[rabbitmq-discuss] Shovel configuration for a clustered broker with mirrored high-availability queues

[Eric]

I have a question about configuring shovel for my application.  We 
currently have two broker clusters that are separated by a network 
partition; one cluster of two brokers in AWS, and another cluster of two 
brokers in a separate network environment.  The non-AWS cluster is the 
"master", so we want to shovel items from the AWS cluster to the master.

The shovel configuration allows for multiple source brokers, but if I want 
redundant shovels running, how should I set them up?  Should I create a 
shovel in each of the clustered AWS brokers that is configured to connect 
only to the broker its installed on, create a non-mirrored queue for that 
shovel to consume, and bind it only to the local exchange?  Or should I be 
trying to configure each shovel to declare a mirrored queue and connect to 
each broker?  It seems like the latter would cause every message to be 
published twice.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20120628/d69495df/attachment.htm>