How to use Dockge with Nvidia and Podman #463
Unanswered
wernight
asked this question in
Ask for Help
Replies: 1 comment
-
I have the same issue. Works on the command line with Docker:
|
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
🛡️ Security Policy
📝 Describe your problem
First to use Podman with Dockge on Arch Linux I had to enable the podman service, which I generally don't but that's okay. Also had to change the docker mounted socket path.
sudo systemctl start podman.service sed -i 's#/var/run/docker.sock:#/run/podman/podman.sock:#' compose.yaml sudo podman-compose up -d
After that most composer files work except those using Cuda, like:
The example above does work if I run it directly using
podman-compose up
. Directly runningpodman run --rm --device nvidia.com/gpu=all nvidia/cuda:11.6.2-base-ubuntu20.04 nvidia-smi
also works. However it gives an error when using Dockge:I tried adding the device to Dockge in case the permissions for devices would be transitive (shouldn't but just in case and to no avail):
Internally, Nvidia-Docker integration is configured in
/etc/cdi/nvidia.yaml
which is generated usingsudo nvidia-ctk cdi generate --output=/etc/cdi/nvidia.yaml
. So it may be related to containers/podman#19338.📝 Error Message(s) or Log
Error response from daemon: container create: stat nvidia.com/gpu=all: no such file or directory
🐻 Dockge Version
1.4.2
💻 Operating System and Arch
Arch
🌐 Browser
N/A
🐋 Docker Version
4.9
🟩 NodeJS Version
No response
Beta Was this translation helpful? Give feedback.
All reactions