Just so you know, we have a customer doing this sort of "cross-replication" thing that you describe (but manually, not auto-failover), but we're going to change it to auto-failover, so we're going to have to go to a regular setup with one server only having destination volumes and the other only having source volumes.