Meeting Notes for 18 July 2007 Wiki/Forum Issue Ruth's Request and Status of CMS e-Lab Swift and Pegasus Upgrade Path to grid-execution Mihael’s refactored code:

Meeting Notes for 18 July 2007

Present: Marge, Liz, Eric, Tom L., Mike W. Mihael, Tibi
Absent: Tom J., Bob

Wiki/Forum Issue

The QuarkNet Fellows (there are approx 50 of them) have requested discussion forums and a wiki to support collaborative work on e-Lab development. Eric had already set up both on the 'prototype' site at http://i2u2.spy-hill.net, but it is not clear if this has all the features that the Fellows have discussed. If so, or if necessary changes are small then this could be migrated to Argonne and integrated with rest of site. There is a need for separate spaces for technical discussion between developers, pedagogical discussion between teachers and educators, and "appropriate" segregation between students, and between teachers and students (however best we determine). These have to be "user friendly" and easy to use.

Action Item: Eric will write up the features of his Wiki/Forum site

Action Item: We will ask Bob whether his Fellows can write a list of their desired features.

Then we compare the two lists. Bob P. and his Fellows will see if Eric’s Wiki provides these, or Eric can estimate how much effort would be needed to provide particular features (short term or long term?), or we can consider Eric's site as a prototype and start over clean.

Eric can give guided tours of the Spy Hill site sometime in the future. Any interested Fellow can also set up an account and look around, just by getting an "invitation code" (ask Eric).

Eric mentioned the Sakai Project, which is an open source content management system for Education, developed by a consortium of universities - see http://www.sakaiproject.org and http://en.wikipedia.org/wiki/Sakai_Project). When we are working toward single-sign on for user accounts we might want to see if we can easily integrate with Sakai. Even though it is aimed at the University level, it is being used by some high schools, and the Sakai developers are actively considering how to facilitate use by high schools. Eric could tell us more later.

Eric also mentioned an Internet2 project to facilitate single-sign-on, but could not remember the name. It is the "Shibboleth" project. See http://shibboleth.internet2.edu/ Both Sakai and Blackboard can use Shibboleth, as can many other educational tools. So we should also look at this when working on single-sign-on.

Ruth's Request and Status of CMS e-Lab

Ruth Pordes has asked if there is a working example site which demonstrates submission of a job to The Grid.

It seems that Tibi’s CMS e-Lab on www12 is the farthest along and thus the most likely candidate. A job is sent to Grid engine (earlier version of Swift), which sends job to the grid site at UChicago (1 machine grid site). The task consists of a ROOT application and data already on the grid machine. To add more grid nodes we would have to copy ROOT and data files to a different grid site.

Features of the CMS e-Lab on www12

Eliminated Dependence on Replica Catalog.
Uses a Web front end.
Provides a queue of jobs as long as you are logged on.

Tom L. experienced problems when he and Ken Cecire tried to run from two different machines. Mihael noticed the server had run out of disk space. Mihael fixed this.

Additional Testing: Returned results in a timely manner when Tom L. did this again. Minor problem of labeling seems to be a problem. Cosmetic problems on the site; links to OGRE.

Tibi could use Jmeter to test the CMS site as he did for the Cosmic e-Lab.

Improvements: We need the newer version of OGRE to be able to do one of the studies. (Dan K., Tom L.)

Action Item: Tom L. will document the status of the CMS e-Lab on the Wiki.

Swift and Pegasus

Eric noted that LIGO grid folks at Caltech are using Pegasus, and wondered what differences there are between using Pegasus and Swift. Would it be possible to also run an analysis using Pegasus, or to use parts of an analysis originally developed for Pegasus with Swift? Mike says yes, that would be possible. They can share the same set of underlying applications and the catalogue of application descriptions. Mike noted that the Pegasus and Swift groups are in some sense competitors. Eric noted that in any case analyses developed for Pegasus were likely for gravitational wave analyses and not environmental data, but there might be something useful we could borrow from them someday. Topic will be continued off-line, but Eric just wanted to mention LIGO use of Pegasus when the name was mentioned.

Upgrade Path to grid-execution

Action Item: Mike W. will put this on the wiki.

Tibi, Eric and Mihael have a plan for how to do the LIGO analysis to run on the grid. Eric had started a wiki page outlining his expectations of an API for grid execution, and he will turn this into documentation of the interface for getting the LIGO Analysis Tool to submit and monitor Grid jobs.

Mihael’s refactored code:

Mike W: Mihael has a new code base, "with most of the cosmics e-Lab converted to the new code base."

We will have the VDC and Swift working together. We will need to update the management software for the references.

Action Item: Identify the unconverted parts. (Liz)

Mihael will need to make changes to code that have been added to CVS since Mihael took the code.

Need to add control to turn on grid or not, and in any case to note run conditions (how was it run)?

Meeting ended at 12:15 CDT.

-- Main.EricMyers - 18 Jul 2007 (from Liz's notes)