There is a *startupMaxReconnectAttempts *option on the failover transport that allow the transport to fail early on the first connect attempt so that spring gets a chance to do retries rather than have the connection block pending the broker start.
see: https://issues.apache.org/activemq/browse/AMQ-2385 On 2 July 2010 09:54, Arjen van der Meijden <acmmail...@tweakers.net> wrote: > Hi, > > We have a simple single-server ActiveMQ-setup with multiple consumers, some > of which use Java and connect via JMS. As the ActiveMQ-server might some > times need a restart or some other down time we configured it to use the > failover protocol (failover:tcp://localhost:61616), so it will reconnect as > soon as its available again and will also keep the connection alive when > there are long periods of silence. > > All this is managed via Spring's normal connection handling. > > This works fine, unless the ActiveMQ-server is not available during the > initial start. Rather than just going into the background and waiting for > the availability of a connection it blocks the start-up without any error > message. > In my case its a spring-managed Servlet that will receive some cache-update > messages via jsm-topics. But rather than connecting in the background to > activemq, the entire Tomcat-instance is blocked and although it will accept > new http-connections, it will do absolutely nothing with them. > > My spring-configuration is like this: > > <bean id="amqfactory" class="o.a.activemq.ActiveMQConnectionFactory"> > <constructor-arg value="" /> <!-- username --> > <constructor-arg value="" /> <!-- password --> > <constructor-arg value="failover:tcp://localhost:61616" /> <!-- broker --> > </bean> > > <jms:listener-container connection-factory="amqfactory" > destination-type="topic"> > <jms:listener destination="some_topic" ref="someMessageListener" /> > </jms:listener-container> > > I'm not really sure what the best solution would be, but I'd expect to get > some ERROR-message at the very least. It is the reason my application won't > start indefinitely (or until activemq is started, whichever comes first). > Its something you easily overlook during debugging and it is very hard to > detect when you don't know what you're looking for (or even if you do, but > just forget to check activemq is running and/or a connection is made). > Logging should be possible, as I do get these kind of messages if it does > work: > > 2010-07-02 09:36:00,095 [ActiveMQ Task] INFO > o.a.activemq.transport.failover.FailoverTransport - Successfully connected > to tcp://localhost:61616 > > Perhaps the easiest fix would be to display a ERROR-message the first time > it can't connect and than fall back to DEBUG-messages. > > Best regards, > > Arjen > -- http://blog.garytully.com Open Source Integration http://fusesource.com
