Bug 178071

Created attachment 92045 [details]
HTML log

Created attachment 92046 [details]
Screenshot

Created attachment 92048 [details]
Another recent failure which may or may not be related

NB-Core-Build #3682 failed too, looks similar.

Created attachment 92074 [details]
Yet another recent failure which may be related

Jesse,
First failure is timeout The test testGCProjects did not finish in 500000ms
as a result of decreasing timeout 

reassigned to Editor for evaluation

same issue on cnd-main builds
http://bertram.netbeans.org/hudson/job/cnd-main/2342/testReport/
http://bertram.netbeans.org/hudson/job/cnd-main/2343/testReport/
http://bertram.netbeans.org/hudson/job/cnd-main/2345/testReport/

testGCDocuments is failing pretty consistently on the cdev builder.

IMO we should not be running MemoryValidationTest on team builders. It suffers from the same reliability issues as other commit validation tests. (While the recent failures may well be due to some real code regression, the test clearly did not catch it reliably enough to stop it from propagating out of the original team area.)

Whatever hotfix is made should be done against a parent changeset from main-silver so it can be quickly pulled into other team repos.

Disabling test in core-main #94e7c9950c88 (can be pulled into other team repos) until it can be evaluated and fixed.

I am not sure but looking at our jet-main repo failures e.g. here
http://bertram.netbeans.org/hudson/job/jet-main/1638/testReport/org.netbeans.test.ide/MemoryValidationTest/testGCDocuments/?
it seems that someone is holding the editor window from main window in the test and thus the document is not GC-ed which sounds like ok from the editor point of view (but not ok from the winsys or the test itself point of view).

There is also a picture in the build pointed by David:
http://bertram.netbeans.org/hudson/job/jet-main/1638/artifact/java.kit/build/test/qa-functional/work/o.n.t.i.M/testGCDocuments/screen.png
It shows no signs of open editors. So there is no reason for the EditorPane to be hold.

Btw. we are mixing two failures. One in testGCProjects:

private static java.util.Collection org.netbeans.modules.parsing.impl.Schedulers.taskSchedulers->
java.util.Collections$UnmodifiableRandomAccessList@a282b9-list->
java.util.ArrayList@791eab-elementData->
[Ljava.lang.Object;@1725e1b-[2]->
org.netbeans.modules.parsing.impl.SelectedNodesScheduler@1e13d2-source->
org.netbeans.modules.parsing.api.Source@70fd39-cache->
org.netbeans.modules.parsing.impl.SourceCache@1cc8531-parser->
org.netbeans.modules.java.source.parsing.JavacParser@1e8415c-cpInfo->
org.netbeans.api.java.source.ClasspathInfo@9bf406-srcClassPath->
org.netbeans.api.java.classpath.ClassPath@11cd164-impl->
org.netbeans.spi.java.project.support.ClassPathProviderMerger$ProxyClassPathImplementation@aa6d7a-mainProvider->
org.netbeans.modules.java.api.common.classpath.ClassPathProviderImpl@199f402-sourceRoots->
org.netbeans.modules.java.api.common.SourceRoots@542767-helper->
org.netbeans.modules.java.api.common.ant.UpdateHelper@1d99ebf-updateProject->
org.netbeans.modules.java.j2seproject.UpdateProjectImpl@19ab7e2-project->
org.netbeans.modules.java.j2seproject.J2SEProject@1e461c

one in testGCDocuments:

Caused by: junit.framework.AssertionFailedError: Are all documents GCed? TextDocument:
static java.util.Vector sun.awt.motif.MFramePeer.allFrames->
java.util.Vector@196444d-elementData->
[Ljava.lang.Object;@2d119d-[0]->
sun.awt.motif.MFramePeer@1ea2c7-target->
org.netbeans.core.windows.view.ui.MainWindow@8a6465-dispatcher->
java.awt.LightweightDispatcher@17f1fcb-mouseEventTarget->
org.openide.text.QuietEditorPane@a013f6-model->
org.netbeans.modules.editor.NbEditorDocument@141f799

Integrated into 'main-golden', will be available in build *200912050227* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
Changeset: http://hg.netbeans.org/main/rev/94e7c9950c88
User: Jesse Glick <jglick@netbeans.org>
Log: #178071 hotfix: disable MVT until failures can be properly evaluated.

(In reply to comment #11)
> Btw. we are mixing two failures. One in testGCProjects:
> 
> private static java.util.Collection
> org.netbeans.modules.parsing.impl.Schedulers.taskSchedulers->
> java.util.Collections$UnmodifiableRandomAccessList@a282b9-list->
> java.util.ArrayList@791eab-elementData->
> [Ljava.lang.Object;@1725e1b-[2]->
> org.netbeans.modules.parsing.impl.SelectedNodesScheduler@1e13d2-source->
> ...

I'm not sure why SelectedNodesScheduler holds reference to Source. Hanzi?


> one in testGCDocuments:
> 
> Caused by: junit.framework.AssertionFailedError: Are all documents GCed?
> TextDocument:
> static java.util.Vector sun.awt.motif.MFramePeer.allFrames->
> java.util.Vector@196444d-elementData->
> [Ljava.lang.Object;@2d119d-[0]->
> sun.awt.motif.MFramePeer@1ea2c7-target->
> org.netbeans.core.windows.view.ui.MainWindow@8a6465-dispatcher->
> java.awt.LightweightDispatcher@17f1fcb-mouseEventTarget->
> org.openide.text.QuietEditorPane@a013f6-model->
> org.netbeans.modules.editor.NbEditorDocument@141f799

Obviously not a problem in the editor.

The not in editor part has been separated into  Bug 178592.

I guess P2 after hotfix in 94e7c9950c88.

I will try this.

http://hg.netbeans.org/jet-main/rev/27ce0d370c33

Integrated into 'main-golden', will be available in build *200912161400* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
Changeset: http://hg.netbeans.org/main/rev/27ce0d370c33
User: David Strupl <dstrupl@netbeans.org>
Log: #178071: SelectedNodesScheduler holds reference to Source for to long.

Failures have been evaluated but somebody forgot to enable the test.