Core Minutes 1/24/2006  

Present:    Joanne Bogart, Philippe Bruel, Toby Burnett, Jim Chiang, Johann Cohen-Tanugi, Seth Digel, Richard Dubois, Dan Flath, Warren Focke, Tom Glanzman, Heather Kelly, Michael Kuss, David Landriu, Julie McEnery, Igor Pavlin, James Peachey, Leon Rochester

• DC2: (Julie, Richard) Kick-off is 5 weeks away.
  • Sky model It's just about ready to go; this has mostly been coordinated by Seth. Previously we were planning to bin only by geomagnetic latitude, but it appears we should also take zenith angle into account. This and other fixes to interleaving will be checked out with a 1-day test run.
  • IRFs Enthusiastic volunteers are fitting IRFS for Bill's three event classes.
  • The data It should take about 2 days to generate the 55-day data set. Expect about 1-1/2 terabytes of generated telemetry; 3 terabytes when you add in the processed data. We'll apply cut zero, excluding 98-99% of the events, (see the discussion in Confluence, particularly the definition of TCut Basic in the code sample) before most people see it, though the full data set will be available to raw data groupies.

    Post-processing will be needed to

    • Fix live time counters in n-tuple to give global livetime for the complete data set, not per-job numbers
    • Remove branches, apply cuts,...
    • Work on pointing history. It should have a snapshot every 30 seconds.
• EM (Heather) There have been only a couple small upgrades recently; this is no time for big changes! AcdUtil and AcdRecon have been patched to be consistent with the change to overlap direction of ACD top tiles, which also affected ribbons. A new TkrRecon with a bug fix has been incorporated. EM is currently based on GlastRelease v7r3p5.
• Core week follow-on (Richard) A summary of what transpired was presented at this week's Analysis Group meeting. Hot topics include
  • gcc upgrade Navid has assembled the machinery for RM builds using gcc 3.4.3 (but with the old glibc version, so these builds are not suitable for installation on users' machines). Old extlib binaries, compiled with 3.2.3, may be used with new binaries of our code.
  • Mac support Navid has built ScienceTools and extlibs it requires on his Mac (no code changes necessary!) and has demonstrated that he can use the Mac batch cluster, so there appears to be no serious impediments to doing Mac builds with RM. He is also working on building GlastRelease.
  • G4 upgrade G4 v8.0 uses a new and incompatible version of CLHEP, making the upgrade process tedious at best. Tracy is most of the way through this, having modified 33 packages so far. Steve asks that we schedule carefully: e.g., don't simulaneously slip in a new G4 and new background model.
  • Gaudi and ROOT upgrades (Heather) Johann has offered to help with Linux build. How much priority does this have? When is it ok to make the move? Similarly for ROOT 5. The binaries are installed on Linux and it has been extensively tested. The same is not true for Windows, but Toby has used it privately with no difficulties.
• Data Server (Igor) The new version, currently running on the development server, has several improvements:
  • Users can add options to Pruner or Peeler jobs.
  • The user interface has been improved by using tabs [browser tabs, that is, not the whitespace character].
  • Versioning has been incorporated, so it will be possible to determine just what was being used at any time.
  • Mail notification has been enhanced to include more information.
  • Tom has asked that the latest versions of Pruner and Peeler, incorporating various bug fixes, be used.
  The new Data Server will be moved to production soon.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:48:22 -0700