Enable resource naming in config

[MondayCha]

Motivation

Volcano v1.9.0 introduces Capacity scheduling capabilities, which makes it possible to configure different quotas for different types of GPU queues (important in production environments). For example:

apiVersion: scheduling.volcano.sh/v1beta1
kind: Queue
metadata:
  name: queue1
spec:
  reclaimable: true
  deserved: # set the deserved field.
    cpu: 2
    memeory: 8Gi
    nvidia.com/t4: 40
    nvidia.com/a100: 20

However, the default Nvidia Device Plugin reports resources as nvidia.com/gpu, which does not support reporting different GPU models as shown in the example.

To address this, we need to customize the device plugin.

Change Details

The NVIDIA community has already had discussions about this issue:

• Issue: Advertising specific GPU types as separate extended resource · Issue #424 · NVIDIA/k8s-device-plugin
• Docs: [External]Custom Resource Naming and Supporting Multiple GPU SKUs on a Single Node in Kubernetes
• Code: k8s-device-plugin/cmd/nvidia-device-plugin/main.go at eb8fd565c3df0caca59bf0ff2ae918e647f46af3 · NVIDIA/k8s-device-plugin

This PR is modified based on the above discussion.

Further Impact

GPU resource renaming will prevent the DCGM Exporter from obtaining pod-level GPU resource usage monitoring, since the DCGM Exporter must exactly match the resource name nvidia.com/gpu or those with a prefix of nvidia.com/mig-.

[volcano-sh-bot]

Welcome @MondayCha!

It looks like this is your first PR to volcano-sh/devices.

Thank you, and welcome to Volcano. 😃

[Monokaix]

Hi，please add more description about this pr，and use git commit -s to sign off your commit.

[william-wang]

Thanks for your contribution. I opened a issue #69 for this pr.

[william-wang]

@MondayCha Would you like to add a doc to guide how to configure and use it ?

[hzxuzhonghu]

/ok-to-test

[Monokaix]

If on device plugin is installed, would helm upgrade be successful?

[MondayCha]

This section is referenced from the installation instructions in the official documentation.

If the user has previously installed an existing gpu device plugin in other ways, then helm upgrade may fail. In this case, it is recommended that the user uninstall and then reinstall it.

[Monokaix]

/lgtm

[william-wang]

/approve

[volcano-sh-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: william-wang

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

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [william-wang]

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

[Monokaix]

/lgtm