Actually it is not an error, because it is normal situation when monitoring tool may see "old" topology when sending task to execute on cluster.