It makes sense to fix it up such that it can work in your case and I
guess to future proof it a bit.
The solution may be to make it configurable or plugable, adding a
fallback transformer implementation that can be delegated to in this
case.
I imagine a new public static
ActiveMQDestination.setFal
This does not sound right. The store and forward should happen async and
should block pending a reconnect of the networkconnector in the event of a
failure of the second broker.
Can you post your activemq configuration?
An active network connector is visible in JMX so you could validate its
state
