gpu-operator-nfd-worker fails to read net interface attribute speed

[blackliner]

The template below is mostly useful for bug reports and support questions. Feel free to remove anything which doesn't apply to you and add more information where it makes sense.

1. Quick Debug Information

• OS/Version(e.g. RHEL8.6, Ubuntu22.04): Ubuntu 22.04
• Kernel Version: 6.5.0-14
• Container Runtime Type/Version(e.g. Containerd, CRI-O, Docker): containerd v1.7.5
• K8s Flavor/Version(e.g. K8s, OCP, Rancher, GKE, EKS): bare metal k8s via kubespray/kubeadm v1.27.7
• GPU Operator Version: nvcr.io/nvidia/gpu-operator:v23.9.1

2. Issue or feature description

On Supermicro motherboards from the X12/H12 series with RoT (Root of Trust) function, an additional, virtual network interface appears in the operating system. Under Linux, its device name is enx+MAC (e.g. enxb03af2b6059f).

It looks like these USB ethernet gadget devices do not support reading out the speed property:

E0119 17:06:12.398921       1 network.go:143] "failed to read net iface attribute" err="read /host-sys/class/net/enxbe3af2b6059f/speed: invalid argument" attributeName="speed"

On the host

cat /sys/class/net/enxbe3af2b6059f/speed
cat: /sys/class/net/enxbe3af2b6059f/speed: Invalid argument

Network details

ip a show enxbe3af2b6059f
4: enxbe3af2b6059f: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
    link/ether be:3a:f2:b6:05:9f brd ff:ff:ff:ff:ff:ff

Ethtool also doesnt show any relevant info

sudo ethtool -i enxbe3af2b6059f
driver: rndis_host
version: 6.5.0-14-generic
firmware-version: RNDIS device
expansion-rom-version:
bus-info: usb-0000:03:00.4-1.2
supports-statistics: no
supports-test: no
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: no

Relevant links:

• https://github.com/NVIDIA/gpu-operator/issues/429
• https://www.thomas-krenn.com/en/wiki/Virtual_network_interface_enx_of_Supermicro_Motherboards

3. Steps to reproduce the issue

Get one of the newest products from Supermicro with this "feature", install gpu operator and see how nfd fails to register any labels on that node.

4. Information to attach (optional if deemed irrelevant)

• [ ] kubernetes pods status: kubectl get pods -n OPERATOR_NAMESPACE
• [ ] kubernetes daemonset status: kubectl get ds -n OPERATOR_NAMESPACE
• [ ] If a pod/ds is in an error state or pending state kubectl describe pod -n OPERATOR_NAMESPACE POD_NAME
• [ ] If a pod/ds is in an error state or pending state kubectl logs -n OPERATOR_NAMESPACE POD_NAME --all-containers
• [ ] Output from running nvidia-smi from the driver container: kubectl exec DRIVER_POD_NAME -n OPERATOR_NAMESPACE -c nvidia-driver-ctr -- nvidia-smi
• [ ] containerd logs journalctl -u containerd > containerd.log

Collecting full debug bundle (optional):

curl -o must-gather.sh -L https://raw.githubusercontent.com/NVIDIA/gpu-operator/master/hack/must-gather.sh 
chmod +x must-gather.sh
./must-gather.sh

NOTE: please refer to the must-gather script for debug data collected.

This bundle can be submitted to us via email: [email protected]

[cdesiniotis]

@blackliner thanks for reporting this. This was also reported to NFD upstream: https://github.com/kubernetes-sigs/node-feature-discovery/issues/1556 and a fix has been merged: https://github.com/kubernetes-sigs/node-feature-discovery/pull/1557. The fix is not in a released version of NFD yet, but we will pick it up as soon as it is out.

cc @ArangoGutierrez