In some cases, yes, but performance may be adversely affected by high connection latency. In the case of a failover partnership sharing a large number of leases between two widely separated partners, it may take unreasonably long for the partners to settle during recovery.
For this reason we do not recommend the practice, but if your design requires it, you can improve the situation somewhat through the use of the max-unacked-updates setting. Increasing this parameter allows the failover partners to send larger batches of updates before waiting for a required acknowledgment when resynchronizing, which can help speed recovery.