[unobtruse]

Hi There

I'm trying to install OpenFOAM 2.3 on a 5 node cluster running Centos 6.4. I've followed the guide at https://openfoamwiki.net/index.php/I...CentOS_SL_RHEL based on Centos 6.5, but I've adjusted my $HOME to point to /opt. The reason for doing this is that the /opt directory is a network shared directory --- the plan is to install OpenFOAM once on a single node and then to simply run it from the others. I have however run into errors that I can not resolve. See attached make.log file.

Any help would be appreciated.

[elvis]

Hi,

I did just have a short look but
MPI seems to be your problem,

does other software with MPI work properly

[unobtruse]

Hi Elvis

I managed to get OpenFOAM running properly in the NFS folder by first installing it in my home directory and then moving the OpenFOAM folder to the desired place. I followed wyldckat's instructions from #5 found here http://www.cfd-online.com/Forums/ope...tml#post361663.

On an unrelated note: when I ran my simulation I kept running into this error:

Setting up information for layer truncation ...
new cannot satisfy memory request.
This does not necessarily mean you have run out of virtual memory.
It could be due to a stack violation caused by e.g. bad use of pointers or an out of date shared library
new cannot satisfy memory request.

.
I think this may have been caused by the NFS folder only being on a 50GB partition... :P Unfortunately I can't run my simulation again. See log file for details.

bmw_snappy.log.tar.gz

[wyldckat]

Greetings to all!

@Colin: I'm sorry to say that the error message was due to a rookie mistake:

/*---------------------------------------------------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  2.3.0                                 |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
Build  : 2.3.0-f5222ca19ce6
Exec   : snappyHexMesh -overwrite

You forgot to add the "-parallel" option when calling snappyHexMesh along with mpirun.

The result was that you had X processes, all trying to create the same mesh, requiring X*N memory in total, instead of just N for the whole parallel mesh generation.

Best regards,
Bruno