LHCb Core Software Meeting

Date and Location

10 Sep 2014
10:30 - 11:50
CERN (2-R-030)


Anastasia Karachaliou, Ben Couturier, Gloria Corti (Vidyo), Illya Shapoval, Joel Closier, Liang Sun, Marco Cattaneo, Marco Clemencic (minutes), Markus Frank, Sofia Qiriazi



  • Need For Speed meeting instead of Core Software once a month (second Wednesday of the month)
  • ROOT release reminder: end of September, feature freeze next week
    • Ben: memory leak in PyROOT when iterating over a TTree
  • Dead line for CHEP abstracts 15 of October.
  • CERN CentOS 7 ready for testing (in virtual machines)
    • ROOT 6 built on slc6 does not work on cc7
  • Quattor to be phased out at end of October
    • still issues with Puppet on physical machines

Specific Topics

Configuration of LCG nightly builds

With run 2 approaching, we need to validate our software stacks against the upcoming versions of the externals. We need the LCG nightlies for that, but they are in a strange shape: the versions used there are older than those used in the latest release. We would also like to test new compilers (for code analysis, performance studies, etc.). In this case it's partially our fault, because we never used the icc and clang builds in the LCG slot dev2, but the versions used there are old.

Our basic request is that the configuration of the LCG nightly builds mirrors the current state of the development and be proactive in providing technology previews (e.g. compilers). In particular we want:

  • slc6 (+ cc7 if enough manpower)
  • gcc48 (no upgrade), gcc49 (with c++14), clang35, icc14 (check c++11)
We also want that the next LCG releases are build on as many as possible of the development and test platforms.

For the configurations we need:

  • ROOT 6 preview with all the externals up to date (dev4)
  • ROOT 5 branch with all the externals up to date (dev4 + ROOT 5)

On our side we will use these new configurations in our slots (lhcb-lcg-head, lhcb-lcg-root5) and clean up the redundant slots.

Side notes

  • Ben: we need clang tools (as contrib)
  • Marco Ca.: we need coverity
  • Marco Cl.: resurrect coverage reports
    • Ben pointed out that wee can do it only on one build system (CMake).

Round Table

Marco Cl.

  • Nightly Builds
    • investigating new configurations to improve builds throughput
    • almost everything build before 8:00 (except the slots waiting for LCG nightlies, those scheduled for late morning and lhcb-prerelease that ended on a machine with some stuck processes eating up the CPU)
    • Joel: will need to benchmark VMs for the builds
  • CMake
    • fixed a few CMake issues in Online
    • improvements in the Gaudi CMake modules, require a new release of LbScripts (to become prod before the first stack built on CMake)
    • preparing the first release built with CMake, using the latest tags
      • need to coordinate with others tools (e.g. Ganga)


  • We need to think about Dirac LHCbDirac and Ganga with CMake
  • We need to plan the switch to Quattor to Puppet


  • Made progress on GaudiHive.
    • to be presented at the Concurrency Forum

Marco Ca.

  • New release stack in preparation, needed for Stripping 21


  • New build system for the releases working well. Still polishing needed.
    • we need to drop the old system
  • Deployment of nightly builds on CVMFS being tested
    • testing garbage collection (for stable deployment(
    • testing S3 storage (nearly as fast as local storage)
    • trying to get lhcbdev.cern.ch available on lxplus/lxbatch
  • Working on LHCbPR
    • running memcheck on the nightlies
    • writing custom analysis (updating/fixing the documentation)


  • Release of Gaudi with ROOT 6 used for the commissioning week
  • Updated numbers from Niko: 200 kCHF needed for upgrading the memory of the farm (10% of the farm cost)


  • Commited some changes in DetCond for the configuration for run 2


  • Need to push new views in CouchDB
  • Trying to understand the LHCbPR documentation


  • Investigating extraction of position information from a geometry.


  • Issue with rivet containing hardcoded paths. Being fixed in place.

-- MarcoClemencic - 10 Sep 2014

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2014-09-10 - 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 & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback