LHCb Core Software Meeting

Date and Location

27-Apr-2011
10:35 - 11:20
CERN (2-R-030)

Attendees

Hubert, Karol, Marco Ca., Marco Cl. (minutes), Markus, Thomas (Hartmann)

Subjects

News

  • Gaudi v22r2 tagged and being released (some problems with the tarballs of the externals).
  • A released version of RELAX disappeared from the LCG release area. Benedikt put it back, but it's still not clear why it disappeared.
  • Introducing Thomas Hartmann
    • he is joining the LHCb Core Software group with some librarian-like tasks: nightlies, releases, tag collector
    • he will be at CERN for the next 2 months, then he will work from his home institute in Germany.

Changes to Improve Memory Usage (Markus)

In the context of bug #70850 (memory leak in stripping), Markus managed to reduce the memory required by our jobs changing the file format. The changes are, essentially those already proposed and discussed, with the addition of the optimization of the basket size.

To sum up, last year we agreed not to change 8yet) our persistency format because it didn't seem to improve enough the speed of our applications and the migration seemed to be tricky. Now we know that we need the new approach to improve out memory footprint, so we should try to change the format as soon as possible.

Changing the format will mean:

  • drop POOL (and GaudiPoolDb) in favor of a direct ROOT persistency
  • the new format will not be readable by old applications, while new applications will be able to read the old format
    • old applications requiring access to new files (e.g. Moore) can be patched on demand to use the new persistency

The change need to be carefully tested, before we can use it in production

  • the new package (currently in Online) will be integrated in LHCb HEAD (can be moved to Gaudi if/when needed)
  • all the configurations relying on GaudiPoolDb options have to be adapted to be able to work with any of the two persistency formats.

The progress is recorded in task #20232.

Round Table

Marco Ca.

  • Coverity builds are successful, but the analysis is not available in the database.
  • Windows' nightly builds are failing.
    • Karol: it is due to missing LCG externals, to be reported to Benedikt

Hubert

  • Problems with the creation of the LCGCMT tarball because of misunderstanding with Benedikt.
  • Need to change mkLCGCMTtar to properly handle the switch between optimized and debug for the externals.
  • Problem building Geant4: the fix to use the new InstallArea layout was in the trunk but never picked up in a tag and never noticed because of the "head of everything" flag being set.
    • trying to find the fix and re-tag
  • It was announced at the LIM that Grigori is preparing a new release of CMT
    • a preview will be available for testing.
  • Jobs failing at IN2P3 because of a bad interaction between local and shared areas. We need a version of AppConfig not distributed for testing: clone v3r99 as v3r99p1.
    • As a side note, it was discussed that the problem is also that the installation at IN2P3 has to be done by hand, and nowhere documented (only Joel and Hubert know how to do it). Luckily, IN2P3 will soon switch to CVMFS.


-- MarcoClemencic - 27-Apr-2011

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2011-04-27 - MarcoClemencic
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback