[Paraview-developers] Cray Dashboard Failure Diagnosis

[pat marion]

Here is a wild guess:  the segfaults are from tests that call
vtkRenderWindowInteractor::New(), causing an infinite loop inside
vtkGraphicsFactory.cxx when compiled with VTK_USE_OSMESA.  I'm on vacation
right now and can't verify any of this, I shouldn't even be on email...

Pat

On Sun, Mar 7, 2010 at 2:08 PM, Burlen Loring <burlen.loring at gmail.com>wrote:

> Hi John,
>
> 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.
>
> Burlen
>
>
> Biddiscombe, John A. wrote:
>
>> Dear Anyone,
>>
>> After much head scratching and recomiling hundreds of times with different
>> options, I'm still no closer to solving an issue (Any many many thanks to
>> Pat who's instructions I've been following, but I still get the same
>> segfaults as when I did it myself using a slightly less 'strict' procedure).
>> http://cdash.cscs.ch/index.php?project=pv-meshless&date=2010-03-06
>> Top dashboard from Cray Compute node linux
>> Faill =414       Pass = 325
>>
>> I don't know why these tests are all segfaulting (a few fail, that's ok,
>> fail is alright, but segfaults I want to knwo why). Does anyone know what
>> theses ones do that the others do not. Is one library in particular called
>> in these tests that the others do not?
>>
>> Any clues gratefully accepted
>>
>> BIG NOTE : when ctest runs, I'm using an interactive login node, so tests
>> which use MPI are being started with "aprun program" and are run on a
>> compute node (some even pass).
>> Tests which do not use MPI are being run on the login/service node - they
>> have the same CPUs etc, so all should be well, but it's a clue anyway
>> (however some of these pass too, so not a very good clue).
>>
>> Thanks
>>
>> JB
>> _______________________________________________
>> Paraview-developers mailing list
>> Paraview-developers at paraview.org
>> http://public.kitware.com/mailman/listinfo/paraview-developers
>>
>>
>
> _______________________________________________
> Paraview-developers mailing list
> Paraview-developers at paraview.org
> http://public.kitware.com/mailman/listinfo/paraview-developers
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview-developers/attachments/20100307/d7366305/attachment.htm>