[heliana60]

Hi Bruno!

Thank you so much, I checked and openFoam settings was trying to compile an earlier version of the open-mpi that was downloaded by ThirdParty. I fixed that and it compiled

I am now facing another little problem when I submit the queue to the cluster, it always says:

Warning: no access to tty (Bad file descriptor).
Thus no job control in this shell

This is the script I am using when I enter qsub viscofoam:

#!/bin/bash
#PBS -l walltime=100:00
#PBS -l select=1:ncpus=8:mpiprocs=8
#PBS -N th10_1e-2_sU
#PBS -q all.q
#PBS -o out # stdout and stderr
#PBS -e err
# redirect stdout and stderr
# -o log
# -e log.err

module load openmpi

mpirun -np 8 viscoelasticFluidFoam -parallel > run_par.$PBS_JOBID

Do you have any idea?

Heliana

[wyldckat]

Hi Heliana,

I moved (copied) your post from here: http://www.cfd-online.com/Forums/ope...llel-mode.html - to this new thread at the correct sub-forum, since this was no longer an issue with installation.

OK, regarding your question: this seems something that should be answered by your system's administrator.
However, I did a quick search on Google for:

qsub Warning: no access to tty

and found out that this is only a warning, meant to tell you that key combinations such as Ctrl+Z will not work, mainly because the job request was sent to the cluster and there is no directly control from the current launch. You'll have to use some other command to stop or control the job you've sent to the cluster.

Best regards,
Bruno

[heliana60]

Hi Bruno!

My administrator already fix it apparently the cluster was not using bash but csh and for some reason it was not following the command at the begging of my script to switch to bash. He fixed that and qsub is running.

The problem that I am facing right now is that when I run it it is like it's running somehow the case n-times (n=number of proccessors-ish) so it is actually taking longer.

I tried this in my own computer with a simple cavity case and it does the same, when I introduce:

mpirun -np 8 icoFoam -parallel > log.parallel &


it runs the case 8 times hence talking longer (it shows when i check with top-command) Am I doing something wrong??

thank you so much for your help

have a nice day!

Heliana

[wyldckat]

Originally Posted by heliana60

I tried this in my own computer with a simple cavity case and it does the same, when I introduce:

mpirun -np 8 icoFoam -parallel > log.parallel &

it runs the case 8 times hence talking longer (it shows when i check with top-command) Am I doing something wrong??

Quick answer:
As long the "-parallel" option is used, then the 8 processes are running the case cooperatively.
If you had not used this option, then it would truly be running the same case 8 times independently.

The problem you're seeing is that you are not at least following the rule of thumb of a minimum of 50000 cells per processor. Have a look at this blog post of mine for a ton more details on this topic: Notes about running OpenFOAM in parallel

[heliana60]

thank you Bruno!

I am reading those notes now

Heliana