Documentation Improvements for NVIDIA GPU Operator on Red Hat OCP

[jensmueller-com]

This issue follows up on this issue. These changes improve automation, reduce manual edits, and make the documentation less error-prone.

The documentation for the NVIDIA GPU Operator on Red Hat OpenShift Container Platform includes a chapter titled Installing the NVIDIA GPU Operator on OpenShift. I have two suggestions for improvement.

1. Suggestion for section: Installing the NVIDIA GPU Operator using the CLI

Current behavior: Steps 3–5 require manual edits for channel and starting CSV.

Suggested improvements:

• Step 3: Set CHANNEL dynamically (similar to step 9):

  $ CHANNEL=$(oc get packagemanifest gpu-operator-certified -n openshift-marketplace -o jsonpath='{.status.defaultChannel}')

  Omit the example output and the redundant command in step 4.

• Step 4: Store the starting CSV in a variable:

  $ STARTING_CSV=$(oc get packagemanifests/gpu-operator-certified -n openshift-marketplace -ojson | jq -r '.status.channels[] | select(.name == "'$CHANNEL'") | .currentCSV')

• Step 5: Use these variables when creating the subscription file:

  $ cat <<EOF > nvidia-gpu-sub.yaml
  apiVersion: operators.coreos.com/v1alpha1
  kind: Subscription
  metadata:
    name: gpu-operator-certified
    namespace: nvidia-gpu-operator
  spec:
    channel: $CHANNEL
    installPlanApproval: Manual
    name: gpu-operator-certified
    source: certified-operators
    sourceNamespace: openshift-marketplace
    startingCSV: $STARTING_CSV
  EOF

  Omit the note about manually editing channel and startingCSV.

2. Suggestion for section: Create the cluster policy using the CLI (below "Create the ClusterPolicy instance")

Current behavior: No note about replacing the hardcoded starting CSV (gpu-operator-certified.v22.9.0).

Suggested improvement: If STARTING_CSV is stored as above, update the command:

$ oc get csv -n nvidia-gpu-operator $STARTING_CSV -ojsonpath={.metadata.annotations.alm-examples} | jq '.[0]' > clusterpolicy.json

Note that to support zsh, .[0] must also be put in quotes.

Apply the same change in the other Create the cluster policy using the CLI section (below "Create the ClusterPolicy instance with NVIDIA vGPU").

[a-mccarthy]

@jensmueller-com thnaks for filing! I've just created #320 with the updates