node-feature-discovery chart parity with gpu-operator

[wcrum]

What happened:

Referencing and using this network operator alongside the nvidia/gpu-operator and deploying node-feature-discovery should there be parity between the versions of node-feature-discovery?

Latest GPU Operator uses v16.6 and latest network-operator uses v15.6.

What you expected to happen:
node-feature-discovery to be the same across both GPU / Network operator.

How to reproduce it (as minimally and precisely as possible):

network-operator/deployment/network-operator/charts/node-feature-discovery/Chart.yaml

Line 2 in 8c2e3bf

appVersion: v0.15.6

https://github.com/NVIDIA/gpu-operator/blob/9afd674d510f2b4e431d9455130d3a8d030e614f/deployments/gpu-operator/charts/node-feature-discovery/Chart.yaml#L2

[wcrum]

Happy to submit PR to match / update node-feature-discovery if this should be on version v16.6 (latest NDF).

[noama-nv]

One instance of NFD should be running at a time in a kubernetes cluster. NFD should be installed as standalone / with Network Operator / with GPU Operator

you should enable NodeFeatureRule by network operator
see
https://mellanox.github.io/network-operator-docs/customizations/helm.html?highlight=deployNodeFeatureRules

[wcrum]

I understand, GPU Operator and the network operator have different versions of NFD. Is there a bump required for network operator which enables and deploys NFD?

[rollandf]

Will be included in 25.1.0