OCamlMPI Problems

OCamlMPI Problems

[Eric Schulte]

Hi,

I've been trying to install ocamlmpi-1.01 with mpich2-1.1.1p1 on a Mac.

I've followed the helpful instructions found in previous email threads
on this list (specifically here [1]).  However after successfully
installing mpich2 and making the changes to my Makefile, although I am
able to compile the testmpi executable with the following [2],
attempting to run the resulting executable segfaults [3].

Does anyone have any suggestions, or an idea of if this problem is most
likely in my configuration of OCamlMPI, or possibly in my mpich2
instillation?

Thanks -- Eric

Footnotes: 
[1]  http://caml.inria.fr/pub/ml-archives/caml-list/2007/10/3e20fe358a148af4095eba48e32c338e.en.html

[2]  
( ~/src/ocamlmpi-1.01 ) ocamlc -cc mpicc -o testmpi unix.cma mpi.cma test.ml -ccopt -L
File "test.ml", line 144, characters 6-9:
Warning Y: unused variable res.

[3]  
( ~/src/ocamlmpi-1.01 ) mpiexec -n 1 testmpi 
[eschulte:02147] *** Process received signal ***
[eschulte:02147] Signal: Segmentation fault (11)
[eschulte:02147] Signal code: Address not mapped (1)
[eschulte:02147] Failing at address: 0x44000088
[ 1] [0xbfffef68, 0x44000088] (-P-)
[ 2] (caml_mpi_init + 0x91) [0xbfffef98, 0x00001b6e] 
[ 3] (caml_interprete + 0x1861) [0xbffff168, 0x000097c8] 
[ 4] (caml_main + 0x3ac) [0xbffff1d8, 0x0000b4ee] 
[ 5] (main + 0x11) [0xbffff1f8, 0x0000b6b5] 
[ 6] (start + 0x36) [0xbffff214, 0x00001a32] 
[ 7] [0x00000000, 0x00000001] (FP-)
[eschulte:02147] *** End of error message ***
mpiexec noticed that job rank 0 with PID 2147 on node eschulte.local exited on signal 11 (Segmentation fault). 

Re: OCamlMPI Problems

[Eric Schulte]

To update,

After replacing mpich2-1.1.1p1 with mpich-1.2.7p1 (which built with no
errors, and passes all of it's own tests) then re-building ocaml-mpi,
I've successfully generated a new slightly less intimidating but still
intractable error [1].  Any suggestions are greatly appreciated.

Thanks -- Eric

Footnotes: 
[1]  
( ~/src/ocamlmpi-1.01 ) mpirun testmpi 
*** An error occurred in MPI_Keyval_create
*** before MPI was initialized
*** MPI_ERRORS_ARE_FATAL (goodbye)
[eschulte.local:96689] Abort before MPI_INIT completed successfully; not able to guarantee that all other processes were killed!

"Eric Schulte" <schulte.eric@gmail.com> writes:

> Hi,
>
> I've been trying to install ocamlmpi-1.01 with mpich2-1.1.1p1 on a Mac.
>
> I've followed the helpful instructions found in previous email threads
> on this list (specifically here [1]).  However after successfully
> installing mpich2 and making the changes to my Makefile, although I am
> able to compile the testmpi executable with the following [2],
> attempting to run the resulting executable segfaults [3].
>
> Does anyone have any suggestions, or an idea of if this problem is most
> likely in my configuration of OCamlMPI, or possibly in my mpich2
> instillation?
>
> Thanks -- Eric
>
> Footnotes: 
> [1]  http://caml.inria.fr/pub/ml-archives/caml-list/2007/10/3e20fe358a148af4095eba48e32c338e.en.html
>
> [2]  
> ( ~/src/ocamlmpi-1.01 ) ocamlc -cc mpicc -o testmpi unix.cma mpi.cma test.ml -ccopt -L
> File "test.ml", line 144, characters 6-9:
> Warning Y: unused variable res.
>
> [3]  
> ( ~/src/ocamlmpi-1.01 ) mpiexec -n 1 testmpi 
> [eschulte:02147] *** Process received signal ***
> [eschulte:02147] Signal: Segmentation fault (11)
> [eschulte:02147] Signal code: Address not mapped (1)
> [eschulte:02147] Failing at address: 0x44000088
> [ 1] [0xbfffef68, 0x44000088] (-P-)
> [ 2] (caml_mpi_init + 0x91) [0xbfffef98, 0x00001b6e] 
> [ 3] (caml_interprete + 0x1861) [0xbffff168, 0x000097c8] 
> [ 4] (caml_main + 0x3ac) [0xbffff1d8, 0x0000b4ee] 
> [ 5] (main + 0x11) [0xbffff1f8, 0x0000b6b5] 
> [ 6] (start + 0x36) [0xbffff214, 0x00001a32] 
> [ 7] [0x00000000, 0x00000001] (FP-)
> [eschulte:02147] *** End of error message ***
> mpiexec noticed that job rank 0 with PID 2147 on node eschulte.local exited on signal 11 (Segmentation fault).