Bare-metal helm installation of Nutanix-csi-storage 3.1.0 label the nodeid with #<ip># instead of acceptable node name

[RatRaceExile]

Helm Deployment of Nutanix-csi-storage 3.1.0 on vanilla kubernetes

To reproduce error:

nutanix-csi-storage
/values.yaml

Set
kubernetesClusterDeploymentType: "bare-metal"

My PC is on 2024.2.
Creation of SC, PV, PVC are all successfully.

But on mount,
AttachVolume.Attach failed for volume rpc error: code = NotFound desc = bad CSI nodeId format: #ip of node#

Nutanix support ticket 01844179.
The SRE does not seem to have knowledge about kubelet registration of CSI driver.

I hope that the dev can take a look at what causes the CSI driver to register nodeid with invalid char #.

---

If I set kubernetesClusterDeploymentType: "non-bare-metal", the new csi went to take the machine-id and attempt to use that as the vm uuid when querying the PC API.

[PALJA09]

we are interested with the solution

[RatRaceExile]

@PALJA09 #175 Seems like the issue still exists in v3.38.

As long as you dont set your VM machine id to be the same as the VM uuid within Nutanix PC/AOS, the issue will continue to persist.

Hence, either you force set your machine id, or dont use Nutanix CSI all together