I went the route of removing the device and target from A (the unhealthy node), removing it from the Replication Manager, then using the Replication Manager on B to recreate the device and target on A. I left it over the weekend to resync but it's still failing.
Sorry for such a delay and hope you enjoyed your holidays! R&D confirmed the slow underlying storage response time to be breaking synchronization again and again. Please log a support call with us using this thread and 611061 as your references. Let us look into this incident together.