[petsc-dev] [petsc-maint #103192] Problem with petsc-dev "make test" failure

[Dave Nystrom]

Barry Smith writes:
 > What the???  That's not how a PETSc error looks! Also if this is the
 > default run it is not using CUSP so it shouldn't possibly generate an
 > error related to CUSP.
 > 
 > This is hopeless, I'm stripping GPU and pthread support from PETSc, there
 > is no way our little team can support all this complexity.

I hope you don't remove the GPU support since I am making heavy use of it and
it has seemed to work very well till today.

I'm not using the pthread support.  I'd be happy to rebuild without it and
try again.

Dave

 > Barry
 > 
 > [0]PETSC ERROR: VecScatterBegin_4() line 53 in src/vec/vec/utils//home/wdn/Projects/Petsc/src/hg/petsc-dev/include/../src/vec/vec/utils/vpscat.h
 > [0]PETSC ERROR: VecScatterBegin() line 1573 in src/vec/vec/utils/vscat.c
 > [1]PETSC ERROR: VecScatterBegin_4() line 53 in src/vec/vec/utils//home/wdn/Projects/Petsc/src/hg/petsc-dev/include/../src/vec/vec/utils/vpscat.h
 > [1]PETSC ERROR: VecScatterBegin() line 1573 in src/vec/vec/utils/vscat.c
 > [1]PETSC ERROR: MatMultTranspose_MPIMAIJ_dof() line 2919 in src/mat/impls/maij/maij.c
 > [1]PETSC ERROR: MatMultTranspose() line 2234 in src/mat/interface/matrix.c
 > [1]PETSC ERROR: MatRestrict() line 7440 in src/mat/interface/matrix.c
 > [1]PETSC ERROR: DMCreateInterpolationScale() line 45 in src/dm/impls/da/dainterp.c
 > [1]PETSC ERROR: DMMGSetSNES() line 763 in src/snes/utils/damgsnes.c
 > [1]PETSC ERROR: DMMGSetSNESLocal_Private() line 980 in src/snes/utils/damgsnes.c
 > [1]PETSC ERROR: main() line 145 in src/snes/examples/tutorials/ex19.c
 > [0]PETSC ERROR: MatMultTranspose_MPIMAIJ_dof() line 2919 in src/mat/impls/maij/maij.c
 > [0]PETSC ERROR: MatMultTranspose() line 2234 in src/mat/interface/matrix.c
 > [0]PETSC ERROR: MatRestrict() line 7440 in src/mat/interface/matrix.c
 > [0]PETSC ERROR: DMCreateInterpolationScale() line 45 in src/dm/impls/da/dainterp.c
 > [0]PETSC ERROR: DMMGSetSNES() line 763 in src/snes/utils/damgsnes.c
 > [0]PETSC ERROR: DMMGSetSNESLocal_Private() line 980 in src/snes/utils/damgsnes.c
 > application called MPI_Abort(MPI_COMM_WORLD, 205796944) - process 1
 > [cli_1]: [0]PETSC ERROR: main() line 145 in src/snes/examples/tutorials/ex19.c
 > aborting job:
 > application called MPI_Abort(MPI_COMM_WORLD, 205796944) - process 1
 > application called MPI_Abort(MPI_COMM_WORLD, 629412112) - process 0
 > [cli_0]: aborting job:
 > application called MPI_Abort(MPI_COMM_WORLD, 629412112) - process 0
 > 
 > 
 > 
 > On Jan 21, 2012, at 3:37 PM, Dave Nystrom wrote:
 > 
 > > See attached.
 > > 
 > > <make_ex19.log><run_ex19.log>
 > > Barry Smith writes:
 > >> Damn thing is not printing the entire error message. Still some wrong use
 > >> of CHKERRCUSP() that should be CHKERRQ().
 > >> 
 > >> cd src/snes/examples/tutorials; make ex19; mpiexec -n 2 ./ex19 and send
 > >> all output to petsc-dev
 > >> 
 > >> Barry
 > >> 
 > >> On Jan 21, 2012, at 3:03 PM, Dave Nystrom wrote:
 > >> 
 > >>> In my latest attempt, petsc-dev seems to build fine but fails the "make test"
 > >>> case of running with 2 processes.  See attached files including my petsc-dev
 > >>> build script and its log file.
 > >>> 
 > >>> Thanks,
 > >>> 
 > >>> Dave
 >