[Condor-users] mpich2 error " '.../condor_exec.exe' with arguments hellow.exe: No such file or directory"

["arash" <anoorghorbani@xxxxxxxxx>]

Dear All,

 

I was configured to quad-core computers (called mpi0 and mpi1) as dedicated resources , which mpi0 are set as scheduler. however I can run simple parallel jobs, but I couldn’t run mpi jobs.

 

And I received the error :

 

'/home/condor/execute/dir_6618/condor_exec.exe' with arguments hellow.exe: No such file or directory

 

In the file        log.#pArAlLeLnOdE#

 

 

I submitted the following file:

#########################################

universe              = parallel

executable         = mp2script.smp

arguments          = hellow.exe

machine_count                = 3

should_transfer_files    = yes

when_to_transfer_output          = on_exit

transfer_input_files       = hellow.exe

+WantParallelSchedulingGroups               = False

notification         =never

log          =log.$(NODE)

error      =err.$(NODE)

output  =out.$(NODE)

queue

#########################################

 

Which hellow.exe is mpicc of

 

*****************************************

/* -*- Mode: C; c-basic-offset:4 ; -*- */

/*

 *  (C) 2001 by Argonne National Laboratory.

 *      See COPYRIGHT in top-level directory.

 */

 

#include <stdio.h>

#include "mpi.h"

 

int main( int argc, char *argv[] )

{

    int rank;

    int size;

   

    MPI_Init( 0, 0 );

    MPI_Comm_rank(MPI_COMM_WORLD, &rank);

    MPI_Comm_size(MPI_COMM_WORLD, &size);

    printf( "Hello world from process %d of %d\n", rank, size );

    MPI_Finalize();

    return 0;

}

 

 

 

And I used mp2script described in CamGrid page  which is:

 

#!/bin/sh

#

# File: mp2script.smp

#

#   Edit MPDIR and LD_LIBRARY_PATH to suit your

# local configuration.

 

_CONDOR_PROCNO=$_CONDOR_PROCNO

_CONDOR_NPROCS=$_CONDOR_NPROCS

 

EXECUTABLE=$1

shift

 

# the binary is copied but the executable flag is cleared.

# so the script have to take care of this

chmod +x $EXECUTABLE

 

# Set this to the bin directory of your mpich2 installation

MPDIR=/usr/local/mpich2

PATH=$MPDIR/bin:.:$PATH

export PATH

 

# When a job is killed by the user, this script will get sigterm

# This script has to catch it and do the cleaning for the

# mpich2 environment

finalize()

{

  mpdallexit

  exit

}

trap finalize TERM

 

# start the mpich2 environment

if [ $_CONDOR_PROCNO -eq 0 ]

then

                # MPICH2 requires an mpd.conf file with a

        # password in it on the host starting the job.

        # We'll generate one on the fly, though we could

                # use a pre-prepared one, e.g:

                # export MPD_CONF_FILE=~/.mpd.conf

 

        export MPD_CONF_FILE=`pwd`/mpd.conf

        echo "secretword=MySecretWord" > $MPD_CONF_FILE

        chmod 600 $MPD_CONF_FILE

 

                # Adjust the following to your needs. I use Intel

                # compilers to build MPICH2

        export LD_LIBRARY_PATH=/lib:/usr/lib:/$MPDIR/lib

 

                mpd --daemon --debug

        val=$?

 

                if [ $val -ne 0 ]

                then

                                echo "mp2script error booting mpd: $val"

                                exit 1

                fi

 

                ## Run the actual mpi job. Note pre-prepared machine file.

                mpiexec -l -machinefile $MPDIR/etc/machfile -envall -n $_CONDOR_NPROCS $EXECUTABLE $@

        mpdallexit

                rm $MPD_CONF_FILE

else

                wait

                exit 0    

fi

 

exit $?

 

###### End of mp2script.smp ######

 

 

 

And the file    log.#pArAlLeLnOdE#  was generated as followed:

 

000 (030.000.000) 02/02 17:28:00 Job submitted from host: <x.x.x.27:54299>

...

014 (030.000.000) 02/02 17:33:04 Node 0 executing on host: <x.x.x.27:39023>

...

014 (030.000.001) 02/02 17:33:04 Node 1 executing on host: <x.x.x.27:39023>

...

014 (030.000.002) 02/02 17:33:04 Node 2 executing on host: <x.x.x.27:39023>

...

001 (030.000.000) 02/02 17:33:04 Job executing on host: MPI_job

...

007 (030.000.000) 02/02 17:33:04 Shadow exception!

                Error from starter on vm3@xxxxxxxxxx: Failed to execute '/home/condor/execute/dir_6618/condor_exec.exe' with arguments hellow.exe: No such file or directory

                0  -  Run Bytes Sent By Job

                1621935  -  Run Bytes Received By Job

...

012 (030.000.000) 02/02 17:33:04 Job was held.

                Error from starter on vm3@xxxxxxxxxx: Failed to execute '/home/condor/execute/dir_6618/condor_exec.exe' with arguments hellow.exe: No such file or directory

                Code 6 Subcode 2

...

 

 

 

I will be pleasured if you have any hint.

 

Regard,

Arash