Hi, for shared filesystem master slave setup you should use either KahaDB (from 5.3.0 on) or ActiveMQ persistence adapter (on earlier versions). If you plan to have shared database master slave, that you should use JDBC persistence adapter (without journaling).
Cheers -- Dejan Bosanac - http://twitter.com/dejanb Open Source Integration - http://fusesource.com/ ActiveMQ in Action - http://www.manning.com/snyder/ Blog - http://www.nighttale.net On Mon, Apr 5, 2010 at 9:36 AM, Feng AtWork <feng8w...@gmail.com> wrote: > Hi, There, > > I have been evaluating ActiveMQ for our new project for the past few weeks, > and I cannot find a conclusive answer about exactly which persistence > adapters can be used with a shared file system master/slave set up. > > There seems to be some conflicting information out there, for example: > > 1) according to Bruce Snyder in this thread, > > http://markmail.org/message/ldjoyewy3jczcaai?q=shared+file+system++list:org.apache.activemq.users > , > it seems journaledJBDC can only be used with shared JDBC master/slave, > > but 2) the example in > http://activemq.apache.org/shared-file-system-master-slave.html shows to > use > just the journaledJDBC adapter with shared file system master/slave; > > and 3), according to the same thread in 1), AMQ persistence should be used > with shared file system master/slave. > > And what about kahaPersistenceAdapter, it is file based, but the above > thread mentions http://activemq.apache.org/kahadb-master-slave.html, which > uses zookeeper and is not supported. > > Finally, what is the different between kahaPersistenceAdapter in > <kahaPersistenceAdapter dir="target/kaha-data"/> and just kaha in <kahaDB > directory="activemq-data" journalMaxFileLength="32mb"/>. Are > kahaPersistenceAdater and kahaDB the same thing? If > kahaPersistenceAdapter > cannot be used with shared file system JDBC, can kahaDB be used instead? > > Any clarification will be much appreciated. We load tested journaledJDBC > with default derby data source and an Oracle data source, the performance > were not satisfactory. > > Thanks. > > -Feng >
