We call server status to verify the way catchup ended before before making sure the former primary has transitioned back to a secondary. This can cause a race where the node hasn't actually exited catchup when we try to check the metrics.