Welcome, Guest
Username: Password: Remember me

TOPIC: Error while loading shared libraries

Error while loading shared libraries 10 years 1 week ago #14814

  • amyrhulbakar
  • amyrhulbakar's Avatar
  • OFFLINE
  • Junior Boarder
  • Posts: 46
  • Thank you received: 4
Hello all,

I need to post this new topic as I can't find the similar discussion. Please direct me to the existing post if this had been discussed.

I have used the configuration of parallel mpi hpc before, but the error of loading the shared libraries was suddenly occurred.
out_t2d_mersey: error while loading shared libraries: libifport.so.5: cannot open shared object file: No such file or directory
out_t2d_mersey: error while loading shared libraries: libifport.so.5: cannot open shared object file: No such file or directory
mpiexec noticed that the job aborted, but has no info as to the process
that caused that situation.

The lines in configuration which related to mpiexec and/or libraries are:
mpi_cmdexec: /opt/mpi/bullxmpi/1.1.17.1/bin/mpiexec -wdir <wdir> -np <ncsize> <exename>
incs_parallel: -I /opt/mpi/bullxmpi/1.1.17.1/include/
libs_all: /opt/mpi/bullxmpi/1.1.17.1/lib/libmpi.so

Was the libifport.so.5 related to the libmpi.so at libs_all?

Please help..
The administrator has disabled public write access.

Error while loading shared libraries 10 years 1 week ago #14817

  • amyrhulbakar
  • amyrhulbakar's Avatar
  • OFFLINE
  • Junior Boarder
  • Posts: 46
  • Thank you received: 4
When tried to change the mpi from bullxmpi to openmpi, the error was still occurred but with different messages from the previous.
[arccacluster102:115354] mca: base: component_find: unable to open /app/libraries/openmpi/1.5.4/gnu-4.6.2/lib/openmpi/mca_ess_lsf: libbat.so: cannot open shared object file: No such file or directory (ignored)
[arccacluster102:115354] mca: base: component_find: unable to open /app/libraries/openmpi/1.5.4/gnu-4.6.2/lib/openmpi/mca_plm_lsf: libbat.so: cannot open shared object file: No such file or directory (ignored)
[arccacluster102:115354] mca: base: component_find: unable to open /app/libraries/openmpi/1.5.4/gnu-4.6.2/lib/openmpi/mca_ras_lsf: libbat.so: cannot open shared object file: No such file or directory (ignored)
[arccacluster102:115355] mca: base: component_find: unable to open /app/libraries/openmpi/1.5.4/gnu-4.6.2/lib/openmpi/mca_mtl_psm: libpsm_infinipath.so.1: cannot open shared object file: No such file or directory (ignored)
[arccacluster102:115356] mca: base: component_find: unable to open /app/libraries/openmpi/1.5.4/gnu-4.6.2/lib/openmpi/mca_mtl_psm: libpsm_infinipath.so.1: cannot open shared object file: No such file or directory (ignored)

The openmpi in configuration:
mpi_cmdexec: /software/libraries/openmpi/1.5.4/gnu-4.6.2/bin/mpiexec -wdir <wdir> -np <ncsize> <exename>
incs_parallel: -I /software/libraries/openmpi/1.5.4/gnu-4.6.2/include/
libs_all: /software/libraries/openmpi/1.5.4/gnu-4.6.2/lib/libmpi.so

Please help..
The administrator has disabled public write access.

Error while loading shared libraries 10 years 5 days ago #14838

  • yugi
  • yugi's Avatar
  • OFFLINE
  • openTELEMAC Guru
  • Posts: 851
  • Thank you received: 244
Could you post your while systel.cfg ?
You could try using mpf90 in the cmd_obj and cmd_exe in your systel.cfg.
And do a compileTELEMAC.py --clean

The error seems to be that the compiled file where not compiled with the same compiler as the one used by mpi.

Hope it helps.
There are 10 types of people in the world: those who understand binary, and those who don't.
The administrator has disabled public write access.

Error while loading shared libraries 10 years 3 days ago #14870

  • amyrhulbakar
  • amyrhulbakar's Avatar
  • OFFLINE
  • Junior Boarder
  • Posts: 46
  • Thank you received: 4
Hi Yugi,

Thanks for your comment. I used ifort for both cmd_obj and cmd_exe, while using bullxmpi for mpi or so. It's now working when I recompiled it again using mpif90 for both, thanks. How could I know that I am using the same compiler for both the compiled files and mpi, let say while using openmpi? Below is my corrected systel.cfg.

# _____                              _______________________________
# ____/ TELEMAC Project Definitions /______________________________/
[Configurations]
configs: bullxmpihpc
#
[general]
root:       $HOME/opentelemac/v6p3r2
version:    v6p3
language:   2
modules:    clean system
options:
#
sfx_zip:    .gztar
sfx_lib:    .a
sfx_obj:    .o
sfx_mod:    .mod
sfx_exe:
#
val_root:   <root>/examples
val_rank:   all
# also possible val_rank:   <3 >7 6
#
# _____               ______________________________________________
# ____/ Bullxmpi hpc /_____________________________________________/
[bullxmpihpc]
#
options: parallel mpi hpc
#
par_cmdexec: <config>/partel < PARTEL.PAR >> <partel.log>
#
mpi_cmdexec: /opt/mpi/bullxmpi/1.1.17.1/bin/mpiexec -wdir <wdir> -np <ncsize> <exename>
#
hpc_stdin: #!/bin/bash
       #PBS -l select=<ncnode>:ncpus=<nctile>:mpiprocs=<nctile>:ompthreads=1
       #PBS -l place=scatter:excl
       #PBS -l walltime=<walltime>
       #PBS -o t2d.hpc.log
       #PBS -e t2d.hpc.err
       #PBS -N <jobname>
       #PBS -q <queue>
       module purge
       module load intel bullxmpi pyenv python/2.7.4
       ulimit -s unlimited
       ulimit -c 0
       cd $PBS_O_WORKDIR
       <mpi_cmdexec>
       exit $?
#
hpc_cmdexec: chmod 755 <hpc_stdin>; qsub <hpc_stdin>
#
cmd_obj: mpif90 -c -O3 -DHAVE_MPI -convert big_endian -cpp <mods> <incs> <f95name>
cmd_lib: ar cru <libname> <objs>
cmd_exe: mpif90 -convert big_endian -lpthread -lm -o <exename> <objs> <libs>
#
mods_all: -I <config>
#
incs_parallel: -I /opt/mpi/bullxmpi/1.1.17.1/include/
libs_partel: $HOME/opentelemac/v6p3r2/optionals/metis-5.1.0/build/Linux-x86_64/libmetis/libmetis.a
libs_all: /opt/mpi/bullxmpi/1.1.17.1/lib/libmpi.so.1.0.1
#
The administrator has disabled public write access.

Error while loading shared libraries 10 years 3 days ago #14871

  • yugi
  • yugi's Avatar
  • OFFLINE
  • openTELEMAC Guru
  • Posts: 851
  • Thank you received: 244
mpif90 -show
will show you what mpif90 actually stands for.
There are 10 types of people in the world: those who understand binary, and those who don't.
The administrator has disabled public write access.
The following user(s) said Thank You: amyrhulbakar
Moderators: borisb

The open TELEMAC-MASCARET template for Joomla!2.5, the HTML 4 version.