[BUG] - User scheduler pods run on user node groups on GCP, Azure, DO deployments

[Adam-D-Lewis]

Describe the bug

These pods prevent scale down of user node group and should be run on general node group ideally.

Expected behavior

Run on general node group

OS and architecture in which you are running Nebari

Linux x86-64

How to Reproduce the problem?

Do a GCP, Azure, or DO deployment and see that 1 user node never scales down

Command output

No response

Versions and dependencies used.

No response

Compute environment

None

Integrations

No response

Anything else?

No response

[viniciusdc]

I agree, as this is bothersome. However, I was under the impression we needed the user scheduler to exist in the same node as where we plan the pods to be scaled in.

[Adam-D-Lewis]

I agree, as this is bothersome. However, I was under the impression we needed the user scheduler to exist in the same node as where we plan the pods to be scaled in.

That is not necessary on Azure, AWS, or GCP at least, and I can't imagine why it would be different for DO. I've switched it to general node group on an Azure deployment and also on a GCP deployment and things work as expected. Additionally, we already have the user scheduler running on the general node group for AWS.