Trying to understand servicemix message routing and fault-tolerance

[William Blackburn <wj...@mac.com>]

I've read the docs on clustering and spent some time reading up on  
activemq brokers and other related topics, so I've got a feel for  
scaling servicemix, but what I'm still missing is fault tolerance.  
How can I ensure that a message will definitely reach its destination.

My concrete example is something like this. Two components are  
orchestrated in some way such that one component prepares a (human  
readable) message and the other connects to a mail server and  
actually send the message. The second component may encounter  
recoverable errors such as a transient network failure or a mail  
server down for maintenance or similar, so the initial attempt may  
fail, but another attempt in the future would succeed.

I know I can code up that retry behavior myself in a number of ways,  
but I'm hoping to find that some behavior of the existing servicemix  
queuing and routing implementation can be used instead to cause the  
first component to retry its message to the second after some  
specified wait.

Any help appreciated.

BJ