Install the Node Feature Discovery Operator (NFD)

The Node Feature Discovery (NFD) Operator identifies hardware features and configurations within an OpenShift cluster. It assigns labels to nodes based on their hardware attributes, ensuring that GPU-enabled nodes are marked appropriately. This allows pods that require GPUs to be scheduled on the correct nodes.

About this task

For more information, see the Node Feature Discovery Operator section in the Red Hat OpenShift Container Platform documentation.

Procedure

  1. In the Openshift Web Console, click Operators > OperatorHub in the left hand menu.
  2. Search for Node Feature Discovery.
  3. Click the Node Feature Discovery tile provided by Red Hat (not Community edition).
  4. Click Install.
  5. On the Install Operator page, make sure that the option "A specific namespace on the cluster" is selected.

    You do not need to create a namespace because one has already been created for you.

  6. Click Install.
  7. Verify that the NFD Operator is successfully installed:
    • Navigate to Operators > Installed Operators.
    • Ensure that either All Projects or openshift-nfd is selected in the Projects drop-down at the top of the page.
    • Verify that Node Feature Discovery appears in the list and its status is Succeeded.
  8. If the NFD Operator is installed successfully, click NFD Operator.
  9. Click the NodeFeatureDiscovery tab.
  10. Select Create NodeFeatureDiscovery.
  11. Scroll to the bottom of the page and click Create.
    Note: No configuration changes are required.
  12. Verify that the worker node with the GPU is successfully labeled.
    • In the left menu, navigate to Compute > Nodes.
    • Select the worker node that you know contains a GPU.
    • Click the Details tab.
    • Under Node labels, based on your operator, verify that one of the following labels are present:
      • NVIDIA: feature.node.kubernetes.io/pci-10de.present=true
      • Spyre: feature.node.kubernetes.io/pci-1014.present=true