When there are more job deletion requests than can be kept in the queue, just stop trying to queue them.