Thanks to visit codestin.com
Credit goes to github.com

Skip to content
This repository was archived by the owner on Dec 9, 2025. It is now read-only.

Conversation

@gauravkghildiyal
Copy link
Member

This should help enable use cases involving device topology alignment.

Currently, we fail to process ResourceClaims that have devices from distinct drivers like the example below, since we don't recognize the customgpu0 device.

apiVersion: resource.k8s.io/v1beta1
kind: ResourceClaim
metadata:
  name: my-claim
spec:
  devices:
    constraints:
    - matchAttribute: resource.kubernetes.io/pcieRoot
      requests:
      - nic-request
      - gpu-request
    requests:
    - allocationMode: ExactCount
      count: 1
      deviceClassName: dranet-cloud
      name: nic-request
    - allocationMode: ExactCount
      count: 1
      deviceClassName: gpu.nvidia.com
      name: gpu-request
status:
  allocation:
    devices:
      results:
      - adminAccess: null
        device: eth1
        driver: dra.net
        pool: my-node
        request: nic-request
      - adminAccess: null
        device: customgpu0
        driver: gpu.nvidia.com
        pool: my-node
        request: gpu-request

Copy link
Collaborator

@aojea aojea left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

great catch, I thought there was something before like this, or I misremember or at some point I deleted it, but this is really needed

@aojea aojea merged commit 923989a into google:main Jul 9, 2025
6 checks passed
@michaelasp
Copy link
Collaborator

Ah this is the reason why I had to split my example into 2 separate claims which the pod then asked for. Good fix!

@gauravkghildiyal gauravkghildiyal deleted the ignore-non-dranet-devices branch July 12, 2025 01:41
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants