LHCb Core Software Meeting

Date and Location

04 Dec 2013
10:30 - 11:20
CERN (2-R-030)

Attendees

Ben Couturier, Gloria Corti, Marco Cattaneo, Marco Clemencic (minutes), Markus Frank, Rob Lambert (Vidyo), Stefan Lohn

Subjects

News

  • AF on Thursday
    • there will be discussion on CERNVM (connected to our request for a a SLC6-compatible CERNVM)
  • ROOT 6.00-beta has been tagged, but it is not usable for us. Waiting/working for beta2.
  • Marco Ca.: we agreed to change the CMTCONFIG to SLC6 in January, but we need Ganga and LHCbDirac to be ready.

CERNVM/CVMFS

  • Accroding to Jakob, the SLC6-compatible CERNVM is 3.0 (AKA uCernVM), currently in beta, to be released by the end of the year
    • it's a real SLC6 with the root partition mounted as CVMFS, with the advantage that you will pull only what you need, and the disadvantage that you cannot add packages to it
    • the lack of configurability of CERNVM 3 seems a show stopper for large adoption (developers tend to want the tools they like)
  • Ben: investigating some observations by Beat, we discovered a few problems
    • we discovered that we didn't have the proper splitting of CVMFS catalogs, which is not difficult, but may cause problems on old (2.0) clients (re-mount may be required)
    • several grid machines at CERN are still on the old client (2.0), but these are SLC5 machines and the CVMFS client is not going to be upgraded
    • SetupProject treated CVMFS in a special way, to avoid entering each project/version directory (avoiding the downloading of unneeded catalogs), but this optimization does not work anymore because it was relying on the filesystem type to be "cvmfs" and now it is called "cvmfs2". A fix is already committed, waiting for a release of LbScripts.
    • Thanks to Beat for reporting his observation, which helped a lot in the understanding of the current CVMFS status

Round Table

Marco Cl.

  • Gaudi v24r2 tagged and built, ready for release.
    • Marco Ca. suggests to use it in lhcb-head
  • Nightly Builds
    • working on optimization of the builds
  • Created DecFilesTests project to test DecFiles without hacking Gauss.

Markus

  • Issue with gdb and gcc48.
    • we need gdb 7.6 to debug gcc48 builds. It's available in the externals, but we need to add it automatically to the path.
    • in the Online environment the problem is slightly different, because they need to run gdb as root. To be investigated
  • Discussion about the proper URI for raw data.
    • Rob had a problem with RAW files on EOS, because the simple URI was working in CASTOR, but not with EOS. It turned out that, because of the way ROOT handles CASTOR access, the "mdf:" qualifier was optional for local files and CASTOR, but it is mandatory for EOS.
      We agreed the correct way to open RAW files (wherever they are) is using the "mdf:" qualifier. It's just annoying that in some cases it can be omitted.

Gloria

  • Testing with new Geant4 builds
  • We still need to find a way to change the generators on the fly too

Ben

  • Slow progress on the periodic tests.
  • Benedikt will provide ASAP the configuration to prepare the RPMs.
  • Discussion with ATLAS
  • Will need a new release of LbScripts to pick up the fix in SetupProject for CVMFS.

Rob

  • gccxml has hardcoded gcc4.8.0 from AFS
  • Discussion with Ben and Marco Cl. on a splitting of the Stripping code. For January.


-- MarcoClemencic - 04 Dec 2013


This topic: LHCb > WebHome > LHCbComputing > CoreSoftwareMeetingMinutes > CoreSoftwareMeetingMinutes20131204
Topic revision: r1 - 2013-12-04 - MarcoClemencic
 
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