Within a cluster, it is desirable to have ability to specify preferred failover nodes. For example if MRG cluster consists of servers A, B,C and server A acts as an active broker, to be able to specify that when/if server A fails, then server B will become next active broker. Server C will become active broker only when/if both A and B fail.
The cluster is active-active so this feature is no longer relevant. All cluster nodes are active, and clients can connect to any active node of the cluster. On failure of a node, all other nodes remain active and can be used as failover targets, there's no need to specify an order.