Net status

[aojea]

Added as beta in 1.33

kubectl get resourceclaims dummy-interface-static-ip  -o=jsonpath='{.status.devices[0].networkData}'
{"hardwareAddress":"8a:09:cd:f9:62:be","interfaceName":"dummy0","ips":["169.254.23.23/32","fe80::8809:cdff:fef9:62be/64"]}

[guptaNswati]

@aojea can you show a diff of how the network interface resourceclaim looked like before adding the status. I am looking into porting this to our nvidia-dra-drivers and still exploring where it fits..

This is what a gpu resourceclaim looks like

status:
    allocation:
      devices:
        results:
        - adminAccess: null
          device: gpu-1
          driver: gpu.nvidia.com
          pool: sc-xx-xxx
          request: gpu

From the PR https://github.com/google/dranet/pull/78/files#diff-e8a7e777d80a14b455bdbf7aae3f28ad8082ffa0a06579e11cc1af741b5f98f7R271 i see your are writing the status whether Ready=True/not to status.allocation.devices.results based on NetworkDeviceReady which is kind of your way of health check. We need to add health checks (xid or ecc errors) to the GPUs before and after allocation to make sure only healthy GPUs are in the AllocatableDevices List.

It seems to be related to this KEP kubernetes/enhancements#4817 because of DRAResourceClaimDeviceStatus FG.

[gauravkghildiyal]

@guptaNswati I'm not sure if this answers your question, but here's what this roughly like today:

status:
  allocation:
    devices:
      results:
      - device: dummy0
        driver: dra.net
        pool: my-node1
        request: req-dummy
    nodeSelector:
      nodeSelectorTerms:
      - matchFields:
        - key: metadata.name
          operator: In
          values:
          - my-node1
  devices:
  - conditions:
    - lastTransitionTime: "2025-08-22T19:58:22Z"
      message: ""
      reason: NetworkDeviceReady
      status: "True"
      type: Ready
    - lastTransitionTime: "2025-08-22T19:58:22Z"
      message: ""
      reason: NetworkReady
      status: "True"
      type: NetworkReady
    device: dummy0
    driver: dra.net
    networkData:
      hardwareAddress: aa:bb:cc:dd:ee:ff
      interfaceName: dummy-renamed
      ips:
      - 10.0.0.1
    pool: my-node1
  reservedFor:
  - name: my-pod1
    resource: pods
    uid: 4f9062a1-8759-420a-bb74-699c9213256c

[guptaNswati]

@gauravkghildiyal its helpful. thank you.

[guptaNswati]

this doc explains more. i wanted to know how Network readiness is decided https://dranet.dev/docs/concepts/interface-status/