel9/x86_64/cuda-rhel9.repo
dnf module enable nvidia-driver:open-dkms
dnf install nvidia-driver nvidia-driver-cuda nvidia-kmod-common nvidia-modprobe nvidia-settings nvidia-xconfig dnf-plugin-nvidia kernel-devel-matched dkms
systemctl set-default graphical.target
systemctl reboot

(Although, I have not used the open driver – my card is too old.)

3

Many thanks. This was exactly what I was looking for.
Some comments:

  • Shortly after the dnf install nvidia-driver… etc step was done, a message poped up on the screen (I running the commands via ssh) saying that the GPU was bond to nouveau. I then rebooted.
  • The system now takes a long time to get to the login screen and when it finally gets to the login screen (still in multiuser mode) the terminal keeps spitting out numbers like:
    …
    [ 186.746545]
    [ 187.654875]
    …

not sure what is going on