Hi Carsten, It looks like we have a different classpath when it fails and when it doesn't Maybe you can put a mvn dependency:tree in the job and check how it looks in the two cases.
Also, AFAIK the fact that the maven dependency resolution mechanism puts the ones defined in the project pom and replaces the transient ones is an implementation detail. Or it was not enforced in maven 2. Or something like that. Or maybe I'm wrong :) anyway, it would make sense to see the maven version in the two cases, in case the job is run on different jenkins slaves. Let me know if I can help in any way. -Andrei On Mon, Sep 19, 2016 at 11:26 AM Carsten Ziegeler <cziege...@apache.org> wrote: > >> Hi, > >> > >> for quiet some time now, it seems that the IT tests of the i18n module > >> fail in reactor build (at least, Jenkins seems to fail all the time). > >> Interestingly, I can only reproduce this from time to time. However when > >> it fails, it fails with: > >> > >> Tests run: 3, Failures: 0, Errors: 2, Skipped: 0, Time elapsed: 26.691 > >> sec <<< FAILURE! - in org.apache.sling.i18n.it.ResourceBundleProviderIT > >> testChangesDetection(org.apache.sling.i18n.it.ResourceBundleProviderIT) > >> Time elapsed: 20.095 sec <<< ERROR! > >> java.lang.NoClassDefFoundError: org/junit/runner/notification/Failure > >> > >> test_dummy(org.apache.sling.i18n.it.ResourceBundleProviderIT) Time > >> elapsed: 0 sec <<< ERROR! > >> java.lang.NullPointerException > >> > >> Especially the NoClassDefFoundError looks strange to me. > >> > >> When building the module directly, it always passes. > >> > >> Anyone an idea of what needs to be changed? > >> > > It seems that the problem is fixed by moving the junit dependency to the > > top. > > Let's see what Jenkins says... > > > > Apparently that didn't fix the problem in Jenkins. > > I suggest we turn off Jenkins completely > > Carsten > > -- > Carsten Ziegeler > Adobe Research Switzerland > cziege...@apache.org > >
