How is ActiveMQ fault-tolerant?

Question

Apache HTTP Server

How to properly ensure the fault tolerance of brokers?
Did I understand correctly based on the docks on the AMQ website that the best option would be a network of brokers, each of which implements a master slave, for example, using ZooKeeper?
Queue messages are distributed over the network of brokers, and in the failover at consumers and producers, the addresses of the masters are indicated. If one of the nodes of the broker network in the person of the master falls, then it loses consumers and producers, and then other brokers take its messages from its slaves and work continues through failover. Then the fallen master rises as an admin, receives some of the new producers and consumers, and some of the messages are sent to him.
Am I understanding everything correctly, or am I misunderstanding something?