Details
Major Description
I've got a configuration with two master/slave setups consisting of 3 ActiveMQ instances each. They are deployed on three servers, with one ActiveMQ instance from each master/slave setup on every server. They are using the leveldb and zookeeper. Everything works fine.
Now I've got the strange behaviour that when I add a network connector to each ActiveMQ instance like this:
networkConnector name="toMasterSlave02" dynamicOnly="true" uri="masterslave:(tcp://host1:61617,tcp://host2:61617,tcp://host3:61617)"
When you now restart the master of the master/slave setup that is targeted by the above network connector the cpu load on the current master goes and stays up at 100%, i.e. it uses one CPU per configured transportConnector.
Now the explanation, mostly copied from http://activemq.2283324.n4.nabble.com/High-CPU-load-with-network-connector-failover-transport-tp4691798.html
When one of the brokers is restarted, the other broker uses ~400% CPU. The cause is the FailoverTransport reconnectTask spinning, and nothing is stopping the task.
Reverting this fix made for
AMQ-5315, while it does reintroduce the NullPointerException, does handle failover properly without spinning:
https://git1-us-west.apache.org/repos/asf/activemq/repo?p=activemq.git;a=commitdiff;h=c391321d1b5b59542d847717654b0d4dba54cf2f <https://git1-us-west.apache.org/repos/asf/activemq/repo?p=activemq.git;a=commitdiff;h=c391321d1b5b59542d847717654b0d4dba54cf2f>The reason it works after reverting that change is the NullPointerException is caught, -> serviceLocalException() -> ServiceSupport.dispose(getControllingService()); with the fix made in
AMQ-5315, the dispose() call is never made.
Sorry, but I've got no clue how to provide a unit test for this. Maybe someone else can help.