On Thu, Sep 20, 2012 at 9:42 AM, Claus Ibsen <[email protected]> wrote: > Hi > > So in recent time I have been polishing the AMQ trunk codebase to > prepare it for the 5.7 release. > > I went as far to run test suites on Windows with Java 7 :) > There were one tests which failed all the time, but the other tests > can pass as well. > > I have also tested with Java7 on my osx, and the Apache CI servers > have jobs for testing with Java7 as well. I think they run on ubuntu. > Anyway I think we are shaping up really well. > > Anyone got anything they would like to get address before we start > considering cutting a RC ? > > I got some tests which identified some leaks in the activemq-pool, > which I would like to polish up and bring into the codebase. > As they require a separate broker to run, we could possible have it to > be run manually. But then ppl can run it from time to time, to avoid > regressions or introducing new leaks. The test is from a JMS client > point of view using Camel. >
I ran the test that identified the leaks in the activemq-pool, and this time it passed with flying colors. The rate of throughput in messaging was constant, as well the memory usage was at a minimum. And when all the messages was done, the GC could claim back to 4mb usage of heap size. The test ran for only 10 minutes, and did 50 x 20 req/reply messaging with AMQ using Camel over TCP. Last time with the leaks after 3 minutes or less it will start to cause memory issues and slow down. I have attached a screenshot. > > > -- > Claus Ibsen > ----------------- > Red Hat, Inc. > FuseSource is now part of Red Hat > Email: [email protected] > Web: http://fusesource.com > Twitter: davsclaus > Blog: http://davsclaus.com > Author of Camel in Action: http://www.manning.com/ibsen -- Claus Ibsen ----------------- Red Hat, Inc. FuseSource is now part of Red Hat Email: [email protected] Web: http://fusesource.com Twitter: davsclaus Blog: http://davsclaus.com Author of Camel in Action: http://www.manning.com/ibsen
