liviu
(diLivio)
February 7, 2025, 11:07am
1
Hello all,
I am trying to install akmod-nvidia but I succesfully failed
I was following this howto :
So the output is:
> # sudo dnf install akmod-nvidia
>
> Updating and loading repositories:
> Repositories loaded.
> Package Arch Version Repository Size
> Installing:
> akmod-nvidia x86_64 3:565.77-1.fc41 rpmfusion-non 90.9 KiB
> Installing dependencies:
> dkms noarch 3.1.4-3.fc41 updates 187.6 KiB
> kmod-nvidia-open-dkms noarch 3:570.86.15-1.fc41 nvidia 100.2 MiB
> nvidia-kmod-common noarch 3:570.86.15-1.fc41 nvidia 87.4 MiB
> nvidia-modprobe x86_64 3:570.86.15-1.fc41 nvidia 54.8 KiB
> xorg-x11-drv-nvidia-kmodsrc x86_64 3:565.77-3.fc41 rpmfusion-non 56.9 MiB
>
> Transaction Summary:
> Installing: 6 packages
>
> Total size of inbound packages is 131 MiB. Need to download 0 B.
> After this operation, 245 MiB extra will be used (install 245 MiB, remove 0 B).
> Is this ok [y/N]: y
> [1/1] akmod-nvidia-3:565.77-1.fc41.x86_64 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> [1/2] xorg-x11-drv-nvidia-kmodsrc-3:565.77 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> [1/3] nvidia-kmod-common-3:570.86.15-1.fc4 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> [1/4] nvidia-modprobe-3:570.86.15-1.fc41.x 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> [1/5] kmod-nvidia-open-dkms-3:570.86.15-1. 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> [1/6] dkms-0:3.1.4-3.fc41.noarch 100% | 0.0 B/s | 0.0 B | 00m00s
> >>> Already downloaded
> -----------------------------------------------------------------------------------
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> -----------------------------------------------------------------------------------
> [7/7] Total 100% | 0.0 B/s | 0.0 B | 00m00s
> Failed to download files
> Librepo error: Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/fedora41/x86_64/7fa2af80.pub (IP: 104.126.37.137)
Any idea please how do I fix this?
Much appreciated!
I believe I recently have seen messages about problems with the upcoming switch to the latest 570.86 driver and problems with the packaging.
I note that you are attempting to install from the developer version of the nvidia repo so it may be that you are trying to install a version that is not yet ready for release.
Please show us the output of dnf repolist
(as preformatted text using the </>
button on the toolbar) [not the block quote "
button]
liviu
(diLivio)
February 7, 2025, 11:37am
4
Thanks but it still happen:
> liviu@bobdenaut î‚° ~ î‚° sudo dnf install akmod-nvidia
> Updating and loading repositories:
> Repositories loaded.
> Package Arch Version Repository Size
> Installing:
> akmod-nvidia x86_64 3:565.77-1.fc41 rpmfusion-non 90.9 KiB
> Installing dependencies:
> dkms noarch 3.1.4-3.fc41 updates 187.6 KiB
> kmod-nvidia-open-dkms noarch 3:570.86.15-1.fc41 nvidia 100.2 MiB
> nvidia-kmod-common noarch 3:570.86.15-1.fc41 nvidia 87.4 MiB
> nvidia-modprobe x86_64 3:570.86.15-1.fc41 nvidia 54.8 KiB
> xorg-x11-drv-nvidia-kmodsrc x86_64 3:565.77-3.fc41 rpmfusion-non 56.9 MiB
>
> Transaction Summary:
> Installing: 6 packages
>
> Total size of inbound packages is 131 MiB. Need to download 131 MiB.
> After this operation, 245 MiB extra will be used (install 245 MiB, remove 0 B).
> Is this ok [y/N]: y
> [1/6] akmod-nvidia-3:565.77-1.fc41.x86_64 100% | 121.8 KiB/s | 40.3 KiB | 00m00s
> [2/6] kmod-nvidia-open-dkms-3:570.86.15-1. 100% | 5.6 MiB/s | 11.7 MiB | 00m02s
> [3/6] nvidia-modprobe-3:570.86.15-1.fc41.x 100% | 41.9 KiB/s | 29.1 KiB | 00m01s
> [4/6] dkms-0:3.1.4-3.fc41.noarch 100% | 352.3 KiB/s | 83.1 KiB | 00m00s
> [5/6] nvidia-kmod-common-3:570.86.15-1.fc4 100% | 17.3 MiB/s | 62.5 MiB | 00m04s
> [6/6] xorg-x11-drv-nvidia-kmodsrc-3:565.77 100% | 13.7 MiB/s | 56.8 MiB | 00m04s
> -----------------------------------------------------------------------------------
> [6/6] Total 100% | 26.9 MiB/s | 131.2 MiB | 00m05s
> [1/7] https://developer.download.nvidia.co ???% | ...
> ...
> ...
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> >>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
> -----------------------------------------------------------------------------------
> [7/7] Total 100% | 26.9 MiB/s | 131.2 MiB | 00m05s
> Failed to download files
> Librepo error: Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/fedora41/x86_64/7fa2af80.pub (IP: 23.36.162.199)
also:
> liviu@bobdenaut î‚° /etc/yum.repos.d î‚° ls -la /etc/yum.repos.d/*cuda*.repo
> -rw-r--r--. 1 root root 202 Feb 7 13:29 /etc/yum.repos.d/cuda-fedora41-12-8-local.repo
> liviu@bobdenaut î‚° /etc/yum.repos.d î‚° cat /etc/yum.repos.d/cuda-fedora41-12-8-local.repo
> [cuda-fedora41-12-8-local]
> name=cuda-fedora41-12-8-local
> baseurl=file:///var/cuda-repo-fedora41-12-8-local
> enabled=1
> gpgcheck=1
> gpgkey=file:///var/cuda-repo-fedora41-12-8-local/FA066D84.pub
> obsoletes=0
> liviu@bobdenaut î‚° /etc/yum.repos.d î‚°
liviu
(diLivio)
February 7, 2025, 11:39am
5
Thanks, here it is:
liviu@bobdenaut î‚° /etc/yum.repos.d î‚° dnf repolist
repo id repo name
1password 1Password Stable Channel
copr:copr.fedorainfracloud.org:lukenukem:asus-linux Copr repo for asus-linux owned by lukenuk
copr:copr.fedorainfracloud.org:phracek:PyCharm Copr repo for PyCharm owned by phracek
cuda-fedora41-12-8-local cuda-fedora41-12-8-local
docker-ce-stable Docker CE Stable - x86_64
fedora Fedora 41 - x86_64
fedora-cisco-openh264 Fedora 41 openh264 (From Cisco) - x86_64
google-chrome google-chrome
libnvidia-container libnvidia-container
nvidia NVIDIA CUDA Repository
nvidia-container-runtime nvidia-container-runtime
nvidia-container-toolkit nvidia-container-toolkit
nvidia-docker nvidia-docker
rpmfusion-free RPM Fusion for Fedora 41 - Free
rpmfusion-free-updates RPM Fusion for Fedora 41 - Free - Updates
rpmfusion-nonfree RPM Fusion for Fedora 41 - Nonfree
rpmfusion-nonfree-nvidia-driver RPM Fusion for Fedora 41 - Nonfree - NVID
rpmfusion-nonfree-steam RPM Fusion for Fedora 41 - Nonfree - Stea
rpmfusion-nonfree-updates RPM Fusion for Fedora 41 - Nonfree - Upda
updates Fedora 41 - x86_64 - Updates
I note that you are installing from the nvidia site (the cuda-fedora41-12-8-local repo)
In my experience at least a portion of your problem is that there are (almost 100%) conflicts between the rpmfusion repos and the cuda-fedora repos.
Please try the installation again but add the option --disablerepo cuda-fedora41-12-8-local
to your command and see what changes there may be.
The rpmfusion repo does not install dkms or any related packages when installing akmod-nvidia.
liviu
(diLivio)
February 7, 2025, 11:43am
7
That is something that I dont want to, somehow I think I’ve messedup my repos.
Help will be much appreciated.
Thanks
liviu
(diLivio)
February 7, 2025, 11:46am
8
liviu@bobdenaut î‚° ~ î‚° sudo dnf install akmod-nvidia --disablerepo cuda-fedora41-12-8-local
Updating and loading repositories:
Repositories loaded.
Package Arch Version Repository Size
Installing:
akmod-nvidia x86_64 3:565.77-1.fc41 rpmfusion-non 90.9 KiB
Installing dependencies:
dkms noarch 3.1.4-3.fc41 updates 187.6 KiB
kmod-nvidia-open-dkms noarch 3:570.86.15-1.fc41 nvidia 100.2 MiB
nvidia-kmod-common noarch 3:570.86.15-1.fc41 nvidia 87.4 MiB
nvidia-modprobe x86_64 3:570.86.15-1.fc41 nvidia 54.8 KiB
xorg-x11-drv-nvidia-kmodsrc x86_64 3:565.77-3.fc41 rpmfusion-non 56.9 MiB
Transaction Summary:
Installing: 6 packages
Total size of inbound packages is 131 MiB. Need to download 57 MiB.
After this operation, 245 MiB extra will be used (install 245 MiB, remove 0 B).
Is this ok [y/N]: y
[1/3] nvidia-kmod-common-3:570.86.15-1.fc4 100% | 0.0 B/s | 0.0 B | 00m00s
>>> Already downloaded
[2/4] nvidia-modprobe-3:570.86.15-1.fc41.x 100% | 0.0 B/s | 0.0 B | 00m00s
>>> Already downloaded 0% | 0.0 B/s | -2.0 B | 00m00s
[3/5] kmod-nvidia-open-dkms-3:570.86.15-1. 100% | 0.0 B/s | 0.0 B | 00m00s
>>> Already downloaded 0% | 0.0 B/s | -2.0 B | 00m00s
[4/6] dkms-0:3.1.4-3.fc41.noarch 100% | 0.0 B/s | 0.0 B | 00m00s
>>> Already downloaded 0% | 0.0 B/s | -2.0 B | 00m00s
[5/6] akmod-nvidia-3:565.77-1.fc41.x86_64 100% | 124.4 KiB/s | 40.3 KiB | 00m00s
[6/6] xorg-x11-drv-nvidia-kmodsrc-3:565.77 100% | 19.1 MiB/s | 56.8 MiB | 00m03s
-----------------------------------------------------------------------------------
[6/6] Total 100% | 14.5 MiB/s | 56.8 MiB | 00m04s
[1/7] https://developer.download.nvidia.co ???% | 0.0 B/s | 0.0 B | 00m00s
[1/7] https://developer.download.nvidia.co ???% | 0.0 B/s | 0.0 B | 00m00s
[1/7] https://developer.download.nvidia.co ???% | 0.0 B/s | 0.0 B | 00m00s
[1/7] https://developer.download.nvidia.co ???% | 0.0 B/s | 0.0 B | 00m01s
[1/7] https://developer.download.nvidia.co ???% | 0.0 B/s | 0.0 B | 00m01s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 10.0 B/s | 10.0 B | 00m00s
[1/7] https://developer.download.nvidia.co 100% | 4.0 B/s | 10.0 B | 00m02s
>>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
>>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
>>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
>>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
>>> Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/f
-----------------------------------------------------------------------------------
[7/7] Total 100% | 14.5 MiB/s | 56.8 MiB | 00m04s
Failed to download files
Librepo error: Status code: 404 for https://developer.download.nvidia.com/compute/cuda/repos/fedora41/x86_64/7fa2af80.pub (IP: 2.23.7.10)
So, how do I revert my repo to “factory settings”, remove everything related to nvidia and start over?
Thanks
augenauf
(Flo)
February 7, 2025, 11:49am
9
Just remove the .repo files from /etc/yum.repos.d
, or open each of the them and set enabled=0
, or use Gnome Software to disable them.
@liviu , btw, what you want when pasting terminal output is “Preformatted Text”, not “quote”. Thanks.
1 Like
I would first consider disabling the cuda-fedora repo permanently along with the nvidia, nvidia-container-runtime, nvidia-containter-toolkit, and libnvidia-container repos.
This can be done with sudo dnf config-manager setopt cuda-fedora41-12-8-local.enable=0
Repeat for each repo you wish to disable, using the repo ID from the “dnf repolist” command and you can do all at once if you wish.
(Disabling the repos can be changed to re-enable them with a similar dnf command)
Only enable those nvidia repos if they have a package you actually need for your work flow, and only after the nvidia drivers have been installed from rpmfusion.
The rpmfusion repo contains the 565.77 driver at present but probably will be updated to the 570.86 driver soon.
liviu
(diLivio)
February 7, 2025, 12:06pm
11
Thanks.
I just removed cuda.
All I need is to get my nvidia running so I can execute docker with nvidia-gpu.
If you run the command dnf install akmod-nvidia
and it shows packages to be installed from any repo other than rpmfusion or fedora you should disable all those repos first before continuing the installation. For the best results that command should only pull from the rpmfusion and fedora repos.
1 Like
liviu
(diLivio)
February 7, 2025, 1:28pm
13
Thanks for the help!
liviu@bobdenaut:~$ sudo dnf repolist
repo id repo name
copr:copr.fedorainfracloud.org:phracek:PyCharm Copr repo for PyCharm owned by phracek
fedora Fedora 41 - x86_64
fedora-cisco-openh264 Fedora 41 openh264 (From Cisco) - x86_64
google-chrome google-chrome
rpmfusion-free RPM Fusion for Fedora 41 - Free
rpmfusion-free-updates RPM Fusion for Fedora 41 - Free - Updates
rpmfusion-nonfree RPM Fusion for Fedora 41 - Nonfree
rpmfusion-nonfree-nvidia-driver RPM Fusion for Fedora 41 - Nonfree - NVIDIA Driver
rpmfusion-nonfree-steam RPM Fusion for Fedora 41 - Nonfree - Steam
rpmfusion-nonfree-updates RPM Fusion for Fedora 41 - Nonfree - Updates
updates Fedora 41 - x86_64 - Updates
liviu@bobdenaut:~$ sudo dnf install akmod-nvidia
Updating and loading repositories:
Repositories loaded.
Package "akmod-nvidia-3:565.77-1.fc41.x86_64" is already installed.
Nothing to do.
liviu
(diLivio)
February 7, 2025, 1:30pm
14
Now, how all this has started?
I need to install docker with nvidia-gpu support and somehow I’ve managed to mess the repo.
Anyidea where do I find some docs to guide me how to install docker with nvidia-gpu support?
Thanks
To verify that the nvidia driver is loaded and functioning please show dnf list --installed \*nvidia\*
and lsmod | grep nvidia
Once that is verified then work on installing docker.
Docker is actually in the fedora repos
$ sudo dnf install docker*
Updating and loading repositories:
Repositories loaded.
Package Arch Version Repository Size
Installing:
docker-buildx x86_64 0.18.0-1.fc41 updates 57.6 MiB
docker-cli x86_64 27.3.1-2.fc41 fedora 26.8 MiB
docker-compose x86_64 2.30.3-1.fc41 updates 59.1 MiB
docker-compose-switch x86_64 1.0.5-1.fc41 fedora 3.3 MiB
docker-distribution x86_64 2.8.1-6.gitb5ca020.fc41 fedora 16.5 MiB
Installing dependencies:
containerd x86_64 1.7.25-1.fc41 updates 139.3 MiB
moby-filesystem noarch 27.3.1-2.fc41 fedora 0.0 B
runc x86_64 2:1.2.4-1.fc41 updates 11.2 MiB
tini-static x86_64 0.19.0-9.fc41 fedora 756.3 KiB
Installing weak dependencies:
moby-engine x86_64 27.3.1-2.fc41 fedora 102.2 MiB
moby-engine-nano noarch 27.3.1-2.fc41 fedora 102.1 KiB
Transaction Summary:
Installing: 11 packages
Total size of inbound packages is 108 MiB. Need to download 108 MiB.
After this operation, 417 MiB extra will be used (install 417 MiB, remove 0 B).
Is this ok [y/N]:
liviu
(diLivio)
February 7, 2025, 2:34pm
16
Jeff V:
lsmod | grep nvidia
liviu@bobdenaut:~$ dnf list --installed \*nvidia\*
Installed packages
akmod-nvidia.x86_64 3:565.77-1.fc41 rpmfusion-nonfree-nvidia-dri
kmod-nvidia-6.12.11-200.fc41.x86_64.x86_64 3:565.77-1.fc41 @commandline
nvidia-gpu-firmware.noarch 20241210-1.fc41 <unknown>
nvidia-modprobe.x86_64 3:565.77-1.fc41 rpmfusion-nonfree-nvidia-dri
nvidia-persistenced.x86_64 3:565.77-1.fc41 rpmfusion-nonfree-updates
nvidia-settings.x86_64 3:565.77-1.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia-cuda.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-cuda-libs.i686 3:565.77-3.fc41 rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-cuda-libs.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia-kmodsrc.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia-libs.i686 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia-libs.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
xorg-x11-drv-nvidia-power.x86_64 3:565.77-3.fc41 rpmfusion-nonfree-nvidia-dri
liviu@bobdenaut:~$
liviu@bobdenaut:~$ lsmod | grep nvidia
nvidia_drm 147456 13
nvidia_modeset 1671168 4 nvidia_drm
nvidia_uvm 3989504 2
nvidia 77516800 58 nvidia_uvm,nvidia_modeset
nvidia_wmi_ec_backlight 12288 0
drm_ttm_helper 16384 2 nvidia_drm,xe
video 81920 6 nvidia_wmi_ec_backlight,asus_wmi,asus_nb_wmi,xe,i915,nvidia_modeset
wmi 28672 4 video,nvidia_wmi_ec_backlight,asus_wmi,wmi_bmof
So my driver is loaded.
As about the docker, yes, I know it is in the repos but when I am trying to load an image it says something like it doesnt see nvidia_gpu (something like that, I dont have the exact msg right now.)
And this happens ONLY if you run docker as normal user (rootless):
docker etc etc
But, if you run it as root using “sudo” then averything works just fine.
So, even after adding a docker grouo and my self to it, I still can not manage to run the docker as dormal user.
Reading here maybe this will help.