Main Page

From FarmShare

(Difference between revisions)
Jump to: navigation, search
(corn info)
(local licensed software)
Line 77: Line 77:
=== local licensed software ===
=== local licensed software ===
-
We have a trial version of MATLAB Distributed Compute Server until 2011-12-25.  It is installed in /mnt/glusterfs/apps/MATLAB/R2011b
+
We have a trial version of MATLAB Distributed Compute Server until 2011-12-25.  It is installed in /mnt/glusterfs/apps/MATLAB/R2011b but it is not yet activated, please mail barley-alpha if you want to try it out.
You will need to write your MATLAB code to use the Parallel Computing Toolbox before using the Distributed Computing Server
You will need to write your MATLAB code to use the Parallel Computing Toolbox before using the Distributed Computing Server

Revision as of 15:26, 11 November 2011

+-+-+-+-+-+-+-+-+-+
|F|a|r|m|S|h|a|r|e|
+-+-+-+-+-+-+-+-+-+

This wiki is intended for the users of the Stanford shared research computing resources. E.g. the "cardinal" and "corn" and "barley" machines.

Contents

How to connect

The machines are available for anyone with a SUNetID. Simply "ssh corn.stanford.edu" with your SUNetID credentials. The DNS name "corn.stanford.edu" actually goes to a load balancer and it will connect you to a particular corn machine that has relatively low load.

The "barley" machines are only accessible via a resource manager (currently Open GridEngine). You'll need to ssh to corn-image-new.stanford.edu and a directory will be created for you on local (shared among barley) storage. E-mail the barley-alpha mailing list for more info.

cardinal info

The "cardinal" machines are older and slower and smaller and intended for long-running processes that are not resource intensive. E.g. mail/chat clients. You could log in to a cardinal and run a screen/tmux session there to do things on other machines...

corn info

Each of the 30 corn machines has 8 cores, 32GB RAM and ~70GB of local disk in /tmp.

barley info

Job submission now works from any corn.

Technical details:

  • 19 new machines, 24 cores each, 96GB RAM
  • 1 new machine, 24 cores, 192GB RAM
  • ~450GB local scratch on each
  • ~3TB in /mnt/glusterfs
  • Grid Engine v6.2u5 (via standard Debian package)
  • 10GbE interconnect (Juniper QFX3500 switch)

To start using the new machines, you can check out the man page for 'sge_intro' or the 'qhost', 'qstat', 'qsub' and 'qdel' commands.

Initial issues:

  • You are limited in space to your AFS homedir ($HOME) and local scratch disk on each node ($TMPDIR)
  • The execution hosts don't accept interactive jobs, only batch jobs for now.
  • You'll want to make sure you have your Kerberos TGT and your AFS token.

If you want to use the newer bigger storage:

  1. log into corn-image-new: "ssh sunetid@corn-image-new.stanford.edu"
  2. cd to /mnt/glusterfs/<your username> (or wait 5mins if it doesn't exist yet)
  3. write a job script: "$EDITOR test_job.script"
    1. see 'man qsub' for more info
    2. use env var $TMPDIR for local scratch
    3. use /mnt/glusterfs/<your username> for shared data directory
  4. submit the job for processing: "qsub -cwd test_job.script"
  5. monitor the jobs with "qstat -f -j JOBID"
    1. see 'man qstat' for more info
  6. check the output files that you specified in your job script (the input and output files must be in /mnt/glusterfs/)

Any questions, please email 'barley-alpha@lists.stanford.edu'

We plan to have "alpha" testing for a month or so, then rebuild the storage nodes using the information we learned, and also rebuild the execution hosts to Ubuntu 11.10. Then we'll have "beta testing" with more users in Nov and Dec and roll out to the full Stanford community on Jan 1.

barley software

stock software

The barley machines are running Ubuntu 11.04, and the software is from the Ubuntu repositories

licensed software

  • /usr/sweet/bin - MATLAB, SAS, stata, etc

local licensed software

We have a trial version of MATLAB Distributed Compute Server until 2011-12-25. It is installed in /mnt/glusterfs/apps/MATLAB/R2011b but it is not yet activated, please mail barley-alpha if you want to try it out.

You will need to write your MATLAB code to use the Parallel Computing Toolbox before using the Distributed Computing Server

more info: http://www.mathworks.com/support/product/DM/installation/ver_current/

current barley policies

  • 100 max jobs per user
  • 3000 max jobs in the system
  • 48hr max runtime for any job
  • no memory/CPU limits yet

If you have any concerns or suggestions, please mail barley-alpha.

Monitoring / Status

Current status of farmshare machines: http://barley-monitor.stanford.edu/ganglia/ More detailed graphs: http://barley-monitor.stanford.edu/munin/

For important announcements, we plan to:

  • modify /etc/motd on the corns
  • send a mail to farmshare-announce
  • duplicate that mail also to the farmshare "blog"

Mailing Lists

We have several mailing lists, all @lists.stanford.edu, most are not used.

  • barley-alpha - temp list till end of Oct/Nov 2011, for discussion around testing the barleys
  • farmshare-announce - announce list (new service name)
  • farmshare-discuss - users discussion (new service name)
  • stanford-timeshare-users - users discussion list for the corn users, list to be retired




Getting started with MediaWiki

Consult the User's Guide for information on using the wiki software.

Personal tools
Toolbox
LANGUAGES