activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "jayson.minard" <jayson.min...@gmail.com>
Subject High volume traffic on a broker can lead to ConcurrentModificationException
Date Sat, 11 Oct 2008 22:55:53 GMT

I posted this on the user side but it is probably better intended for the
developer list so posting here.  Sorry for the double-post.  More
information has been added to this versino.

Hello everyone, 

We have a network of 5 brokers communicating via TCP on the default
connector.  We are running around 300,000 messages per hour on each broker
through a series of queues.  On every broker there is a point in time where
it hits a ConcurrentModificationException (see
https://issues.apache.org/activemq/browse/AMQ-1971) and they no longer
dispatch messages afterwards thereby freezing all of the queues. 

This seems like a critical bug in a release candidate (or soon to be
candidate) so I wanted to post it to the list to see if others have
encountered it, or if there were ideas on a quicker fix.  The code is in the
midst of a few resource locks and is in a very important part of the broker,
so I am nervous about attempting a patch without knowing more about how the
inner workings behave 

Broker is 5.2.0 built from the SVN tag of that name.  Client uses the 5.2.0
Maven POM on the staging server for the 5.2.0 release.  Running JDK 1.6 on 8
core linux machines.  

As an additional test case I ran just one broker on one machine with a local
set of producer/consumers and the problem exists there as well without the
network of brokers.

Any thoughts on the problem? 

-- Jayson

-- 
View this message in context: http://www.nabble.com/High-volume-traffic-on-a-broker-can-lead-to-ConcurrentModificationException-tp19937232p19937232.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.


Mime
View raw message