You were right:

The mistake was in the PETSC_DIR variable.


configuring this way



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

gives me, among other errors, the messages below.

Makefile_config_petsc:1: /home/tobias/PETSc/petsc-3.6.2/conf/variables: No such 
file or directory
make: *** No rule to make target 
`/home/tobias/PETSc/petsc-3.6.2/conf/variables'.  Stop.
Makefile_config_petsc:1: /home/tobias/PETSc/petsc-3.6.2/conf/variables: No such 
file or directory
make: *** No rule to make target 
`/home/tobias/PETSc/petsc-3.6.2/conf/variables'.  Stop.

This is why I ended up using the incorrect PESC_DIR value:

"/conf/variables"  is located in "/home/tobias/PETSc/petsc-3.6.2/lib/petsc/"


I also attached the new  config.log file

Again, thank you for all your help

________________________________
Von: John Peterson <jwpeter...@gmail.com>
Gesendet: Mittwoch, 4. November 2015 16:30
An: tschroeder
Cc: libmesh-users
Betreff: Re: [Libmesh-users] Neumann boundary problem for Laplace equation



On Wed, Nov 4, 2015 at 2:04 PM, tschroeder 
<tschroe...@daad-alumni.de<mailto:tschroe...@daad-alumni.de>> wrote:

Here it is.

OK, so one issue is that your PETSc was not found properly:

configure:29442: checking 
/home/tobias/PETSc/petsc-3.6.2/lib/petsc/include/petscversion.h usability
configure:29442: mpicc -c   conftest.c >&5
conftest.c:143:75: fatal error: 
/home/tobias/PETSc/petsc-3.6.2/lib/petsc/include/petscversion.h: No such file 
or directory
 #include </home/tobias/PETSc/petsc-3.6.2/lib/petsc/include/petscversion.h>
                                                                           ^

it looks like your PETSC_DIR variable may not be set quite right.

This doesn't really explain the segfault you are seeing, but ensuring libmesh 
finds PETSc -- and uses the same MPI that PETSc is using -- may end up fixing 
the problem.

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

Reply via email to