[petsc-users] 3.5.1 fail with mumps

Mon Aug 4 06:47:39 CDT 2014

Can you send make.log and configure.log?

  Thanks,

     Matt

On Sun, Aug 3, 2014 at 10:40 PM, Pascal Paschos <
pascal.paschos at northwestern.edu> wrote:

>  Hi,
>
>  I've already seen a thread where someone recently was trying to build
> with mumps and was failing at test.
> The recommendation that seemed to have worked for that user, at least on
> this point, was
> a fresh install and shared libraries off. Well, I am having the same
> problem with 3.5.1 with the same error message he was getting.
> The build can not interpret symbols in some mumps and parmetis references.
> Whether shared libraries is set to 0 or 1 is irrelevant, at least in my
> case. The problem persists.
> The error messages are in the bottom of this post.
> And yes, all packages are there, otherwise petsc would not build in the
> first place. It fails a test with an error message that is
> typical of the build missing some include files that it expects to have
> present.
> Without mumps - metis & parmetis - the install is fine and tests complete
> fine.
> So, I guess I am wondering if the developers have any more insight to
> spare...
> Thanks.
>
>  Here is my configure:
>
> ./configure --prefix=/software/petsc/3.5.1.1  \
>
> --with-blas-lib=/software/lapack/3.5.0/lib/libblas.so
> --with-lapack-lib=/software/lapack/3.5.0/liblapack.so \
>
> --with-hdf5=1 --with-hdf5-dir=/software/hdf5/1.8.10 \
>
> --with-valgrind=1 --with-valgrind-dir=/software/valgrind/3.8.1 \
>
> --with-boost=1 --with-boost-dir=/software/boost/1.41.0 \
>
> --with-cc=mpicc --with-cxx=mpicxx --with-fc=mpif90 \
>
> --with-debugging=1 \
>
> --download-fftw=yes \
>
> --download-scalapack=yes  \
>
> --download-mumps=yes \
>
> --download-parmetis=yes \
>
> --download-metis=yes \
>
> --CFLAGS="-I/usr/include -I/usr/include/openssl" \
>
> --LDFLAGS=" -L/usr/lib64" \
>
> --LDLIBS=" -lcrypto -lssl" \
>
> --with-shared-libraries=0
>
>
>  I found it necessary to pass references to libssl to the linker because
> otherwise the make is then complaining with an SSL related issue, that it
> can not find some include file.
>
>
>
>  make PETSC_DIR=/software/petsc/3.5.1.1 test
>
> Running test examples to verify correct installation
>
> Using PETSC_DIR=/software/petsc/3.5.1.1 and PETSC_ARCH=arch-linux2-c-debug
>
> ********************Error detected during compile or
> link!********************
>
> *See http://www.mcs.anl.gov/petsc/documentation/faq.html
> <http://www.mcs.anl.gov/petsc/documentation/faq.html>*
>
> */software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/snes/examples/tutorials
> ex19*
>
>
> ***********************************************************************************
>
> **********************W-a-r-n-i-n-g**************************
>
> *Your PETSC_DIR may not match the directory you are in*
>
> *PETSC_DIR  /software/petsc/3.5.1.1 <http://3.5.1.1> Current directory
> /software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/snes/examples/tutorials*
>
> *Ignore this if you are running make test             *
>
> ********************************************************
>
> mpicc -o ex19.o -c -I/usr/include -I/usr/include/openssl -g3 -O0
> -I/software/petsc/3.5.1.1/include -I/software/petsc/3.5.1.1/include
> -I/software/valgrind/3.8.1/include -I/software/hdf5/1.8.10/include
> -I/software/boost/1.41.0/include
> -I/software/mpi/openmpi-1.6.3-gcc-4.6.3-trq4/include    `pwd`/ex19.c
>
> mpicc -I/usr/include -I/usr/include/openssl -g3 -O0  -o ex19  ex19.o
> -L/software/petsc/3.5.1.1/lib  -lpetsc -Wl,-rpath,/software/petsc/
> 3.5.1.1/lib -lcmumps -ldmumps -lsmumps -lzmumps -lmumps_common -lpord
> -lscalapack -Wl,-rpath,/software/lapack/3.5.0 -L/software/lapack/3.5.0
> -llapack -Wl,-rpath,/software/lapack/3.5.0/lib -L/software/lapack/3.5.0/lib
> -lblas -lparmetis -lmetis -lX11 -lssl -lcrypto -lpthread -lfftw3_mpi
> -lfftw3 -Wl,-rpath,/software/hdf5/1.8.10/lib -L/software/hdf5/1.8.10/lib
> -lhdf5_fortran -lhdf5_hl -lhdf5
> -L/software/mpi/openmpi-1.6.3-gcc-4.6.3-trq4/lib64
> -L/hpc/software/gcc/4.6.3-rhel6/lib/gcc/x86_64-unknown-linux-gnu/4.6.3
> -L/hpc/software/gcc/4.6.3-rhel6/lib/gcc
> -L/hpc/software/gcc/4.6.3-rhel6/lib64 -L/hpc/software/gcc/4.6.3-rhel6/lib
> -lmpi_f90 -lmpi_f77 -lgfortran -lm -lgfortran -lm -lgfortran -lm -lgfortran
> -lm -lquadmath -lm -lmpi_cxx -lstdc++ -ldl -lmpi -lnuma -lrt -lnsl -lutil
> -lgcc_s -lpthread -ldl
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(baij.o): In function
> `MatCreate_SeqBAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/baij/seq/baij.c:3006:
> undefined reference to `MatGetFactor_baij_mumps's'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(mpibaij.o): In function
> `MatCreate_MPIBAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/baij/mpi/mpibaij.c:3049:
> undefined reference to `MatGetFactor_baij_mumps'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(sbaij.o): In function
> `MatCreate_SeqSBAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/sbaij/seq/sbaij.c:1966:
> undefined reference to `MatGetFactor_sbaij_mumps'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(mpisbaij.o): In function
> `MatCreate_MPISBAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/sbaij/mpi/mpisbaij.c:1803:
> undefined reference to `MatGetFactor_sbaij_mumps'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(aij.o): In function
> `MatCreate_SeqAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/aij/seq/aij.c:4136:
> undefined reference to `MatGetFactor_aij_mumps'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(mpiaij.o): In function
> `MatCreate_MPIAIJ':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/impls/aij/mpi/mpiaij.c:5578:
> undefined reference to `MatGetFactor_aij_mumps'
>
> /software/petsc/3.5.1.1/lib/libpetsc.a(spartition.o): In function
> `MatPartitioningRegisterAll':
>
> /hpc/software/sources/builds/petsc-3.5.1_PP/petsc-debug/petsc-3.5.1/src/mat/partition/spartition.c:45:
> undefined reference to `MatPartitioningCreate_Parmetis'
>
>
>

-- 
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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140804/29ac5ca0/attachment-0001.html>