Currently, each connection manager needs to have configured see-other-host separately, which means that each connection manager may have different conditions for redirection. While this is acceptable, it may cause issues with maintaining configuration as each connection manager may need to be configured separately.

It would be a lot better to have a "single" see-other-host configuration (bean) which would be used by all connection managers.

Additionally, it this "master" clustering strategy, could be used to select processing node for packets in clustering strategy even without knowledge about the fact where client is actually connected and could be used as a single "node" for generation of packets to be delivered to the bare JID.

This task was created as a result of the issue #acspubsub-21 which was fixed but with this feature added, that task would be a lot simpler to solve and would be less error prone.