LIGO Data Grid

Note: These tools and instructions have been tested for Fedora Core 4. If you are interested in the client package for other platforms please send mail to Junwei Cao.

• Setting up your environment
• Grid OpenSSH
• Grid FTP
• Condor (local)
• Condor (remote)

Setting up your environment

Once at the beginning of each session you should do (if you use bash or sh as your shell)

source ldg-4.3/setup.sh
grid-proxy-init

Users that use csh or tcsh should do

source ldg-4.3/setup.csh
grid-proxy-init

When prompted for your certificate pass phrase enter it to generate a proxy certificate, used to authenticate around the LSC DataGrid.

Grid OpenSSH

Make sure that the Grid SSH is the one in your path:

$ which ssh
~/ldg-4.3/globus/bin/ssh

If so then you can just ssh to one of the LSC DataGrid server sites and will not need a password:

$ ssh hydra.phys.uwm.edu
[skoranda@hydra ]$

Note that you can also use this OpenSSH client to SSH to any host in the normal way using a password.

Grid FTP

The GridFTP is globus-url-copy. To copy a file from a remote GridFTP server do something like this:

globus-url-copy gsiftp://hydra.phys.uwm.edu/home/skoranda/foo file:/home/skoranda/foo

To put a file onto a GridFTP server do something like this:

globus-url-copy file:/home/skoranda/foo gsiftp://hydra.phys.uwm.edu/home/skoranda/foo

If the file is large add the following arguments to use multiple data channels and large TCP windows:

-p 4 -tcp-bs 1048576 -vb

For more details do

globus-url-copy -help

Condor (locally)

First start Condor if it is not running:

condor_on

Your local pool, ie. your machine should be ready:

[skoranda@kanga skoranda]$ condor_status

Name          OpSys       Arch   State      Activity   LoadAv Mem ActvtyTime

kanga.phys.uw LINUX       INTEL  Unclaimed  Idle       0.000   501 0+00:00:04

                     Machines Owner Claimed Unclaimed Matched Preempting

         INTEL/LINUX        1     0       0         1       0          0

               Total        1     0       0         1       0          0

Write a Condor submit file to describe your job. See also the Condor Manual:

[skoranda@kanga skoranda]$ cat hello.sub 
universe = vanilla
executable = /bin/hostname
output = hello.out
error = hello.err
log = hello.log
queue

Then submit your job using condor_submit:

[skoranda@kanga skoranda]$ condor_submit hello.sub
Submitting job(s).
Logging submit event(s).
1 job(s) submitted to cluster 2.

Watch your job's progress using condor_q:

[skoranda@kanga skoranda]$ condor_q

-- Submitter: kanga.phys.uwm.edu : <129.89.57.179:39331> : kanga.phys.uwm.edu
 ID      OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD               
   2.0   skoranda        8/18 07:00   0+00:00:00 R  0   0.0  hostname          

1 jobs; 0 idle, 1 running, 0 held

The stdout from your job will go into the file you listed in your submit file for output =:

[skoranda@kanga skoranda]$ cat hello.out  
kanga.phys.uwm.edu

When you are done you may turn Condor off if you wish:

[skoranda@kanga skoranda]$ condor_off
Sent "Kill-All-Daemons" command to local master

Condor (remote)

You can submit your Condor jobs to a remote Condor pool using your local Condor tools. Edit your submit script something like this:

[skoranda@kanga skoranda]$ cat hello.sub
universe = globus
globusscheduler = hydra.phys.uwm.edu/jobmanager-condor
executable = /bin/hostname
output = hello.$(Process).out
error = hello.$(Process).err
log = hello.$(Process).log
transfer_executable = false
queue 10

This will queue up 10 jobs on the Condor pool at UWM. To submit the jobs do

[skoranda@kanga skoranda]$ condor_submit hello.sub
Submitting job(s)..........
Logging submit event(s)..
10 job(s) submitted to cluster 3.

Your jobs will queue up and can be monitored using condor_q:

[skoranda@kanga skoranda]$ condor_q

-- Submitter: kanga.phys.uwm.edu : <129.89.57.179:39396> :
kanga.phys.uwm.edu
 ID      OWNER            SUBMITTED     RUN_TIME ST PRI SIZE CMD               
   3.0   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.1   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.2   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.3   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.4   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.5   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.6   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.7   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.8   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          
   3.9   skoranda        8/18 07:13   0+00:00:00 I  0   0.0  hostname          

10 jobs; 10 idle, 0 running, 0 held

The stdout from your jobs will be written back to your local machine:

[skoranda@kanga skoranda]$ cat hello.*.out
medusa-slave285.medusa.phys.uwm.edu
medusa-slave010.medusa.phys.uwm.edu
medusa-slave126.medusa.phys.uwm.edu
medusa-slave017.medusa.phys.uwm.edu
medusa-slave009.medusa.phys.uwm.edu
medusa-slave223.medusa.phys.uwm.edu
medusa-slave281.medusa.phys.uwm.edu
medusa-slave087.medusa.phys.uwm.edu
medusa-slave005.medusa.phys.uwm.edu
medusa-slave118.medusa.phys.uwm.edu

---

$Id: quicktools.html,v 1.16 2007/02/06 04:48:42 patrick Exp $

Supported by the National Science Foundation. Any opinions, findings and conclusions or recomendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF)

---

$Id: quicktools.html,v 1.16 2007/02/06 04:48:42 patrick Exp $