On May 12, 2013, at 9:29 AM, David Blevins <david.blev...@gmail.com> wrote:
> And it was a good fix. This change was to AnnotatedType itself which is > where the root issue lived. > > Until OWB is rewritten or changed to not keep and use AnnotatedType at > runtime, it needs to be thread safe. And to be fair, it's still not fully thread-safe. Just "more" thread safe :) -David > > > -David > > On May 11, 2013, at 11:42 PM, Romain Manni-Bucau <rmannibu...@gmail.com> > wrote: > >> I already fixed it David... >> Le 12 mai 2013 02:14, "David Blevins" <david.blev...@gmail.com> a écrit : >> >>> >>> On May 3, 2013, at 7:18 AM, Thiago Veronezi <thi...@veronezi.org> wrote: >>> >>>>>> The pool ones are weird and i never have it locally :( >>>> I have a theory: clock issues. It seems that the clock runs slower in VMs >>>> (google "clock drift virtualbox" or "clock drift vmware") >>> >>> Very interesting! >>> >>> Found the root issue which is that the AnnotatedType object in OWB was not >>> thread safe. Was doing lazy initialization without proper synchronization. >>> There was a synchronized block on the init() method, but no protection on >>> the read methods. If you only synchronize the writes and not the reads, >>> you don't get very far :) >>> >>> The pool tests work really hard to create race conditions. >>> >>> >>> -David >>> >>> >>>> On Thu, May 2, 2013 at 3:44 PM, Romain Manni-Bucau < >>> rmannibu...@gmail.com>wrote: >>>> >>>>> Yep fixed rest issue ;) >>>>> >>>>> The pool ones are weird and i never have it locally :( >>>>> Le 2 mai 2013 21:01, "Thiago Veronezi" <thi...@veronezi.org> a écrit : >>>>> >>>>>> Today the linux box runs the rest-example tests fine. But now it >>>>> complaints >>>>>> about the "StatelessPoolStatsTest", like you've mentioned. The windows >>>>> box >>>>>> complains about the "PoolTest". >>>>>> >>>>>> Note that they can complain about different tests, depending on their >>>>> mood. >>>>>> Probably it is something related to the performance and load of the >>>>>> computer in which the tests run. >>>>>> >>>>>> []s, >>>>>> Thiago. >>>>>> >>>>>> >>>>>> >>>>>> On Wed, May 1, 2013 at 10:18 AM, Thiago Veronezi <thi...@veronezi.org >>>>>>> wrote: >>>>>> >>>>>>> Hi Romain, >>>>>>> >>>>>>> I don't get successful tests in my localhost for quite some time. :) >>>>>> Oddly >>>>>>> the same code runs fine on our build servers (at least, some times). >>>>>>> I've just tested it again on a windows 8 and linux box. Both failed, >>>>> but >>>>>>> for different reasons. >>>>>>> >>>>>>> On win8, something went wrong in the tests of the openejb-core module. >>>>>>> The linux build complains about the rest-example tests. >>>>>>> >>>>>>> I will be able to post more details later this evening. >>>>>>> >>>>>>> []s, >>>>>>> Thiago >>>>>>> >>>>>>> >>>>>>> >>>>>>> On Tue, Apr 30, 2013 at 4:50 AM, Romain Manni-Bucau < >>>>>> rmannibu...@gmail.com >>>>>>>> wrote: >>>>>>> >>>>>>>> Hi guys, >>>>>>>> >>>>>>>> our pool tests regularly fail (StatelessPoolStatsTest for instance) >>> on >>>>>> our >>>>>>>> CI and stops the full build of tomee. >>>>>>>> >>>>>>>> locally it passes for me, anyone tested? >>>>>>>> >>>>>>>> do we comment them or use a custom runner to retry if failing to >>> force >>>>>> the >>>>>>>> build to go ahead? >>>>>>>> >>>>>>>> PS: cdi-tomee passes now locally ;) >>>>>>>> >>>>>>>> Romain Manni-Bucau >>>>>>>> Twitter: @rmannibucau >>>>>>>> Blog: http://rmannibucau.wordpress.com/ >>>>>>>> LinkedIn: http://fr.linkedin.com/in/rmannibucau >>>>>>>> Github: https://github.com/rmannibucau >>>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>> >>> >>> >