Add support to docker runtime for OCI access to NVIDIA GPUs

[aaron-nall]

fixes #20597

Adds support for OCI interface to NVIDIA GPUS with a new valid --gpus option "nvidia.com".

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: aaron-nall / name: Aaron Nall (8f037a9, f4e9719, b84c2eb, 7de2b58)

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: aaron-nall
Once this PR has been reviewed and has the lgtm label, please assign medyagh for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[k8s-ci-robot]

Welcome @aaron-nall!

It looks like this is your first PR to kubernetes/minikube 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes/minikube has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

[k8s-ci-robot]

Hi @aaron-nall. Thanks for your PR.

I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[minikube-bot]

Can one of the admins verify this patch?

[medyagh]

@aaron-nall can you plz help me undrestand this PR more ? maybe a Before/After this PR ? what exactly is being enabled in this PR ? what couldn't have been done before it ?

and also I think if you do --gpus=all it will automatically enable the addon and not needed to specifiy it

[medyagh]

/ok-to-test

[ComradeProgrammer]

containers/podman#21156 I think it is related to this pr

[aronahl]

@aaron-nall can you plz help me undrestand this PR more ? maybe a Before/After this PR ? what exactly is being enabled in this PR ? what couldn't have been done before it ?

and also I think if you do --gpus=all it will automatically enable the addon and not needed to specifiy it

If CDI Devices are enabled for dockerd, at least in the latest stable NixOS release, the --gpus=all flag has been replaced with --device=nvidia.com/gpu=all.

The dockerd feature is listed as experimental, but it seems to be the only way to access Nvidia GPUs in current NixOS docker installs.

[medyagh]

@aaron-nall can you plz fix the unit test

"make test"

--- FAIL: TestValidateGPUs (0.00s)
start_test.go:830: validateGPUs(cat, docker, docker) = "The gpus flag must be passed a value of "nvidia", "nvidia.com", "amd" or "all""; want = "The gpus flag must be passed a value of "nvidia", "amd" or "all""
FAIL

[minikube-pr-bot]

kvm2 driver with docker runtime

+----------------+----------+---------------------+
|    COMMAND     | MINIKUBE | MINIKUBE (PR 20959) |
+----------------+----------+---------------------+
| minikube start | 50.7s    | 50.8s               |
| enable ingress | 15.4s    | 14.9s               |
+----------------+----------+---------------------+

Times for minikube start: 50.5s 51.3s 53.8s 49.9s 48.2s
Times for minikube (PR 20959) start: 51.8s 49.1s 53.3s 49.7s 50.3s

Times for minikube ingress: 16.1s 15.5s 16.0s 15.0s 14.5s
Times for minikube (PR 20959) ingress: 14.5s 14.5s 15.0s 15.5s 15.0s

docker driver with docker runtime

+----------------+----------+---------------------+
|    COMMAND     | MINIKUBE | MINIKUBE (PR 20959) |
+----------------+----------+---------------------+
| minikube start | 23.9s    | 23.1s               |
| enable ingress | 12.6s    | 12.8s               |
+----------------+----------+---------------------+

Times for minikube ingress: 12.8s 12.3s 13.3s 12.3s 12.3s
Times for minikube (PR 20959) ingress: 13.8s 12.3s 13.3s 13.3s 11.3s

Times for minikube start: 22.5s 25.6s 24.1s 24.9s 22.6s
Times for minikube (PR 20959) start: 23.0s 23.1s 22.4s 23.0s 23.8s

docker driver with containerd runtime

+----------------+----------+---------------------+
|    COMMAND     | MINIKUBE | MINIKUBE (PR 20959) |
+----------------+----------+---------------------+
| minikube start | 22.6s    | 23.0s               |
| enable ingress | 29.7s    | 23.2s               |
+----------------+----------+---------------------+

Times for minikube (PR 20959) start: 22.4s 22.0s 23.4s 24.3s 23.0s
Times for minikube start: 24.2s 21.6s 23.1s 21.9s 22.4s

Times for minikube ingress: 24.3s 38.8s 22.8s 22.8s 39.8s
Times for minikube (PR 20959) ingress: 23.3s 22.8s 23.3s 23.8s 22.8s

[medyagh]

/retest-this-please