On Mon, 26 Jan 2015 12:57:51 +0100 Mikael Mortensen <[email protected]> wrote:
> > > On 26 Jan 2015, at 12:36, Jan Blechta <[email protected]> > > wrote: > > > > On Mon, 26 Jan 2015 12:11:18 +0100 > > Mikael Mortensen <[email protected]> wrote: > > > >> > >>> On 26 Jan 2015, at 11:53, Jan Blechta <[email protected]> > >>> wrote: > >>> > >>> On Mon, 26 Jan 2015 11:42:42 +0100 > >>> Mikael Mortensen <[email protected] > >>> <mailto:[email protected]>> wrote: > >>> > >>>> Hi, > >>>> > >>>> With recent (today) hashdist installation I run the following > >>>> test.py script > >>>> > >>>> from dolfin import * > >>>> > >>>> mesh = UnitSquareMesh(4, 4) > >>>> V = VectorFunctionSpace(mesh, 'CG', 1) > >>>> u = interpolate(Expression(('x[0]', 'x[1]')), V) > >>>> print “Hi" > >>>> > >>>> > >>>> [mikael@ubuntu tests (master)]$ python test.py > >>>> Hi > >>>> > >>>> and then it just hangs there without exiting. Same thing in > >>>> ipython. I am currently using host python, but I got the same > >>>> when I compiled with hashdist python. Anyone else experiencing > >>>> this weirdness? Any idea what is going on? I’m on Ubuntu 14.04 > >>>> and I’ve done instant-clean. > >>> > >>> To debug it, you could try attaching debugger (if it is allowed on > >>> your system) > >>> > >>> gdb python <pid> > >>> > >>> hit ^C, and print stacktrace > >>> > >>> (gdb) bt > >>> > >>> Jan > >> > >> Thanks for the tip. Not really familiar with gdb so I tried pdb > >> first > >> > >> python -m pdb test_GaussDivergence.py > >>> /home/mikael/MySoftware/fenicstools/tests/test.py(5)<module>() > >> -> from dolfin import * > >> (Pdb) continue > >> Hi > >> The program finished and will be restarted > >> > >> and then it just hangs there. Ctrl+C does nothing. > >> > >> > >> With gdb I am getting somewhere though: > >> > >> gdb python > >> Reading symbols from python...(no debugging symbols found)...done. > >> (gdb) run test.py > >> Starting > >> program: /home/mikael/.hashdist/bld/profile/6axink4nal3d/bin/python > >> test.py process 16999 is executing new > >> program: /home/mikael/.hashdist/bld/python/qlvjlzdbishm/bin/python2.7 > >> [Thread debugging using libthread_db enabled] Using host > >> libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". > >> [New Thread 0x7fffdc7e3700 (LWP 17005)] Hi ^C > >> Program received signal SIGINT, Interrupt. > >> __lll_lock_wait () > >> at ../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:135 > >> 135 ../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S: No > >> such file or directory. (gdb) bt #0 __lll_lock_wait () > >> at ../nptl/sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:135 #1 > >> 0x00007ffff7bc6657 in _L_lock_909 () > >> from /lib/x86_64-linux-gnu/libpthread.so.0 #2 0x00007ffff7bc6480 > >> in __GI___pthread_mutex_lock (mutex=0x7ffff063fa10 > >> <attribute_lock+16>) at ../nptl/pthread_mutex_lock.c:79 #3 > >> 0x00007ffff03a140c in ompi_attr_get_c () > >> from /home/mikael/.hashdist/bld/mpi/vrn4awnke2cx/lib/libmpi.so.1 #4 > >> 0x00007ffff03cba27 in MPI_Attr_get () > >> from /home/mikael/.hashdist/bld/mpi/vrn4awnke2cx/lib/libmpi.so.1 #5 > >> 0x00007ffff07b2afb in Petsc_DelComm_Outer () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #6 0x00007ffff03a1e88 in ompi_attr_delete () > >> from /home/mikael/.hashdist/bld/mpi/vrn4awnke2cx/lib/libmpi.so.1 #7 > >> 0x00007ffff03cb93c in MPI_Attr_delete () > >> from /home/mikael/.hashdist/bld/mpi/vrn4awnke2cx/lib/libmpi.so.1 #8 > >> 0x00007ffff07a2baa in PetscCommDestroy () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #9 0x00007ffff07a52bc in PetscHeaderDestroy_Private () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #10 0x00007ffff081372c in ISLocalToGlobalMappingDestroy () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #11 0x00007ffff0819d08 in PetscLayoutDestroy () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #12 0x00007ffff0871718 in VecDestroy () > >> from /home/mikael/.hashdist/bld/petsc/dfw5fwbfp6hl/lib/libpetsc.so.3.5 > >> #13 0x00007ffff22bb15c in dolfin::PETScVector::~PETScVector() () > >> from /home/mikael/.hashdist/bld/dolfin/wt5zll4ojwwy/lib/libdolfin.so.1.5 > >> #14 0x00007ffff22bb229 in dolfin::PETScVector::~PETScVector() () > >> from /home/mikael/.hashdist/bld/dolfin/wt5zll4ojwwy/lib/libdolfin.so.1.5 > >> #15 0x00007ffff221c34a in dolfin::Function::~Function() () > >> from /home/mikael/.hashdist/bld/dolfin/wt5zll4ojwwy/lib/libdolfin.so.1.5 > >> #16 0x00007ffff221c4d9 in dolfin::Function::~Function() () > >> from /home/mikael/.hashdist/bld/dolfin/wt5zll4ojwwy/lib/libdolfin.so.1.5 > >> #17 0x00007ffff308f999 in > >> std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release > >> (this=0x19bbc60) at /usr/include/c++/4.8/bits/shared_ptr_base.h:144 > >> #18 0x00007fffdf8093cc in _wrap_delete_Function () > >> from > >> /home/mikael/Software/hashstack/default/lib/python2.7/site-packages/dolfin/cpp/_function.so > >> #19 0x00007fffdf7fc15e in SwigPyObject_dealloc () > >> from > >> /home/mikael/Software/hashstack/default/lib/python2.7/site-packages/dolfin/cpp/_function.so > >> #20 0x00000000005392ff in ?? () #21 0x00000000004d914b in ?? () #22 > >> 0x00000000004fdb96 in PyDict_SetItem () #23 0x000000000055a9e1 in > >> _PyModule_Clear () #24 0x00000000004f2ad4 in PyImport_Cleanup () > >> #25 0x000000000042fa89 in Py_Finalize () #26 0x000000000046ac10 in > >> Py_Main () > > > > Can you reproduce it with PETSc example > > http://www.mcs.anl.gov/petsc/petsc-current/src/vec/is/examples/tests/ex1.c.html > > which also calls ISLocalToGlobalMappingDestroy? > > Just navigate to <petsc>/src/vec/is/examples/tests/, type make and > > execute. > > > > Jan > > > > Problem went away with openmpi 1.6.5. Btw, is the petsc source stored > somewhere with hashdist? I don't know. But it would be helpful for debugging. I don't know how optimization flags are handled by hashdist but every FEniCS developer should have a build with at least PETSc --with-debugging=1 DOLFIN -DCMAKE_BUILD_TYPE:STRING=Developer apart from production builds. Then it is useful having PETSc source in place corresponding to debugging symbols in libpetsc.so. How is this handled, Johannes? Sorry for OT but I'd like to know before switching to new hashdist/fenics-developer-tools framework. Jan > > M > > >> > >> So I’m trying to google this now. I think it’s because I’m using a > >> newer version of openmpi than in the default hashdist script. > >> Default is 1.6.5, but I have compiled with 1.8.2. 1.6.5 gives me > >> problems with h5py, which is why I upgraded it in the first place. > >> So at the moment h5py works, but not fenics:-( > >> > >> M > >> > >>> > >>>> > >>>> Mikael > >>>> > >>>> > >>>> _______________________________________________ > >>>> fenics-support mailing list > >>>> [email protected] > >>>> <mailto:[email protected]> > >>>> http://fenicsproject.org/mailman/listinfo/fenics-support > >>>> <http://fenicsproject.org/mailman/listinfo/fenics-support> > > > > _______________________________________________ > fenics-support mailing list > [email protected] > http://fenicsproject.org/mailman/listinfo/fenics-support _______________________________________________ fenics-support mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics-support
