I'm getting this error from the scheduler once a minute
Error getting pod default/<deployment name>-<old replica set number>-<random pod number> for retry
Then it says that the error it got back then was that there where no node that could fit the cpu requirements.
Since then this replica set is long gone.
Can I make the scheduler to stop?
Are you sure that the pod is not existing? Since the scheduler will not retry scheduling if the binding request returns NotFound error, I think the pod may is still there.
I think you should delete the pod manually to notify the scheduler not to retry that pod.