In the RHEV environment, every host is part of a RHEV cluster. In a local high availability scenario, hosts forming a VCS cluster should be part of a single RHEV cluster. However, in disaster recovery scenarios, you can configure all hosts on the primary site in one RHEV cluster and all hosts on the secondary site in a different RHEV cluster, though they are all part of the same datacenter. During a site failover, when the DROpts attribute is set, VCS changes the virtual machine host as per the new RHEV cluster.

If the DROpts attribute is not set, VCS does not allow a host from a different RHEV cluster to start the virtual machine. This issue occurs because virtual machine migration does not work across RHEV clusters. Therefore, VCS fails to start the virtual machine on a host that is part of a different cluster.

Veritas recommends configuring hosts in different clusters only in a disaster recovery configuration, and setting the DROpts attribute of the KVMGuest agent. For a local high availability scenario, you do not need to set the DROpts attribute, and all the hosts forming a VCS cluster should be part of the same RHEV cluster.