I'm using the following script to run a single processor fluent job on a linux cluster machine using Sun Grid Engine:

#!/bin/bash #$ -S /bin/bash #$ -cwd export PATH=$PATH:/usr/global/Fluent.Inc/bin:$HOME export FLUENT_ARCH=lnx86_glibc22 export LM_LICENSE_FILE=7241@egr8-w2k export FLUENT_INC=/usr/global/Fluent.Inc /usr/global/Fluent.Inc/bin/fluent 3ddp -g -i run1.jou >& run1.out

Batch jobs work fine on the main processor (linn1). However, when I use the above script to send to another processor on the cluster via Sun Grid Engine I get a directory path error. It appears that the other units of the cluster can't find the software. What should I do?

Thanks for any and all help. Worth

Hi, the command to run fluent on a cluster should be like: fluent 3ddp -g -pnet -i run1.jou >&run1.out& where -pnet is linked to the RSHD communications software. If you are using MPICH then you should use -pnmpi If you are using MPI/Pro then you should use -pvmpi

[joerik]

i actually did try to run the same simulation i had ran on a windows environment on the clusters with the following command line

fluent 3ddp -t8 -cnf=/home/name/node -g -i pipe_sf.jou > result.out

but got a lot of error messages.
The /home/name/node is the file path name on my computer and the pipe_sf.jou has the script to read the case and data file with a number of iteration.

My initial problem is that the case file did not even read to the end