How to run jobs on linux through pbs_script

harry · April 8, 2010, 21:51

Dear Sir/Madam,
I am going to run some fluent based simulations on a linux cluster. This cluster requires that (1) jobs need to submitted through pbs script, (2) each job last only for 48 hours (wall time). 2 months are required for per sequential job in my simulations.
In my case, how should I write a pbs script to run fluent on liunx, and let fluent dump restart file every 48 hours, stop, and restart based on previous simulation data.

I would appreciate your pbs script file, if possible.

Best regards,
Harry

pranab_jha · April 8, 2010, 23:38

Its a set of commands (just like in linux) you have to give to the server by typing it out in a pbs file. Contact your server administrator and he/she will let you know what all you will have to include. Generally it will contain your job name, # of processors you are using per node (if more than one), memory and other stuff. Also you will have to add the fluent module and read the journal file that you will create with fluent.

harry · April 8, 2010, 23:53

Thanks a lot. I have only experiences to run in-house code in clusters. Below is my understanding of your suggestions.

To run fluent in linux based on pbs_script, I need to use journal to let fluent read mesh and setup case file, right?

Another problem is that how can I let fluent stop after 48-hour calculation, and restart based on the previous results?

This is the critical problem I need to overcome.

If possible, I would appreciate you could send me a simple case which shows me the entire setup.

Best regards,
Harry

April 8, 2010, 21:51	How to run jobs on linux through pbs_script	#1
harry New Member Harry Join Date: Mar 2009 Posts: 27 Rep Power: 17	Dear Sir/Madam, I am going to run some fluent based simulations on a linux cluster. This cluster requires that (1) jobs need to submitted through pbs script, (2) each job last only for 48 hours (wall time). 2 months are required for per sequential job in my simulations. In my case, how should I write a pbs script to run fluent on liunx, and let fluent dump restart file every 48 hours, stop, and restart based on previous simulation data. I would appreciate your pbs script file, if possible. Best regards, Harry

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to run fluent in LINUX	Selina	FLUENT	26	December 4, 2019 17:25
proam problem in Star 3.20 Linux	David Baker	Siemens	5	February 7, 2007 17:23
Parallel Fluent on Linux clusters	Chengwang Lei	FLUENT	3	November 24, 2004 02:05
linux cluster	Johan Carlsson	Main CFD Forum	2	June 21, 2003 07:14
Running 2 CFD jobs on one PC	steve podleski	Main CFD Forum	17	February 16, 2000 14:40

April 8, 2010, 23:38		#2
pranab_jha Member Pranab N Jha Join Date: Nov 2009 Location: Houston, TX Posts: 86 Rep Power: 16	Its a set of commands (just like in linux) you have to give to the server by typing it out in a pbs file. Contact your server administrator and he/she will let you know what all you will have to include. Generally it will contain your job name, # of processors you are using per node (if more than one), memory and other stuff. Also you will have to add the fluent module and read the journal file that you will create with fluent.

April 8, 2010, 23:53		#3
harry New Member Harry Join Date: Mar 2009 Posts: 27 Rep Power: 17	Thanks a lot. I have only experiences to run in-house code in clusters. Below is my understanding of your suggestions. To run fluent in linux based on pbs_script, I need to use journal to let fluent read mesh and setup case file, right? Another problem is that how can I let fluent stop after 48-hour calculation, and restart based on the previous results? This is the critical problem I need to overcome. If possible, I would appreciate you could send me a simple case which shows me the entire setup. Best regards, Harry