Meeting Notes for 29 August 2007 Issues with www11 Wiki/Forum Issue Mihael's refactored code: LIGO Analysis CMS Update

Meeting Notes for 29 August 2007

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

Issues with www11

Bob noticed that www11 was not working properly. (e.g., view.jsp was failing). Mihael checked the version number and it was correct. He discovered that NFS filesystems were not mounted. Mike suggests sending it to systems to let them know that this needs checking. It could possibly be related to the recent DHCP problem. For this machine to be part of the load-balancing pool it needs to be added to the server list in the load-balancing perl script.

Wiki/Forum Issue

(Carry over from last telecon)

Bob now says that his teachers want a place to store their documents (Word, PowerPoint, text, etc...) Eric suggested that whatever is used, it would be good to have a single sign-on, because this is a part of the idea of "one stop shopping". Eric is also interested in interacting with teachers to get feedback on what they like and don't, and to find out answers to questions like "what is simple?". Eric will also be coming out to Illinois so Eric and Bob could make progress on this then. Tom is also interested in using the forums Eric has set up.

Action Item: Bob, Eric and Tom L. will discuss this offline. Tom L. will set up a telecon.

Mihael's refactored code:

We have had ten iterations of testing and are very close to calling it ready. Mihael: "We are very close to calling everything besides the analysis part ready."

Mihael will put minor fixes on the server for Liz to test both on Macs and on PCc.

He will then put changes in Nick's changes to the analysis code to get production.

Bob will test that thoroughly.

Mihael will make changes for SWIFT that can be used for LIGO.

LIGO Analysis

The LIGO analysis was run on www13 from the command line, so all the software is working. Eric wants to get it working from the Web interface, using the PHP code he's written to collect the analysis parameters and launch a task. The original versions of Apache and PHP on www13 are not current and cannot be used, so he has built newer versions from source.

To launch a task on the grid the plan is to have Eric's PHP code perform an HTTP POST to a form handling script written by Mihael. It's as if an expert were filling in the form on behalf of the student, and indeed a bare form could be used for debugging. This is similar to what's done now for cosmics, only that uses HTTP GET (so all the parameters are passed in one long URL). This provides as clear demarcation between where the LIGO user interface ends and the CI grid interface begins. Eric will set up a test of this to run a task on www13 under Unix, just to demonstrate the idea.

Mihael suggested that he might set up something which uses ssh to launch tasks, to get something working quickly, and then plan to use GRAM later one for a production system.

Question: What would constitute success in running an analysis "on the grid?"
Mike: Running it on one or more OSG sites, not just our own machines.

CMS Update

New tools have been created for OGRE that allow doing the lateral size study. We need to add these. Tom L's students have used version 4 on Dan's machine rather than what is currently within the e-Lab. We discussed the feasibility of Dan adding the analysis code changes following Mihal's model. Tom L. noted the steep learning curve for Dan to understand what Tibi did. Mihael will talk to Mike about how to approach updating the CMS e-Lab.

Click!

We were cut off after 55 minutes and we did not get to finish the discussion of the Clean up of the CMS Test data. Tom L sent the following email:

"Just to finish the last point I was making prior to rude termination of the conference call: the CMS data already been "cleaned up" with respect to particle identification. There is an annoying source of noise which could be cut from the data, as well, and perhaps that should be done by Sept 30th. There is need to update the data set with test beam '06 data (adding to the current TB '04 data), but that isn't of near-term importance.

The only cosmetic clean-up of the e-Lab is some work on the references--at least one is incomplete--and some links which are misdirected. These I suspect I could send to Liz and ask her kindly to edit. I'll do a thorough review by next week--let's put it on the agenda for next week's telecon, in fact.

Items of that sort can be completed by Sept 30th.

Mihael, perhaps Dan and I can meet with you in Chicago, when the time seems right to you. We do need some sort of decision regarding the direction Dan should take: pile into Tibi's redaction of the OGRE analysis code, or await Mihael's adjustment to that code. But this is not a Sept 30 issue, unless the desire is to update the www12 (Tibi’s) OGRE with changes Dan has made to accommodate the "lateral size" study. I wouldn't recommend that as a high priority."

Meeting ended at 11:55 AM CDT.

-- Main.LizQuigg - 29 Aug 2007
-- Main.EricMyers - 04 Sep 2007