Well, I just gave it a try myself and it didn't quite do the trick. With the failover, the slave does wait for the master, and when the master does start, the two appear to connect okay. However, if you bring down the master, the slave fails to open up it transports.
Joe Joe Fernandez wrote: > > Try using a failover connector like this. > > <broker ... masterConnectorURI="failover://(tcp://master:61616)"> > > Joe > http://www.ttmsolutions.com > > > > > Andreas.Weber wrote: >> >> Hi, >> >> in a cluster szenario we run a distributed application containing >> ActiveMQ. We configure a pure master/slave, with one cluster host as >> ActiveMQ master (waitForSlave="true") and another one as slave. All >> cluster nodes are started simultaneous (more or less, not really >> predictable). >> >> Now, when slave node starts before master node, slave node (activemq) >> gives an error that it can't connect to master. ("MasterConnector: >> Failed to start network bridge...") >> Even if master starts just a few seconds after the slave, this error >> occurs. >> >> This also causes the master to hang up, because it waits for the slave. >> (We have to set waitForSlave="true" to assure consistency.) >> >> So, is there any way to avoid that the slave node has to be started >> first, and the master node has to wait until the slave node comes up? >> >> My expectation would have been, that the slave is polling for the >> master, with retrying that again, if connection fails in the first >> attempt (?) >> >> Any ideas? >> >> Best Regards, >> Andreas >> >> >> >> >> >> >> > > -- View this message in context: http://old.nabble.com/Pure-Master---Slave---start-order-tp28880946p28881460.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
