initResource() failed; findAllTfJobs returned error:

[jlewi]

Today I hit the following error running on a GKE cluster with 1.8

E1103 01:44:19.709603       1 controller.go:229] initResource() failed; findAllTfJobs returned error: unknown
E1103 01:44:19.710058       1 controller.go:92] initialization failed: unknown
I1103 01:44:19.710063       1 controller.go:93] retry in 30s...

[jlewi]

I suspect this is the line that's failing

jobList, err := c.TfJobClient.List(c.Namespace)

I suspect its failing because of insufficient RBAC permissions.

I had originally deployed the operator without setting rbac.install=true. I then purged it and reinstalled it but my guess is something didn't get cleaned up.

[jlewi]

I suspect the problem is here

rules:
- apiGroups:
  - mlkube.io

That should probably have been updated to tensorflow.org