[petsc-dev] Improving nightly builds for maint

[Satish Balay]

On Wed, 18 Oct 2017, Lisandro Dalcin wrote:

> On 18 October 2017 at 22:36, Satish Balay <balay at mcs.anl.gov> wrote:
> > On Wed, 18 Oct 2017, Lisandro Dalcin wrote:
> >
> > Alternate is to remove/increase the timout. It doesn't work for some
> > builds anyway [esp valgrind runs]
> >
> > This feature relies on sending 'kill' message to the process it
> > started - i.e mpiexec. But some mpiexecs when killed - don't kill the
> > child processes cleanly]
> >
> > $ git diff |cat
> > diff --git a/bin/maint/buildtest b/bin/maint/buildtest
> > index 31b5a3a028..e4bbf96db8 100755
> > --- a/bin/maint/buildtest
> > +++ b/bin/maint/buildtest
> > @@ -104,7 +104,7 @@ if ( $? ) then
> >  else
> >    set VALGRIND=`echo MPIEXEC="${PETSC_DIR}/bin/petscmpiexec -valgrind"`
> >    setenv PETSCVALGRIND_OPTIONS "--suppressions=${PETSC_DIR}/bin/maint/petsc-val.supp"
> > -  set TIMEOUT=720
> > +  set TIMEOUT=7200
> >  endif
> >
> 
> No, please don't!  Be green, Satish!
> 
> What's the point of running a 3D problem under valgrind for 2 hours?


BTW: Just so clarify - eventhough you see a 'killed' message -
currently those jobs are running to completion. [however long it takes
under valgrind]

Satish


> 
> I think we should somehow support
> 
> requires: !valgrind
> 
> and add special code in the test harness (maybe using some environment
> var?) to skip running these tests...
> 
> 
>