[zfaraday]

As the title suggests, I can't run parallel cases although the last week I could do it without any problem... :S

This is the error message provided:

--------------------------------------------------------------------------
Open RTE was unable to open the hostfile:
    /usr/lib64/mpi/gcc/openmpi/etc/openmpi-default-hostfile
Check to make sure the path and filename are correct.
--------------------------------------------------------------------------
[linux-cfd:05355] [[64188,0],0] ORTE_ERROR_LOG: Not found in file base/ras_base_allocate.c at line 200
[linux-cfd:05355] [[64188,0],0] ORTE_ERROR_LOG: Not found in file base/plm_base_launch_support.c at line 99
[linux-cfd:05355] [[64188,0],0] ORTE_ERROR_LOG: Not found in file plm_rsh_module.c at line 1167

I checked the path of the filename and, as expected it doesn't exist. Actually, etc/ directory within /usr/lib64/mpi/gcc/openmpi/ doesn't even exist, so I don't know where to find the openmpi-default-hostfile file. The point is that I run my last parallel case the last week and everything was fine! The only thing I did these days, that I remember, that may have broken something is a git pull and recompilation, but I did it before without a problem, so I don't know what to do now...

I will appreciate any advice.

Regards,

Alex

[wyldckat]

Greetings Alex,

I know I saw a solution for this just a few days ago... let's see if I'm lucky and I can easily find it again...
OK, one I saw back then was this: http://www.openfoam.org/archive/2.2.2/download/suse.php

Note: There is a problem with mpirun on OpenSUSE 12.3 (https://svn.open-mpi.org/trac/ompi/ticket/3554) where a default file is not installed correctly. Where this problem exists, it is possible to run mpirun using the following flag:

mpirun --default-hostfile none

The other solution is here: https://forums.opensuse.org/showthre...76#post2604776 - namely, to run (as root) the following commands:

cd /usr/lib64/mpi/gcc/openmpi/
mkdir etc
cd etc
ln -s /etc/openmpi-default-hostfile openmpi-default-hostfile

Best regards,
Bruno

[zfaraday]

Dear Bruno,

Many thanks for your quick response. After having opened this thread I kept searching in the net for a solution and I got to this page. Where a method it is explained. Firstly, Alberto Passalacqua says:

usually this error happens when MPI wasn’t compiled properly or it is not set up properly. You may check if the MPI setup for your user is correct with the command mpi-selector-menu, and see if both your user and the system have the same version of openmpi selected. If not, mpi-selector-menu lets you change this setting. Note that you must logout and login again to see the effect of any changes you make.

Other question: did you re-compile OpenFOAM on this installation of GeekoCFD? Knowing this would help me to debug.

I run on the terminal the command mpi-selector-menu as suggested and found out that, obviously, something wasn't going well...

Current system default: openmpi-1.6
Current user default:   <none>

    "u" and "s" modifiers can be added to numeric and "U"
    commands to specify "user" or "system-wide".

1. openmpi-1.6
U. Unset default
Q. Quit

I changed the Current user default to openmpi-1.6 but it kept crashing...

I kept reading the conversation of Alberto till the end till I found this:

Try to create this symbolic link. It used to be required:

ln -s /etc /usr/lib64/mpi/gcc/openmpi/

Which I guess it does exactly the same that the second option you gave me:

Originally Posted by wyldckat

cd /usr/lib64/mpi/gcc/openmpi/
mkdir etc
cd etc
ln -s /etc/openmpi-default-hostfile openmpi-default-hostfile

Am I right? I did what Alberto says in his page because I saw it first...

Well, the problem is that now it keeps crashing but the error shown the following:

--> FOAM FATAL ERROR: 
Trying to use the dummy Pstream library.
This dummy library cannot be used in parallel mode

    From function 

--> FOAM FATAL ERROR: 
Trying to use the dummy Pstream library.
This dummy library cannot be used in parallel mode

    From function UPstream::init(int& argc, char**& argv)
    in file UPstream.C at line 37.

FOAM exiting



--> FOAM FATAL ERROR: 
Trying to use the dummy Pstream library.
This dummy library cannot be used in parallel mode

    From function UPstream::init(int& argc, char**& argv)
    in file UPstream.C at line 37.

FOAM exiting


UPstream::init(int& argc, char**& argv)
    in file UPstream.C at line 37.

FOAM exiting


--> FOAM FATAL ERROR: 
Trying to use the dummy Pstream library.
This dummy library cannot be used in parallel mode

    From function UPstream::init(int& argc, char**& argv)
    in file UPstream.C at line 37.

FOAM exiting

--------------------------------------------------------------------------
mpirun noticed that the job aborted, but has no info as to the process
that caused that situation.
--------------------------------------------------------------------------

Am I doing anything wrong?

Many thanks for your insight Bruno.

[zfaraday]

Ok, finally solved!

I just recompiled Pstream within $FOAM_SRC as it's said here in the 4th point and now I can run in parallel again!

I wonder if it happened because of the update I did a couple of days ago, because the last week I had no problem in running parallel cases...

Thanks for your tip Bruno (althoug I just almost-solved a few minutes ago )

Regards,

Alex