[Bug]: GPU Error shows up on Non-GPU Clusters

[andrewballantyne]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

On some clusters, we get an error on the Workbench page in relation to GPUs. This was reported on the Dev Sandbox as well as in another RHODS cluster.

It's unclear what the actual cause is -- but it looks to be a 404 from the backend /api/gpu endpoint. Which means it likely is one of the many calls that happens on the backend that causes it to happen. The backend code is not safe from promise fails in k8s calls so it could just be a fail to fetch some details.

Expected Behavior

Silently stop 404s and just say there is no GPUs on this cluster.

Steps To Reproduce

Unknown.

Workaround (if any)

No response

What browsers are you seeing the problem on?

No response

Open Data Hub Version

Dashboard: v2.11.0

Anything else

This will be hard to reproduce -- best guess is it can't find the scale machines.

Investigation of the GPU endpoint will be needed to see if there is some triggering factor. There are several calls that are unbounded and can throw into a non-existent catch which turns it into a thread breaking error.

[Gkrumbach07]

This issue is will no longer relevant as the api/gpu backend endpoint is deprecated in favor for accelerator profiles

tracker for accelerators merging into main:

• [UI] Habana Support Part 1 #1450

[Gkrumbach07]

This can be closed with the completion of:

• [Feature Request]: accelerator feature cleanup #1894
  this issue will remove the deprecated gpu endpoint