Satish can provide more details on how you can easily run in the EXACT 
environment where something broke to debug it much faster.  The model is 

1)  ssh [email protected]

2) ssh testmachine  (testmachine is always the end part of the name of the log 
file)

3) cd to either  /sandbox/petsc/petsc.test  or /home/petsc/petsc.test depending 
on the machine

4) git fetch

5) git checkout the broken branch 

6) set PETSC_ARCH=arch of broken machine, this is in the name of the log 
filename

7) ./config/examples/${PETSC_ARCH}.py

8) build PETSc and debug away.

  If you have any trouble with this let Satish and I know. It is the intention 
that debugging on the test machines should be very straightforward and not 
require getting help from anyone or following convoluted instructions.

  Barry

> On Oct 30, 2015, at 2:15 PM, Matthew Knepley <[email protected]> wrote:
> 
> I ran it through valgrind on my machine with no problems. Checking logs
> 
>    Matt
> 
> On Thu, Oct 29, 2015 at 5:09 PM, Barry Smith <[email protected]> wrote:
> 
> <   [3] Roots referenced by my leaves, by rank
> < Symmetric gradient null space: PASS
> < Function tests pass for order 0 at tolerance 1e-10
> < Function tests pass for order 0 derivatives at tolerance 1e-10
> ---
> > [2]PETSC ERROR: #1 PetscCommDuplicate() line 178 in 
> > /usr/home/balay/petsc.clone-3/src/sys/objects/tagm.c
> > [2]PETSC ERROR: #2 PetscHeaderCreate_Private() line 60 in 
> > /usr/home/balay/petsc.clone-3/src/sys/objects/inherit.c
> > [2]PETSC ERROR: #3 PetscSFCreate() line 44 in 
> > /usr/home/balay/petsc.clone-3/src/vec/is/sf/interface/sf.c
> > [2]PETSC ERROR: #4 DMPlexDistribute() line 1562 in 
> > /usr/home/balay/petsc.clone-3/src/dm/impls/plex/plexdistribute.c
> > [2]PETSC ERROR: #5 CreateMesh() line 232 in 
> > /usr/home/balay/petsc.clone-3/src/dm/impls/plex/examples/tests/ex3.c
> > [2]PETSC ERROR: #6 main() line 911 in 
> > /usr/home/balay/petsc.clone-3/src/dm/impls/plex/examples/tests/ex3.c
> > [2]PETSC ERROR: PETSc Option Table entries:
> > [2]PETSC ERROR: -dim 3
> > [2]PETSC ERROR: -dm_plex_max_projection_height 2
> > [2]PETSC ERROR: -dm_view ascii::ASCII_INFO_DETAIL
> > [2]PETSC ERROR: -malloc_dump
> > [2]PETSC ERROR: -nox
> > [2]PETSC ERROR: -nox_warning
> > [2]PETSC ERROR: -num_comp 3
> > [2]PETSC ERROR: -petscpartitioner_type simple
> > [2]PETSC ERROR: -petscspace_order 1
> > [2]PETSC ERROR: -petscspace_poly_tensor
> > [2]PETSC ERROR: -qorder 1
> > [2]PETSC ERROR: -simplex 0
> > [2]PETSC ERROR: -test_fe_jacobian
> > [2]PETSC ERROR: -tree
> > [2]PETSC ERROR: ----------------End of Error Message -------send entire 
> > error message to [email protected]
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 2
> > [cli_2]: aborting job:
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 2
> >
> > ===================================================================================
> > =   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
> > =   PID 15231 RUNNING AT wii
> > =   EXIT CODE: 1
> > =   CLEANING UP REMAINING PROCESSES
> > =   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
> > ===================================================================================
> /usr/home/balay/petsc.clone-3/src/dm/impls/plex/examples/tests
> Possible problem with with runex3_nonconforming_tensor_3, diffs above
> 
> 
> 
> 
> -- 
> What most experimenters take for granted before they begin their experiments 
> is infinitely more interesting than any results to which their experiments 
> lead.
> -- Norbert Wiener

Reply via email to