There's almost certainly something wrong with your MPI installation.

How did you install it?  Can you send me the config.log file?

-- 
John



---------- Forwarded message ----------
From: tschroeder <tschroe...@daad-alumni.de>
Date: Wed, Nov 4, 2015 at 1:48 PM
Subject: AW: [Libmesh-users] Neumann boundary problem for Laplace equation
To: John Peterson <jwpeter...@gmail.com>


John,


Sorry to bug you again.


I installed PETSc and MPI. Then , I configured libMesh as follows:


 ./configure PETSC_DIR=/home/tobias/PETSc/petsc-3.6.2/lib/petsc
PETSC_ARCH=arch-linux2-c-debug --enable-everything --enable-petsc
--enable-mpi


Then, I ran into an issue when executing "make check."  This is the error
message that I get (when running example introduction_ex1):



...

==================
All 5 tests passed
==================
make[3]: Leaving directory
`/home/tobias/libMesh/libmesh-0.9.4/contrib/nanoflann'
make[2]: Leaving directory
`/home/tobias/libMesh/libmesh-0.9.4/contrib/nanoflann'
make[2]: Entering directory `/home/tobias/libMesh/libmesh-0.9.4/contrib'
make[2]: Nothing to be done for `check-am'.
make[2]: Leaving directory `/home/tobias/libMesh/libmesh-0.9.4/contrib'
make[1]: Leaving directory `/home/tobias/libMesh/libmesh-0.9.4/contrib'
Making check in examples
make[1]: Entering directory `/home/tobias/libMesh/libmesh-0.9.4/examples'
Making check in introduction/introduction_ex1
make[2]: Entering directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
make  example-dbg example-devel example-opt   run.sh
make[3]: Entering directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
  CXX      example_dbg-introduction_ex1.o
  CXXLD    example-dbg
/usr/bin/ld: skipping incompatible /lib/libz.so when searching for -lz
/usr/bin/ld: skipping incompatible /usr/lib/libz.so when searching for -lz
  CXX      example_devel-introduction_ex1.o
  CXXLD    example-devel
/usr/bin/ld: skipping incompatible /lib/libz.so when searching for -lz
/usr/bin/ld: skipping incompatible /usr/lib/libz.so when searching for -lz
  CXX      example_opt-introduction_ex1.o
  CXXLD    example-opt
/usr/bin/ld: skipping incompatible /lib/libz.so when searching for -lz
/usr/bin/ld: skipping incompatible /usr/lib/libz.so when searching for -lz
make[3]: Nothing to be done for `run.sh'.
make[3]: Leaving directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
make  check-TESTS
make[3]: Entering directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
***************************************************************
* Running Example introduction_ex1:
*   example-dbg -d 3
/home/tobias/libMesh/libmesh-0.9.4/reference_elements/3D/one_hex27.xda
***************************************************************

[usbosnavdev-ludwig:09698] *** Process received signal ***
[usbosnavdev-ludwig:09698] Signal: Segmentation fault (11)
[usbosnavdev-ludwig:09698] Signal code: Address not mapped (1)
[usbosnavdev-ludwig:09698] Failing at address: 0x44000098
[usbosnavdev-ludwig:09698] [ 0]
/lib64/libpthread.so.0(+0xf130)[0x7f4ee8983130]
[usbosnavdev-ludwig:09698] [ 1]
/usr/local/lib/libmpi.so.12(MPI_Comm_size+0x46)[0x7f4eeb21a576]
[usbosnavdev-ludwig:09698] [ 2]
/home/tobias/libMesh/libmesh-0.9.4/.libs/libmesh_dbg.so.0(_ZN7libMesh8Parallel12Communicator6assignERKi+0x44)[0x7f4eecdc28b8]
[usbosnavdev-ludwig:09698] [ 3]
/home/tobias/libMesh/libmesh-0.9.4/.libs/libmesh_dbg.so.0(_ZN7libMesh8Parallel12CommunicatoraSERKi+0x2f)[0x7f4eecdc286d]
[usbosnavdev-ludwig:09698] [ 4]
/home/tobias/libMesh/libmesh-0.9.4/.libs/libmesh_dbg.so.0(_ZN7libMesh11LibMeshInitC2EiPKPKci+0x4f4)[0x7f4eecdc05ae]
[usbosnavdev-ludwig:09698] [ 5]
/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1/.libs/example-dbg[0x40f447]
[usbosnavdev-ludwig:09698] [ 6]
/lib64/libc.so.6(__libc_start_main+0xf5)[0x7f4ee85d4af5]
[usbosnavdev-ludwig:09698] [ 7]
/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1/.libs/example-dbg[0x40f249]
[usbosnavdev-ludwig:09698] *** End of error message ***
/home/tobias/libMesh/libmesh-0.9.4/examples/run_common.sh: line 50:  9698
Segmentation fault      $LIBMESH_RUN ./$executable $options $LIBMESH_OPTIONS
FAIL: run.sh
====================================================
1 of 1 test failed
Please report to libmesh-users@lists.sourceforge.net
====================================================
make[3]: *** [check-TESTS] Error 1
make[3]: Leaving directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
make[2]: *** [check-am] Error 2
make[2]: Leaving directory
`/home/tobias/libMesh/libmesh-0.9.4/examples/introduction/introduction_ex1'
make[1]: *** [check-recursive] Error 1
make[1]: Leaving directory `/home/tobias/libMesh/libmesh-0.9.4/examples'
make: *** [check-recursive] Error 1
tobias@usbosnavdev-ludwig ~/libMesh/libmesh-0.9.4 $



This is how I configure libMesh:


./configure PETSC_ARCH=arch-linux2-c-debug
PETSC_DIR=/home/tobias/PETSc/petsc-3.6.2 --enable-everything --enable-petsc
--enable-mpi





------------------------------
*Von:* tschroeder
*Gesendet:* Dienstag, 3. November 2015 16:59
*An:* John Peterson
*Betreff:* Re: [Libmesh-users] Neumann boundary problem for Laplace equation

Perfect,

Thank you



Am 03.11.2015 um 16:41 schrieb John Peterson <jwpeter...@gmail.com>:



On Tue, Nov 3, 2015 at 2:23 PM, tschroeder <tschroe...@daad-alumni.de>
wrote:

> This is what happens:
>
>
>   CXX      src/geom/libmesh_dbg_la-face_quad9.lo
>   CXX      src/geom/libmesh_dbg_la-face_tri.lo
> src/geom/elem_cutter.C: In constructor 'libMesh::ElemCutter::ElemCutter()':
> src/geom/elem_cutter.C:43:49: error: 'MPI_COMM_SELF' was not declared in
> this scope
>    _comm_self.reset (new Parallel::Communicator (MPI_COMM_SELF));
>                                                  ^
> make[1]: *** [src/geom/libmesh_dbg_la-elem_cutter.lo] Error 1
> make[1]: *** Waiting for unfinished jobs....
> make[1]: Leaving directory `/home/tobias/libMesh/libmesh-0.9.4'
> make: *** [all-recursive] Error 1
>

Ah, looks like an issue with the --disable-mpi code path, which as you can
imagine doesn't get tested very often.

Installing MPI/PETSc will "fix" the problem, but I'll open an issue to make
sure the library builds in this configuration before the next release.

-- 
John
------------------------------------------------------------------------------
_______________________________________________
Libmesh-users mailing list
Libmesh-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to