Welcome to the Mykopat Grid Wiki

[Old Wiki page]​https://my-mgrid.mykopat.slu.se/trac/grid/wiki/oldwiki

What is the Mykopat Computing Grid?

The computing grid has been setup for users to be able to analyse very large genetic data sets, and other analysis requiring long computing times on Linux systems. A grid of computers is essentially a number of computers linked together by a common grid management software. The management software delegates how the job requests should be handled by the computer resources and most importantly makes sure the resources are shared in a fair manner.

Currently our grid consists of five computers available around the clock for long running high-resource jobs. These provide 40 cpu cores and 296GB ram. In addition to this we are also adding normal desktop computers as a part of the grid during non-office hours. Currently this adds somewhat more than 33 extra cores for short-running low-memory tasks.

Available Software

Here we have a list of installed Software. Please check the list before asking to have software installed. If you miss anything specific. Create an issue here where you specify what you need.

How to log in: There are several ways to access the grid:

1. From Macs and other unix-like computers you can login using the ssh-command in the terminal. You will then interact with the grid though commands in the terminal.
2. From Windows computers, you can login using Putty to login to a terminal session. You will then interact with the grid though commands in the terminal.
3. There is also an option to start a Remote Desktop on the grid login node.

Don't forget to run your jobs through the grid engine, otherwise they might just be stopped by the system when the default resource limits are exceeded.

Running commands

Once you have logged in to my-mgrid you will encounter a Linux system. (It is a Debian GNU/Linux system, quite similar to Ubuntu, but with some customisations to make it more suitable to act in a grid environment. If you are unfamiliar with work in a unix shell, you should consult som unix shell tutorial. We have a small Unix CheatSheet or quick reference that might be useful. You should also check the GridEngine pages for information on how to run long-running analysis.

Projects

You should register for a project before running analyses on the grid. This will give you a project directory available for all members, as well as provide a fair sharing of run time on the grid. Without any project assigned, the priority of your grid jobs will be very low.

There are some commands related to projects described on the ProjectCommands page.

In short:

• All project files should go into the project directory (/proj/project-name)
• Before submitting jobs, use the proj command to set your current project (e.g. once logged in proj my-project-name and the cd /proj/my-project-name)
• To conveniently change set current project and change directory into the project home, use proj -c my-project-name

Project Web Services

For each project a number of WebServices are available to use. These include:

• Project web folder?
• Subversion for version control of project files?
• Trac for project documentation (includes a wiki like this, issue handling and a nice browser for subversion)?

Frequently Asked Questions

Go here to find out information about common problems (for example: installing software in your local folder)

Technical documentation for Administrators is found in the Admin section of the wiki.