Hi,
have you already been able to solve this? Do you have separate networks for Corosync/Cluster, Ceph, Administration etc? Because it sounds like the migrations saturate your network connection, then the Corosync packages do not arrive in time
and then all sort of bad stuff can happen (like...