Running from the command it's probably not the end of the world. But I always run this to make sure long-lived processes work and it generally finds any nasty stuff hiding because using the same Maven instance over and over doesn't last very long with anything leaking badly. Igor and I are getting the same results but I think it's a function of a couple things it seems. First we don't dispose of the plugin class realms and purge them like we do in m2e, and then I believe there are some plugins that are loading more things and just blowing the permgen. So we're trying to figure out how to purge the class realm cache. This will take care of the memory problem, and then I'll start looking at the speed.
On Nov 24, 2012, at 1:02 PM, Olivier Lamy <[email protected]> wrote: > 2012/11/24 Jason van Zyl <[email protected]>: >> Most things are ready release-wise but I found a rather nasty performance >> regression. Seems to be two contributing factors: differing versions of >> Wagon seem to contribute to a large slow down and there appears to be some >> massive classloader leakage which is probably something I did recently. >> Running the ITs in embedded mode always reveals these problems and they show >> the following. >> > > What do you mean exactly with "Running the ITs in embedded " ? > Using embedded profile when running core its ? (if yes that runs fine > here: > https://builds.apache.org/view/M-R/view/Maven/job/core-integration-testing-maven-3-embedded/) > > ? > >> 3.0.3 >> Tests run: 714, Failures: 4, Errors: 2, Skipped: 0, Time elapsed: 342.985 >> sec <<< FAILURE! >> >> 3.0.4 with Wagon 2.2 >> Tests run: 714, Failures: 4, Errors: 2, Skipped: 0, Time elapsed: 371.001 >> sec <<< FAILURE! >> >> 3.0.4 with Wagon 2.3 >> Tests run: 714, Failures: 4, Errors: 2, Skipped: 0, Time elapsed: 472.893 >> sec <<< FAILURE! >> >> 3.1.0 with Wagon 2.3 >> Tests run: 656, Failures: 13, Errors: 233, Skipped: 0, Time elapsed: 576.268 >> sec <<< FAILURE! >> >> With trunk everything is tanking with permgen failures so I'm trying to >> track this down with Igor. So I would have like to have put 3.1.0 up for >> vote today but I'm going to track this down first and try and get the >> performance back to at least 3.0.3 levels. I'm also going to put the >> performance framework Igor developed back up on a machine. >> >> Thanks, >> >> Jason >> >> ---------------------------------------------------------- >> Jason van Zyl >> Founder & CTO, Sonatype >> Founder, Apache Maven >> http://twitter.com/jvanzyl >> --------------------------------------------------------- >> >> I never make the mistake of arguing with people for whose opinions I have no >> respect. >> >> -- Edward Gibbon >> >> >> >> >> > > > > -- > Olivier Lamy > Talend: http://coders.talend.com > http://twitter.com/olamy | http://linkedin.com/in/olamy > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > Thanks, Jason ---------------------------------------------------------- Jason van Zyl Founder & CTO, Sonatype Founder, Apache Maven http://twitter.com/jvanzyl --------------------------------------------------------- believe nothing, no matter where you read it, or who has said it, not even if i have said it, unless it agrees with your own reason and your own common sense. -- Buddha
