I have 2 nodes cluster A and B and 1 producer connected to node A. Also I have 1 consumer conncted to node A und 10 consumers
connected to node B.
When I send messages to A I observe following behaviour:
- it doesn't matter how many messages are already in the queues
- it doesn't matter how many consumers (1 or 10) are connected to the queue
The load balancing behaviour is always the same. It doesn't take into account the different loads of the queues / nodes
Actually the only one factor is "does the destination node have a consumer or no"
How does the load balancing mechanism select the node exactly? Is there any formula for it?
On producing It's even distribution. We used to only do message redistribution but we always get users asking for even distribution.
there's a JIRA to add it as a feature, but I think that's something we will be able to do only after 2.3 (whatever that will be caleld.. 3.0, 2.4.. dunno)