[Fauster]

Dear Foam users,


I am kind of lost with OpenFOAM and clusters . This is mainly due to my poor knowledge of cluster behavior. I have read some previous posts and they not provide me the solutions to my questions.


OpenFOAM Cluster Setup for Beginners
openfoam on 2 machines with passwordless ssh
https://www.cfd-online.com/Forums/bl...-parallel.html

1. Assuming that I would like to install OpenFOAM on a cluster composed with 4 nodes what would be the best OS ? I would like also to install also ANSYS so I guess a red hat based distribution would be the best one...I have read that Rocks (http://www.rocksclusters.org/) is well suited for science HPC. But I am not able to find out if it will be problematic to install OpenFOAM or ANSYS on this distribution.
2. I have understood that some company advice to use Bright for cluster administration (or other non free tools). What does exactly do Bright ? Is Rocks doing the same things as Bright ? Should I go Open source for the cluster administration ?
3. Is there any guides line or tutorials for OpenFOAM cluster installation ? I guess I need to install it on each nodes...Is there any difference on the installation for the front node ?
4. I would like to run OpenFOAM from my desktop computer. How I can send the job to the cluster ? I guess I need to connect to the cluster (maybe using ip adress) from a terminal and then type something like :

   mpirun -np 168 –hostfile ~/hostfile.14 interFoam -paralle

   (http://www.hpcadvisorycouncil.com/pd..._Practices.pdf). Is this command still ok ?
5. Usually the case will be located on my desktop computer. Do I need to copy it to all the cluster nodes before running the case on parallel ? Where OpenFOAM will write the results files ? On the cluster or my desktop computer ? (this for disk space)
6. If I have 64=4(node numbers)*16 processor directories on my cluster how I can postprocess the results ? If the results size is very large I guess I can't reconstruct it to my desktop computer. Can I use paraview to visualize the results located on my cluster directly to my desktop computer screen ?

I know it's a lot of questions but it would help me a lot ! If someone using OpenFOAM on a cluster can help me to understand the whole process it will be fantastic !


Thanks