Looking through docker containers to Nvidia GPU hardware

As background, I used the following helpful resources :

Key points being :

Even though Nvidia suggests using new docker, you still need their container pass-through to get to the host's drivers
Fedora's version of docker is too old : You need to use the official docker repos (for CentOS...)

Install the docker repo, and test

dnf -y install dnf-plugins-core

dnf remove docker   # Just to make sure that we're not on the Fedora official one

dnf config-manager --add-repo https://download.docker.com/linux/fedora/docker-ce.repo
dnf install docker-ce docker-ce-cli containerd.io

systemctl start docker
docker run hello-world
#
# Hello from Docker!
# This message shows that your installation appears to be working correctly.
#
# ...

Set up user permissions for docker

We need to add each user that requires access to the docker system to the docker group :

grep dock /etc/group
#docker:x:991:
#dockerroot:x:990:

[root]# usermod -aG docker user
[root]# su - user
[user]$ id -a # Make sure user is in docker group

If you're already logged in as an individual user, you can 'revitalise' your group membership without re-logging in, by :

sudo -k # reset sudo timeout
exec sudo -i -u $(whoami) # no password necessary

Install the nvidia repo, and test

Now on to the crucial Nvidia piece :

dnf config-manager --add-repo https://nvidia.github.io/nvidia-docker/centos7/nvidia-docker.repo
dnf install nvidia-container-toolkit

systemctl restart docker

docker run -it --rm --gpus all fedora nvidia-smi -L
# GPU 0: GeForce GTX 1060 6GB (UUID: GPU-b69d6083-PIIa-PIIa-PIIf-a33c1fa106c2)

All done.