Add runtimeClassName

[thomaspetit]

This option might be interesting for those that are running a GPU within their kubernetes cluster and want to be able to specify the runtimeClassName manually.

---
apiVersion: v1
kind: Pod
metadata:
  name: nbody-gpu-benchmark
  namespace: default
spec:
  restartPolicy: OnFailure
  runtimeClassName: nvidia
  containers:
  - name: cuda-container
    image: nvcr.io/nvidia/k8s/cuda-sample:nbody
    args: ["nbody", "-gpu", "-benchmark"]
    resources:
      limits:
        nvidia.com/gpu: 1
    env:
    - name: NVIDIA_VISIBLE_DEVICES
      value: all
    - name: NVIDIA_DRIVER_CAPABILITIES
      value: all

[cilindrox]

LGTM 👍 - thanks @thomaspetit

Could you bump the chart's version to 0.4.0?

[thomaspetit]

Fixed and also signed my commits (as per requirement for allowing this PR)

[thomaspetit]

@cilindrox can you have a look at the latest commit? I'd love to re-align my homelab to the latest version with gpu acceleration 😄

[cilindrox]

LGTM 👍

[thomaspetit]

I'm validating some local stuff and just noticed that the securityContext is not configureable which basically means that the GPU can be detected but not used by Plex. Probably need to make it configureable too.

[cilindrox]

I'm validating some local stuff and just noticed that the securityContext is not configureable which basically means that the GPU can be detected but not used by Plex. Probably need to make it configureable too.

Good call - we went with something a bit too lenient but maybe that whole block can be made configurable and the default value just uses the current settings?

LMK and I can keep this PR open

[thomaspetit]

The security context that you're referring to ise indeed required for rclone but I you don't plan to use the rclone or gpu acceleration you maybe do not want to make your plex container privileged from a security pov.

I'll check how I can nicely make it fit and ensure we have something non-breaking 👍

[cilindrox]

whoops, bad container - even more straightforward then ;)

[thomaspetit]

I made an error locally by not adding my encoding package libnvidia-encode-550. It works nicely now without the privileged mode.

We can merge it if ok for you 🥳

[thomaspetit]

🥳