Create quadlet configuration for Ollama

[p5]

I wanted to provide the Ollama quadlet definitions because it shows how we can attach GPUs to the quadlets, plus changing the selinux configuration and using the .d directory for switching between Nvidia and ROCm.

[rhatdan]

@p5 these are great, I think the ollama.container.d/ only effects the parts outside of the quadlet, IE Things like [Service] and other fields.

The quadlet is changing the IMAGE=foobar line into a podman command to execute foobar. Systemd will read the ollama.container.d/ when the service is starting.

[rhatdan]

@p5 whats up with this one?

[p5]

Will close this as I don't know when I'll get round to finishing it (and I completely forgot I opened it in the first place :facepalm:). Apologies

[rhatdan]

You could always take a look at RamaLama... :^)

[p5]

Starting some investigation into getting CUDA running on ramalama, and will certainly look into contributing a quadlet if it all works :)