Meeting Notes for 27 February 2008

Present: Liz, Mihael, Tom L., Tom J.;
Absemt: Eric, Dan K, Mike W., Randy., Kris, Joao, Bob

Supporting Workshops on Saturday (Japan, California)

Kris Whelan wondered if there would be anyone available to give her support during her afternoon session on March 1 in Los Angeles, Bob adviced her to follow the instructions in the wiki (ClusterOps). She does not have the correct permissions; she has an account, but probably not an sshkey.

Consensus: Kris should use www18; it should be stable enough for her needs as Tom Jordan understands them. It is probably not appropriate for her to restart servers. She can use the production machines assessible at as fall-backs.

Refactored Code Status

Is it Ready for Prime Time

Liz asked if we could think about switching production to re-factored code on Monday. Others think we need more testing of a large group of users so Tom L. is going to have his students test it Feb. 28th using the protocol he has defined at:

We will also add an invitation on the production machines for others to try it.

Mihael reminded us that in the last round of testing on www12 at Tom’s school a few months ago, there was a network problem. Mihael was never able to understand this. He suspects that there might be something on the local side that limited the students from connecting.

The step to make production would be to replace 18 for 14 and 15 in the load-balancing. Keep the same entry point. Code is still in Tibi’s area. This approach will make it easy to switch back to the current production version if there is a serious problem.

Mihael's Calendar:
  • March 4th – arrives in Romania.
  • March 5th – procures some internet (and Romanian beer?)
  • March 6th – Target date for conversion of refactored code to production.

Releases: Re-factored and production

rc9 is running on www18:
  • flux fix (only channels that have at least one event are selectable)
  • addGroup.jsp login fix - won't mess up the url after logging in
  • CPLD fixes
  • new links for CMS and LIGO
  • Uploads progress bar is still NOT fixed.

VERSION_4_29 is running on 14 and 15:
  • new links for CMS and LIGO and pointers to www18.

Tom is still slowly systematically poking at the files that don’t have cpld_frequencies. If we can’t do it reliably, then we will assign it according to the board number. We are currently updating the ones in the database. We will commit those changes to production.

Joao's diagram for ATLAS/Amelia

The original diagram is in MeetingNotes20080220

Eric's comment

Eric made the following comment in e-mail that reflected the view of others:
I'd change the lable "Grid" of the big green blob to something like "ELabs server(s)".

Then create another blob under CERN and to the left of Amelia (forming a square), and label that "Grid".

The web interface, database of of events catalogued by users, and tools for statistical analysis will all be done by us on our servers -- except that we may farm out the big analysis tasks to "the grid". And those tasks may well have direct access to the Tier N (probably N=2) dataset, if we are able to do so.

So draw lines of interaction between the ELabs servers and "the grid", and between the grid and the ATLAS Events data pool.
Action Item: Liz will send an e-mail to Joao suggesting that he look at the diagram that Mihael made in ExecutionChoices to understand the parts we have in the Cosmic e-Lab. This might help him see why we are concerned about the label “Grid”.

Mihael/Liz's Background for Execution Choices

Tom L. is going to use this with his students and give feedbacks.

Helping Taiwanese

Mihael has been invited to go to Taiwan to help them with their e-Lab. He is trying to work out the logistics.

-- Main.LizQuigg - 27 Feb 2008
-- Main.EricMyers - 29 Feb 2008
Topic revision: r6 - 2019-05-22, AdminUser
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback