During the failover/failback there is a short outage in iSCSI connection. There are lots of error events in Windows event log (e.g. "Initiator failed to connect to the target." ...) on cluster nodes and the cluster disks are going offline and online again. Only a few seconds, but the VMs running on cluster (with CSV) are crashing a few seconds after that occures.

I've opened a trouble ticket at open-e support, but no answer yet ...

So my workaround is to save the VMs, doing the DSS failover and start the VMs again. It's bad for a HA scenario, but the only way at the moment ...