[rabbitmq-discuss] use of an NFS for broker failover

[Shane]

Given two brokers (Broker A and Broker B) use the same file system
space via NFS and the following events occur:

1. Broker A goes down with producers/consumers connected to it.
2. Broker B starts up
3. Producers/consumers reconnect to Broker B.

All exchanges/queues are created upon startup of the producers/
consumers.

What else should I be considering to achieve failover?