Hi,

There is a lucene index stored in a server. In the same server there are two application servers Jboss 7.1 and Wildfly 8 beta1 running on different ports. In JBoss 7.1 there are two war using different data sources. In Wildfly there is a one war. All 3 wars has to use(read-write) same lucene index through hibernate search. All three wars have the same hibernate search version.

What is the proper way to configure hibernate search for all wars having in mind that the will read y/o write concurrently to the index?

Hi,

have you checked out the discussion of the different backends in the reference guide [1]?

You could get started with the simple Lucene backend [2] which directly uses a shared index (also see [3] for locking options). If the shared index writes show up as bottleneck you could switch to the JMS backend [4] which performs all writes through a single node, serves reads from index copies and propagates index updates to the master node via JMS.

--Gunnar

[1] http://docs.jboss.org/hibernate/search/4.4/reference/en-US/html_single/#d0e633
[2] http://docs.jboss.org/hibernate/search/4.4/reference/en-US/html_single/#d0e644
[3] http://docs.jboss.org/hibernate/search/4.4/reference/en-US/html_single/#search-configuration-directory-lockfactories
[4] http://docs.jboss.org/hibernate/search/4.4/reference/en-US/html_single/#d0e666