https://issues.redhat.com/browse/ENTMQBR-4530 has introduced a pluggable mechanism to coordinate broker activation during replication based on 2 roles:

• primary

• backup

This mechanism works well in asymmetric scenario where a primary machine has better hardware (disk, network, CPU or memory) then its configured backup and while restarted after a failover of its backup, expect to take the lead and serve clients as live again. This process is called "failback".

In scenarios were brokers hardware is the same and/or no broker is a preferred choice to act as a live (meaning that failing-back isn't needed), brokers can be said to be "peers" and they can just agree on a ID to coordinate their activation(s) (and ensure just a single live broker exists, replicated): this ID is called coordination-id and must be statically specified in the brokers configurations.

Thanks to the symmetric behaviour of brokers, no brokers need to be configured as backup, but both can just be configured as primary, sharing the same coordination-id.