can you raise a jira issue[1] for #3 and possibly #2 as the initiator of the discovery should wait for it to start imho.
[1] http://activemq.apache.org/contributing.html#Contributing-Ifyoufindabugorproblem 2009/1/15 BigPic <nab...@jgro.net>: > > > BigPic wrote: >> >> I'm using ActiveMQ 5.2.0 but I can't get clients to discover the broker. >> My goal is to run a single broker with multiple clients all >> auto-discovering it. >> >> I have the broker config: >> >> <transportConnectors> >> <transportConnector name="openwire" >> uri="tcp://localhost:61616" >> discoveryUri="multicast://default?group=test"/> >> </transportConnectors> >> >> and the broker is running on a Win2K system. >> >> Then on another system (Win XP SP3), I have a Java 1.5 client: >> >> String brokerUrl = >> "discovery:(multicast://default)?group=test&maxReconnectAttempts=2"; >> ActiveMQConnectionFactory connectionFactory = >> new ActiveMQConnectionFactory(brokerUrl); >> >> > > OK, found a few problems. #1 is that the client broker URL should be > > String brokerUrl = > "discovery:(multicast://default?group=test)?maxReconnectAttempts=2"; > > #2 is that it appears the discovery agent isn't started until the first > connection attempt and the connection attempt does not in any way wait for > the discovery agent to be started before trying to connect, so the 2 > reconnection attempts pass before the discovery agent discovers the broker. > > #3 is definitely a bug. If I use > > String brokerUrl = > "discovery:(multicast://default?group=test)?maxReconnectAttempts=30&initialReconnectDelay=1000&useExponentialBackOff=false"; > > the connection still fails, and in far less than the 30 seconds that should > be necessary to do 30 attempts with a 1000 ms (= 1 second) delay between > attempts. To get a connection I need to have at least 10 reconnect attempts > and set useExponentialBackOff=true. > > > > -- > View this message in context: > http://www.nabble.com/Multicast-Discovery-not-working-tp21344587p21484957.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. > > -- http://blog.garytully.com Open Source SOA http://FUSESource.com
