[Paraview-developers] Cray Dashboard Failure Diagnosis

pat marion pat.marion at kitware.com
Tue Mar 9 10:39:35 EST 2010


Did you compile cmakeified python?  The cmakelist doesn't build the time
module.  To fix, just edit the top level python CMakeList.txt, search for
datetime, and right after it add the line:

  add_python_module(time          MOD_SRCS Modules/timemodule.c )

Then you can turn it on by enabling BUILD_UNTESTED_MODULES.

Pat

On Tue, Mar 9, 2010 at 9:57 AM, Biddiscombe, John A. <biddisco at cscs.ch>wrote:

>  Pat et al
>
>
>
> Changing the interactor fixes the problem. Nice work.
>
>
>
> Seems like in vtkGraphicsFactory, that when OSMESA is defined, we could
> always return a TestingInteractor? does offscreen rendering ever use normal
> interactors?
>
>
>
> New problem:
>
> if I do (from an interactive MOM node)
>
> ctest –R Python
>
> then 360 tests run and 356 pass (not bad) (NB. in the python testing there
> is module “time”, which could not be found, so I had to comment out a bunch
> of walltime checks – is this a python build issue or a paraview one, I guess
> the datetime module which didn’t compile in python 2.5.2 needs fixing -
> yes?)
>
>
>
> if I do
>
> aprun –n 1 /full/path/to/ctest –R Python
>
>
>
> then the first Python test passes, but all the rest time out. It’s as
> though something is stopping me forking two python jobs one after the other.
> Ideally, I’d need to tell ctest to call
>
> aprun –n 1 test for all tests individually. Is there  a way of adding an
> extra command to the start of the test command line?
>
>
>
> nearly there ...
>
>
>
> Thanks in advance
>
>
>
> JB
>
>
>
>
>
> *From:* pat marion [mailto:pat.marion at kitware.com]
> *Sent:* 08 March 2010 23:05
> *To:* Biddiscombe, John A.
> *Cc:* Burlen Loring; ParaView Developers
> *Subject:* Re: [Paraview-developers] Cray Dashboard Failure Diagnosis
>
>
>
> Hi John,
>
> I'm back at the office now and I can confirm that it is
> vtkRenderWindowInteractor::New causing the segfaults.  You noted that the
> CellLocator test seems to crashes at exit, well if you look at the code it
> calls vtkRenderWindowInteractor::New after is does 'cout << 'Passed: a total
> of 9802 ray-sphere intersections detected.'
>
> The Widgets tests pass because they use vtkTestingObjectFactory to override
> vtkRenderWindowInteractor::New() to instantiate a vtkTestingInteractor.
>
> Here is a temp fix- edit VTK/Rendering/vtkGraphicsFactory.cxx and replace:
>
>     return vtkRenderWindowInteractor::New();
>
> with
>
>     return vtkTestingInteractor::New();
>
> Also add an #include "vtkTestingInteractor.h", then recompile and try your
> tests again.  Maybe sometime we'll have time to setup a static, python,
> osmesa, no x11 dashboard and fix these tests properly.
>
> Pat
>
> On Mon, Mar 8, 2010 at 4:42 AM, Biddiscombe, John A. <biddisco at cscs.ch>
> wrote:
>
> Pat,
>
>
>
> > the segfaults are from tests that call vtkRenderWindowInteractor::New(),
> causing an infinite loop inside vtkGraphicsFactory.cxx
>
>
>
> This is a good theory, Can anyone *who is not on holiday* J confirm if any
> of these tests (the passing ones) instantiate interactors (list of Widget
> test results at end)
>
>
>
> Burlen :
>
> >Can you run one of the tests in a debugger and see specifically where it
> segfaults? I'm sure that will give you the clue you're after. I guess this
> is obvious so maybe there is a reason you haven't done it.
>
>
>
> Quite right. For reasons unknown to me, I don’t get any symbol information
> and gdb tells me nothing. I’ve been asking around to see why my build has no
> debug info, but so far I’m clueless.
>
>
>
> Anyone :
>
> Important observation, looking at various outputs, I noticed that for
> example, CellLocator test does this
>
>
>
> 76: Test command: /users/biddisco/build/paraview/bin/GraphicsCxxTests
> CellLocator -D /apps/sandbox/VTKData -T
> /users/biddisco/build/paraview/VTK/Testing/Temporary -V
> Baseline/Graphics/CellLocator.png
>
> 76: Test timeout computed to be: 1500
>
> 76: Passed: a total of 9802 ray-sphere intersections detected.
>
> 1/1 Test #76: CellLocator ......................***Exception: SegFault
> 0.23 sec
>
>
>
> It actually does the test OK, but fails on exit. So now I’m lookingat
> cleanup code etc.
>
>
>
> JB
>
>
>
> ctest -R Widget
>
> Test project /users/biddisco/build/paraview
>
>       Start 533: HeaderTesting-Widgets
>
>  1/38 Test #533: HeaderTesting-Widgets .....................   Passed
> 2.72 sec
>
>       Start 534: BoxWidget
>
>  2/38 Test #534: BoxWidget .................................   Passed
> 6.04 sec
>
>       Start 535: BoxWidget2
>
>  3/38 Test #535: BoxWidget2 ................................***Failed
> 3.42 sec
>
>       Start 536: TestAngleWidget2D
>
>  4/38 Test #536: TestAngleWidget2D .........................   Passed
> 0.49 sec
>
>       Start 537: TestAngleWidget3D
>
>  5/38 Test #537: TestAngleWidget3D .........................   Passed
> 1.10 sec
>
>       Start 538: TestBalloonWidget
>
>  6/38 Test #538: TestBalloonWidget .........................   Passed
> 0.10 sec
>
>       Start 539: TestCaptionWidget
>
>  7/38 Test #539: TestCaptionWidget .........................   Passed
> 2.41 sec
>
>       Start 540: TestCheckerboardWidget
>
>  8/38 Test #540: TestCheckerboardWidget ....................   Passed
> 0.23 sec
>
>       Start 541: TestContourWidget2
>
>  9/38 Test #541: TestContourWidget2 ........................   Passed
> 0.13 sec
>
>       Start 542: TestDistanceWidget
>
> 10/38 Test #542: TestDistanceWidget ........................   Passed
> 1.37 sec
>
>       Start 543: TestImplicitPlaneWidget
>
> 11/38 Test #543: TestImplicitPlaneWidget ...................   Passed
> 4.48 sec
>
>       Start 544: TestImplicitPlaneWidget2
>
> 12/38 Test #544: TestImplicitPlaneWidget2 ..................   Passed
> 4.78 sec
>
>       Start 545: TestImplicitPlaneWidget2b
>
> 13/38 Test #545: TestImplicitPlaneWidget2b .................   Passed
> 0.15 sec
>
>       Start 546: TestOrientationMarkerWidget
>
> 14/38 Test #546: TestOrientationMarkerWidget ...............   Passed
> 4.59 sec
>
>       Start 547: TestRectilinearWipeWidget
>
> 15/38 Test #547: TestRectilinearWipeWidget .................   Passed
> 2.96 sec
>
>       Start 548: TestSeedWidget
>
> 16/38 Test #548: TestSeedWidget ............................   Passed
> 0.71 sec
>
>       Start 549: TestSliderWidget
>
> 17/38 Test #549: TestSliderWidget ..........................   Passed
> 1.74 sec
>
>       Start 550: TestSliderWidget2D
>
> 18/38 Test #550: TestSliderWidget2D ........................   Passed
> 2.08 sec
>
>       Start 551: TestTextWidget
>
> 19/38 Test #551: TestTextWidget ............................   Passed
> 0.10 sec
>
>       Start 552: TestParallelopipedWidget
>
> 20/38 Test #552: TestParallelopipedWidget ..................   Passed
> 0.28 sec
>
>       Start 553: TestSeedWidgetNonUniformRepresentations
>
> 21/38 Test #553: TestSeedWidgetNonUniformRepresentations ...   Passed
> 0.35 sec
>
>       Start 554: ImagePlaneWidget
>
> 22/38 Test #554: ImagePlaneWidget ..........................   Passed
> 4.51 sec
>
>       Start 556: TestImageActorContourWidget
>
> 23/38 Test #556: TestImageActorContourWidget ...............***Failed
> 2.03 sec
>
>       Start 557: TestSurfaceConstrainedHandleWidget
>
> 24/38 Test #557: TestSurfaceConstrainedHandleWidget ........   Passed
> 2.66 sec
>
>       Start 558: TestPolygonalRepresentationHandleWidget
>
> 25/38 Test #558: TestPolygonalRepresentationHandleWidget ...   Passed
> 2.52 sec
>
>       Start 560: TestSphereHandleWidget
>
> 26/38 Test #560: TestSphereHandleWidget ....................   Passed
> 21.94 sec
>
>       Start 561: TestImageTracerWidget
>
> 27/38 Test #561: TestImageTracerWidget .....................   Passed
> 4.37 sec
>
>       Start 562: TestLineWidget
>
> 28/38 Test #562: TestLineWidget ............................   Passed
> 0.96 sec
>
>       Start 563: TestLineWidget2
>
> 29/38 Test #563: TestLineWidget2 ...........................   Passed
> 1.15 sec
>
>       Start 564: TestLogoWidgetAlphaBlending
>
> 30/38 Test #564: TestLogoWidgetAlphaBlending ...............***Failed
> 0.12 sec
>
> File "../../../Testing/Temporary/TestLogoWidgetAlphaBlending.png" not
> found.
>
> File "../../../Testing/Temporary/TestLogoWidgetAlphaBlending.diff.png" not
> found.
>
>       Start 565: TestLogoWidgetDepthPeeling
>
> 31/38 Test #565: TestLogoWidgetDepthPeeling ................***Failed
> 0.11 sec
>
> File "../../../Testing/Temporary/TestLogoWidgetDepthPeeling.png" not found.
>
> File "../../../Testing/Temporary/TestLogoWidgetDepthPeeling.diff.png" not
> found.
>
>       Start 567: TestPlaneWidget
>
> 32/38 Test #567: TestPlaneWidget ...........................   Passed
> 2.77 sec
>
>       Start 568: TestPointWidget
>
> 33/38 Test #568: TestPointWidget ...........................   Passed
> 0.86 sec
>
>       Start 569: TestScalarBarWidget
>
> 34/38 Test #569: TestScalarBarWidget .......................   Passed
> 2.58 sec
>
>       Start 570: TestSplineWidget
>
> 35/38 Test #570: TestSplineWidget ..........................   Passed
> 1.86 sec
>
>       Start 571: TestTensorProbeWidget
>
> 36/38 Test #571: TestTensorProbeWidget .....................   Passed
> 3.68 sec
>
>       Start 573: TestHandleWidget
>
> 37/38 Test #573: TestHandleWidget ..........................   Passed
> 1.97 sec
>
>       Start 575: TestImagePlaneWidgetPython-image
>
> 38/38 Test #575: TestImagePlaneWidgetPython-image ..........***Failed
> 5.44 sec
>
>
>
> 87% tests passed, 5 tests failed out of 38
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20100309/c68d6405/attachment-0001.htm>


More information about the Paraview-developers mailing list