0 Replies Latest reply on Aug 12, 2013 1:07 PM by ksimpson

    Slave broker always starts first

    ksimpson

      Scenario: 

      • Two RHEL6 VMs, each with an installation of JBoss Fuse
      • Each JBoss Fuse has two brokers, master and slave
      • master01/slave01 point to same mounted NFS, as are master02/slave02
      • First VM has master01 and slave02
        • master01 network uri = "masterslave(:(MASTER02_URI, SLAVE02_URI))?randomize=false"
        • slave02 network uri = "masterslave(:(MASTER01_URI, SLAVE01_URI))?randomize=false"
      • Second VM has master02 and slave01
        • master02 network uri = "masterslave(:(MASTER01_URI, SLAVE01_URI))?randomize=false"
        • slave01 network uri = "masterslave(:(MASTER02_URI, SLAVE02_URI))?randomize=false"
      • Broker configuration includes 'networkConnectorsStartAsync=true'

       

      My expectation after restarting both VMs is that the master brokers will start on each machine, as the randomize option is set to false.  But slave02 always starts first on vm01, which is additionally confusing because master02 and slave02 point to same data store, which means they shouldn't both have lock.

       

      Can anyone shed light on this stuff?  None of the examples I've found online work the way they claim...