activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Leonard Gestrin <Leonard.Gest...@markettools.com>
Subject Failover question (hangs appserver)
Date Tue, 15 Sep 2009 00:17:00 GMT
Hello, 
We are using activemq 5.2.0. Our webapp runs in tomcat 5. We don't use spring JMS support,
however we do configure JMS listener class as a singleton in spring factory.

Recently, we configured failover capabilities as 

jms.broker.url=failover:(tcp://sfltlg-halsg3b1:61616,tcp://sfltlg-hlsag3b1:61616)?randomize=false&maxReconnectAttempts=1&initialReconnectDelay=30000

We set maxReconnectAttempts as 1 because we have our own connection retry mechanism that gets
triggered on JMS exception; 
However, if both brokers are down, tomcat seems to hang and goes to 100% CPU. It appears as
if client connection code tries to connect, fails and tries to reconnect right away. 
This behavior happens only if we use failover notation. If we revert back to 

jms.broker.url=://sfltlg-halsg3b1:61616 

and if we shutdown broker, webapp behaves OK -no memory/CPU spikes.

I can post our custom code, if it would help. 
Any suggestions what might be wrong with our configuration would be greatly appreciated.
Thanks
Leonard


Mime
View raw message