When I run the .cmd file in the cluster I am getting an error message as

/bin/bash: : No such file or directory

The below is the PBS script in the .cmd file

#! /bin/bash
#PBS -o logfile.log
#PBS -e errorfile.err
#PBS -l cput=740:00:00
#PBS -l select=2:ncpus=8
tpdir=`echo $PBS_JOBID | cut -f 1 -d .`
tempdir=$HOME/work/job$tpdir
mkdir -p $tempdir
cd $tempdir
cp -R $PBS_O_WORKDIR/* .
fluent 2ddp -g -t16 -pib -cnf=$PBS_NODEFILE -ssh -i srik.jou >> out
mv ../job$tpdir $PBS_O_WORKDIR/.

Please help me to figure out the problem.

It means that one of the commands in your script is trying to access a file or directory that does not exist. Most likely it's one of the commands that uses a variable to represent a file or directory, and that variable is empty.

What's with the first line that uses $PBS_JOBID. That variable is never set. Is it set in the calling environment?

Also note:
You have a space between #! and /bin/bash, not sure how that is interpreted.
The lines 2 to 5 are not evaluated but treated as comments due to the # at the beginning.

/path/to/fluent

@Habitual

This is the output when I give in the command /path/to/fluent

-bash: /path/to/fluent: No such file or directory

@joe_2000

I removed the space but now I am getting this error.

-bash: /var/spool/PBS/mom_priv/jobs/18406.gnr.SC: /bin/bash^M: bad interpreter: No such file or directory

18406 is the Job ID which is automatically created when I execute the .cmd file

I took your file and changed the two logs to my preferred output in the test environment:

#PBS -e logs
#PBS -o logs

I then added:

set -vx

Just after the last PBS directive. To avoid trying to actually run fluent I added an echo to the front:

echo "fluent 2ddp -g -t16 -pib -cnf=$PBS_NODEFILE -ssh -i srik.jou" >> out

The job submitted ($ qsub X) and ran with no problems. Therefore:

1) As Habitual says, replace the bare command "fluent" with an absolute path to it.
2) Add set -vx as above.
3) Check that all referenced directories and the fluent command is visible across all nodes on the cluster.

If these don't work you'll have to post the error log showing the output from the set -vx so that we have a chance of seeing which line is failing.

^M means you have dos EOL characters, so you need to remove them. There is a command dos2unix to do that.

The PBS was given to me in our college website so there only the bare command "fluent" is given.
I wrote the following code in Notepad and saved it as "fluentpl.cmd" on my windows desktop PC.
I ran the commamd "qsub fluentpl.cmd" from SSH secure shell after uploading my files to the remote cluster.

I am getting the same error again as
"/bin/bash: : No such file or directory"

This is the PBS script file. I am attaching the screen shot of the error message.

#! /bin/bash
#PBS -o logfile.log
#PBS -e errorfile.err
#PBS -l cput=740:00:00
#PBS -l select=2:ncpus=8
set -vx
tpdir=`echo $PBS_JOBID | cut -f 1 -d .`
tempdir=$HOME/work/job$tpdir
mkdir -p $tempdir
cd $tempdir
cp -R $PBS_O_WORKDIR/* .
fluent 2ddp -g -t16 -pib -cnf=$PBS_NODEFILE -ssh -i srik.jou >> out
mv ../job$tpdir $PBS_O_WORKDIR/

Attached Thumbnails

 

pan64 was spot on with the ^M character, I must confess I'd assumed it was a transcription artefact. You confirmed it with your reference to Notepad. way back when DOS was first written they tried to avoid using a proper print driver (think one 5.25" floppy and no hard drive) by using <carriage return><line feed> to mark the end of text lines. The rest of the known (well, to me at least) universe uses a simple <line feed> as a delimeter and let the print driver add carriage control as required. The upshot of this is that text files are not portable between the M$ and *nix worlds without dos2unix and unix2dos programs.

Can you ssh onto the Linux host and issue the command:
You will see the actual physical path IF it can be found. It ought to be something like /usr/local/bin/fluent or /opt/fluent/fluent. Use that in the script.

I'm having to read this on a winbox and I'm not seeing your screenshots. Cut-and-paste is far easier for others to read.

You'll probably need to cut-and-paste the script into an editor on the *nix box. Don't bother with Notepad again, it'll never work properly for this task. The quickest to learn is vi, but it is pretty basic. On the plus side once learnt it ought to be on every *nix box you come across.

you may try:
qsub dos2unix fluentpl.cmd
(only once)
and afterward qsub fluentpl.cmd should work.

Thank you all for the help ..It is running..

One more question..In the below script I am requesting a huge wall time in the cluster so i am getting a message like

"qsub: Job rejected by all possible destinations"
Is it possible to somehow overcome this problem so that my file runs in the cluster as soon as possible like making it a priority file or routing it or something

#! /bin/bash
#PBS -o logfile.log
#PBS -e errorfile.err
#PBS -l cput=740:00:00
#PBS -l select=2:ncpus=8
set -vx
tpdir=`echo $PBS_JOBID | cut -f 1 -d .`
tempdir=$HOME/work/job$tpdir
mkdir -p $tempdir
cd $tempdir
cp -R $PBS_O_WORKDIR/* .
fluent 2ddp -g -t16 -pib -cnf=$PBS_NODEFILE -ssh -i srik.jou >> out
mv ../job$tpdir $PBS_O_WORKDIR/

First, cput is CPU time; the total across all processors. You need walltime if you really mean wall time.

Second, in the absence of further information the message indicates that no part of your cluster will accept such a large CPU request. You need to talk to your system administrator to determine the best way forward. Without sysadmin assistance the only thing you can do is to try and split the job into smaller parts.