How to use Code_Aster on the University parallel clusters

Here you can find detailed informations on how you can use Code_Aster on the University parallel clusters directly from your PC.

Which cluster you can use

At the present Code_Aster is installed and maintained on the following parallel clusters:

Which version of Code_Aster you can use

Both the operating and development versions are installed and maintained.

Each version has a stable (STA) and a new branch (NEW). The stable branch is updated only when a new stable branch is released (about every six months) starting from the latest NEW branch available. In this case, the NEW branch becomes the STA branch and the STA branch becomes the old stable branch (OLD). The NEW branch is updated more frequently, as follows:

  • for the development version, a NEW branch is released each week in order to follow the evolutions of the code. It includes both bug corrections and new features for the development version of Code_Aster
  • for the operating version, a NEW branch is released each time a bug correction is neeeded for the actual operating version of Code_Aster. No new features are introduced in the operating version
The development version of Code_Aster must be considered an unstable release of the code and the user is usually interested in the use of the operating version. Moreover the syntax of the commands used in the NEW development version can undergoes changes at each release. However if you need to use some new features of the code or if you are developing into Code_Aster, then you have no choice: the NEW development version is for you!

Step 1. Get an account on the cluster

Yeah, it seems a stupid thing but... first of all, you should have an account on the cluster in order to use it! You can find how following these links:

Step 2. Login to the cluster

You must use SSH to login to the cluster. Here you can find instructions on how to login and how to transfer files between the cluster and your PC. Briefly, what you should do depends on the OS you are using:

  • If you are using a unix-like system (Linux, MacOS ...), you can login using ssh from the command line and you can transfer files using scp on the command line or a graphical tool, like Dolphin/Konqueror in KDE, gFTP and so on.
  • If you are using Windows, you can login using Putty and you can transfer files using WinSCP.

Step 3. Configure aliases

You need to configure aliases just one time. This step is not mandatory and you can jump directly to step 4. However it facilitates the way you use to run Code_Aster. Once you have logged in, edit the bash configuration file, as follows:

  • type: vi .bashrc and then press ENTER
  • press i
  • move the cursor at the end of the file using the arrow keys on your keyboard and create a new line as you usually do in an editor (press Enter/Return on your keyboard)
  • add the following two lines:
alias astk='/software/Code_Aster/bin/astk'
alias as_run='/software/Code_Aster/bin/as_run'
  • press Esc on your keyboard and then press ENTER
  • press : and w on your keyboard and then press ENTER
  • press : and q on your keyboard and then press ENTER
  • type: source .bashrc and then press ENTER

Step 4. Submitting a job using ASTK

You just need to type astk (or type /software/Code_Aster/bin/astk if you didn't go through step 3) and the application used to run Code_Aster simulation opens. If you are using Windows, please remember to install and start up on your PC the Xming server in order to display ASTK GUI on your desktop.

If you get an error like "Unable to open display", than you must do the following:

  • in Linux, login using the command ssh -X username@cluster_address
  • in Windows, be sure that X11Forwarding is enabled in Putty options and that you have Xming server up and running

Everything concerning the simulation is controlled from ASTK. You can use both interactive and batch mode. In interactive mode the job is run on the frontal machine. In batch mode the job is submitted to the cluster queue system automatically. Please, use interactive mode only to compile your development version of Code_Aster. For all your simulations, use the batch mode. Keep in mind that this request is enforced by allowing limited resources to interactive jobs (1 CPU only, 128 MB of RAM and no more than 5 minutes job run).

In batch mode, the cluster batch system is automatically managed by ASTK and you don't have to bother with "esotic" configuration files for your simulations. You can choose between two groups of batch queues: parallel and serial. When you submit a job to a queue in the parallel group, you can use the parallelism built into the code (by means of openMP and openMPI). On the contrary, only one processor is used when a job is submitted to a queue in the serial group. You can specify which queue you want to use by selecting the menu Options->Classe and typing the name of the queue you want to use. If you don't specify the queue, the job will be submitted to a serial queue. The list of the queues available in the cluster can be retreived by selecting the menu Configuration->Servers. Here's an example:

queues.png

In this example, three parallel queues (parallel-R2, parallel-R4 and parallel-R5) and one serial queue (serial) are available. You can also read the time limit and the maximum number of processors that the job can use.

You can specify the number of CPUs you want to use by selecting the menu Options->ncpus.

Step 5. Monitor the execution of your jobs

After the submission of your job, you can monitor its execution in realtime. The first step is to get the job number and the job name. You can retreive these informations from the asjob window (if it's not displayed, press "Progress of jobs" button in ASTK). Here's an example:

asjob.png

There's one job running (RUN). It has been submitted on 27/05/2010 at 11:37:56. The job name is SEB and the job number is 29793.

In order to monitor its execution in realtime, just type the following command:

tail -f $HOME/flasheur/jobname.ojobnumber

substituting the correct values for your job. In the example above, the command would be: tail -f $HOME/flasheur/SEB.o29793

You can press CTRL+C to stop the realtime monitoring.

If you want to see all the message file:

nedit $HOME/flasheur/jobname.ojobnumber


Current Tags:
create new tag
, view all tags
Topic attachments
I Attachment Action Size Date Who Comment
pngpng asjob.png manage 12.3 K 2010-05-27 - 11:53 DanieleColombo Example of a running job
pngpng queues.png manage 23.0 K 2010-05-27 - 16:36 DanieleColombo  
Topic revision: r12 - 2011-02-21 - 13:47:48 - DanieleColombo
Aster Web
17 Dec 2017
Links
Home
Download
FAQ
News
Test Cases
Developments
Publications

tip Saturne Tip of the Day
Create multiple cases
You can create multiple cases with a single command, just add as many cases as you want in your cree ... Read on Read more

 

Code_Aster collaborative website
@ the University of Manchester
Copyright & by the contributing authors. Unless noted otherwise, all material on this web site is the property of the contributing authors.