Home > Research Support Group > WestGrid > Site specific new user information

New User Message as of July 23, 2004

Greetings,

We have set up a WestGrid account for you
on nexus.westgrid.ca.

This is the gateway to WestGrid's large
shared-memory parallel computer facility
located at the University of Alberta.

You can use grid-enabled tools to access this
site's resources or login directly with the
username $USERNAME and password you supplied
during registration.

Please visit http://www.westgrid.ca for
general information about using WestGrid and
follow the links for site-specific details.
Contact support@westgrid.ca support@westgrid.ca
Thank you for your interest in WestGrid.

Regards,
   WestGrid Account Management Team
                                            

Here are some UofA Site-specific Details:

This local Westgrid facility consists of six SGI Origin machines with a current total of 452 processors:

• nexus 8 MIPS R16000 700 MHz processors with 8 GB of RAM
• arcturus 256 MIPS R16000 700 MHz processors with 256 GB of RAM
• helios 32 MIPS R14000 500 MHz processors with 16 GB of RAM
• corona* 32 MIPS R14000 500 MHz processors with 16 GB of RAM
• australis 64 MIPS R12000 400 MHz processors with 32 GB of RAM
• borealis 64 MIPS R12000 400 MHz processors with 16 GB of RAM

(*At the present time usage of corona is reserved for exclusive use of Dr. Rankin and his group.)

All of these machines may be used in batch mode, but only one, nexus, is available for interactive use. Nexus is the only one users are allowed to log on to, the rest are used by submitting a batch job to a PBS queue.

Getting Started, Logging In, and Submitting Batch Jobs:

You may access nexus interactively using ssh, sftp, or scp, but not using telnet or ftp, for security reasons.

ssh nexus.westgrid.ca

Interactive usage is allowed so that users may develop and submit batch jobs, and do limited testing. It is expected that users will run their jobs in batch mode.

Interactive users on nexus are asked to use at most 2 processors, and to limit their walltime to 1 hour. Interactive use is monitored.

Most of the processors are available only to users who submit their jobs to the batch queueing system (PBS). Using PBS is not difficult, however you will need to write and submit PBS job scripts. To learn how to do this and how to monitor your job's status please read our local users guide on:

PBS (Portable Batch System)

PBS will schedule your job according to your priority, so there will be a delay before your job starts running, but when your job does run it will have exclusive use of the processors it is running on.

The queues on all UofA WestGrid machines have a walltime limit of 24 hours. The other limits (such as minimum number of processors used) vary among the machines.

If you estimate that a particular job will run longer than the 24 hours, then you will need to incorporate checkpointing and/or signal handling into your program. We can help you with this; please feel free to contact us with any special requirements you might have.

There are also other limits, such as a limit of 6 simultaneous jobs running per user id. These are intended to increase the fairness of the system.

The Westgrid e-mail address for support is: support@westgrid.ca

Your $HOME disk space and working disk:

Storage space is provided through CXFS - a high performance filesystem shared by all the UofA Westgrid machines. CXFS is available on Nexus, Australis, and Arcturus. For Borealis, Corona, and Helios this space is served by NFS. You maight experience I/O performance degradation on NFS served space.

Please remember to remove files when they are no longer needed.

Regarding Back-ups:

We back up all files on a regular basis. However, clients must be aware that as with all file systems and backup facilities, problems can still occur. Users should still ensure that their critical files are archived in a separate location.

More site specific information is available at: http://www.ualberta.ca/AICT/RESEARCH/WestGrid/