[REQ] Respect priorityClass when scheduling eraser pods

[eranreshef]

What kind of request is this?

Improvement of existing experience

What is your request or suggestion?

Scenario:

The eraser-controller-manager tries to schedule pods on nodes which do not have sufficient resources (OutOfPods for example).

Current behaviour:

Since kube-scheduler is not involved in the scheduling process, the new pods remain "stuck" in that error state and no eviction/preemption are executed in order to get the pod to run and do its work, even if eraser pods have higher priority then other running pods on that node.

Desired state:

Lower priority pods are preempted/evicted in order to enable eraser pods to run.

Are you willing to submit PRs to contribute to this feature request?

• Yes, I am willing to implement it.