Hi. I'm using Hibernate Search 4.1.1.Final with JMS master-slave cluster configuration.

I need a standard JMS master-slave configuration where slave also updates it's own (local) indexes immediately after (or before) sending notification to JMS queue.
Master refreshes slaves every 60 seconds and its not optimal for my system to decrease that value. Other slaves can wait the refresh time, but the slave originating work needs his indexes updated immediately.

So far I implemented my custom BackendQueueProcessor which wraps both JMSBackendQueueProcessor and LuceneBackendQueueProcessor and delegates each work to both of them.
However, when I use filesystem-slave directory provider, which seems proper for me, the slave indexes are correctly refreshed after those 60 seconds but there is no immediate change. When using filesystem direcotry provider, slave indexes are refreshed immediately, but not updated from master.
What configuration should I use or what should I implement to achieve my goal?

I'm using native locking strategy in slaves, if that's important in the case.