Meeting Notes for 28 November 2007

Present: Eric, Mihael, Randy, Bob, Liz, Tom L. , Tom J.

Port 8080 and Reverse Proxy

Dale Ingram and Bob Peterson have noticed a few problems with port 8080 being blocked at schools. Eric M. set up a reverse proxy on www13 as a demo so Dale’s school thinks the content is coming from port 80. This will be fine for now for www13, but we will need to understand how to handle the different content that is on port 8080 and 80. One has the official i2u2 site and the other wiki. Eric would like to see the main i2u2.org page redesigned to include elements he's demonstrating on www13.i2u2.org. He suggests we take a look at designing the overall site from the top down. Liz pointed out this will involve a discussion with Marge. We will have to wait until she gets back in January.

We may not need to understand how we can combine this with the URL for load-balancing if we use Mihael's approach of having one main server and using the other machines for analysis only.

Mihael did a test that showed the feasibility of this approach: He added SSH support on our own local cluster. SWIFT is then able to farm out work to the other machines in the cluster. While the task is slower, this method might be useful for balancing load, instead of or in addition to balancing load with the web server.

Finding i2u2 with Google

Putting i2u2.org in Google will yield www.i2u2.org as the top result, while just searching for "i2u2" does not. This is useful when you are working in a workshop environment and want teachers to find e-Labs quickly.

Calender

Eric put up a sample Google calendar for us to use for i2u2. It works with an account with non-Google email; you do not need to have a gmail account. Eric suggested we might also want a QuarkNet calendar as well as the e-Lab calendar. Tom L. and Liz are both able to edit it, Bob will be too when he creates a Google account. Eric will check whether you can upload from iCal to the Google calendar. [Added later: it seems you can't. You get a copy of the calendar in iCal, not a subscription.]

Refactored Code

Mihael, Eric, Dale and Liz will have a telecon on Friday to discuss converting LIGO e-Lab to the refactored version.

New Data Server at Argonne

Mihael said it should be up today; he will check.

Joao

We will invite Joao from ATLAS to join the telecon now that he is back working on Amelia.

LIGO analysis

Eric announced a new release of Bluestone, the LIGO Analysis tool. Here is his email:

I'm happy to announce the rollout of a new version of Bluestone, the LIGO Analysis Tool (at http://tekoa.ligo-wa.caltech.edu/tla/ ). As you'll see from the version number in the lower left of every page, this is version 0.41.

While there are some small changes to the interface, most of the real changes are internal, including all those that were required to get this to also run on the ELabs cluster at Argonne. Some of the changes also lead toward being able to run an analysis on the Grid, though that is not quite all working yet. It should be soon.

See Tool,LIGOAnalysis(TLA)#Release_versions for more about Bluestone release versions.

CMS Analysis

Tom L. is going to make an educators’ wish list for CMS e-Lab to help Dan. This will involve engaging teachers including LHC and Cosmic e-Lab fellows in the discussions. They will be brainstorming something reasonable. New ideas they are batting around for CMS include studying di-muons. They can use Monte-Carlo data and then a trigger stream.

We also need feedback from teachers on the interfaces for the e-Lab analysis pages. Beth Marchant and Danielle could provide input on the interface.

Randy discussed the status of the CMS Test beam e-Lab. With the development of a full CMS e-Lab, what will be the testbeam e-Lab lifecycle? Is it a proof of concept or will it have a life of its own? Primary beam and a knowledge of the energy of the beam. A good precursor to the large scale data. Nice to have a testbeam environment as an introduction, but if it is not really “clean’, then it could be hard to use educationally (Tom J.) We had a problem with the labeling of a testbeam data set. That’s an error in a file name.

--A follow-up post: Test beam 04 data was correctly labeled last spring. But my experience was that piloting students did not have trouble with the data labeling. They understood the context quickly (incorrect entries in a new electronic logbook at the test beam facility), and took on the challenge of correctly identifying the particles from their signatures, just as CMS scientists will have to do when they examine the original data. --Tom L

One approach with the testbeam e-Lab; We could ask what an unknown beam is after they have seen how two named beams look.

Part of the concern is how much time Dan should spend on the testbeam e-Lab if we need to get on to the full CMS e-Lab. Liz pointed out that maybe with Mihael's implementation of a way to keep the analysis on Dan's machine, it would be not be as large a job for Dan to get the testbeam e-Lab upgraded. Mihael said this was true, but they still have to figure out:

• What the data files are?
• How you build the interface. Needs to be transformed into html.

Dan K is to be invited to the next telecon.--Follow-up post: Done--he'll join us 12/5/07--Tom L

Alina Bejan

Liz and Bob met with Alina Bejan, the OSG Education and Outreach Training Coordinator. They discussed some areas in which we might work with Alina. Alina gave a presentation at SC2007 for OSG and discovered interest in training on the use of e-Labs.

Action Items from last week

Action Item: Mihael will write his proposal for the use of the machines. DONE: See ServerReshuffling

Action item: Mihael will write a deployment plan for the refactored version DONE: See NewVersionDeploymentPlan20071121

-- Main.LizQuigg - 29 Nov 2007
-- Main.EricMyers - 02 Dec 2007