NA62 Monte Carlo Simulations

This wiki describes how to set up and distribute the NA62 Monte Carlo simulation code.

Software

• Test tarball from Antonino Sergi: NA62MC.tgz
• Job submission and output management system code components are held on and SVN repo at IC (access list: Dan, Janusz, Stuart). The files attached to this wiki are not the latest versions.

Code compilation

• Versions:
  • v4 - Contains bug in the MC which Tonino spotted after giving us the tar ball
  • v5 - Installed by Janusz on May 10 (r170, see table below)
  • v6 - September 2012 version (r188)
  • v7 - r193 uses GEANT4 9.5p01, improves processing speed roughly by a factor of 5 wrt the previous version
• Unpack tarball to NA62MC
• Fix libs path problems in GNUMakefile-s (Generator/ and Beam/)

     cp ../NA62/backups/Beam.GNUmakefile Beam/GNUmakefile
     cp ../NA62/backups/Generator.GNUmakefile Generator/GNUmakefile 

• Make changes to env.sh: source 'offline' environment, clear XAWLIBS variable, rearrange paths (move system paths after NA62-specific paths), basically pay attention to these two lines:

     source /afs/cern.ch/na62/offline/geant4.9.5.p01/env.sh
     export XAWLIBS=" "

• Source the the environment file (source scripts/env.sh)
• Type make and be patient

Code distribution

• Create tarballs for NA62MC, geant4, root, X11 and sw (CreateTarballs.sh). For example:

      tar -czvf NA62MC_v7.tgz NA62MC
      cd /afs/cern.ch
      tar -czvf ~/G4-libs.tgz sw/lcg/external/geant4/9.5.p01/x86_64-slc5-gcc43/lib
      tar -czvf ~/Qt-libs.tgz sw/lcg/external/qt/4.6.3p2/x86_64-slc5-gcc43-opt/lib

• Run a special job (with an SGM Admin role) to wget and unpack these tarballs on sites (InstallSoftware.sh)
• Run test job

Job submission

• Run parameters given in the StandardRun.mac file: run number, number of events, random seed, root output file name
• Job environment parameters are set by sourcing the na62jobenv.sh script created by InstallSoftware.sh
• The .mac is modified using sed on the worker node to modify paths to LAV matrix files, which are installed in the software area by InstallSoftware.sh
• The actual job script is attached (needs editing and streamlining)

Automated jobs scripts creator

There's now a suite of tools for submitting jobs either one-by-one or in bulk. Jobs submitted via this interface have the advantage that everything about the jobs (input and output files, scripts, run parameters and statistics) are archived for later reference.

• Toolbox located at: http://na62.gla.ac.uk/scripter.php
• Submitted jobs list (with a MySQL DB backend) is located here
• Jobs status iPhone webapp.
• Production rounds: listed here.

The script creator interface lets the production manager (PM) set up the job parameters (both MC parameters and output options) and choose the number of 'runs' to launch. The scripter prints out in the browser a list of commands that the PM has to run on his grid UI in order to launch the job or production with his credentials. For individual jobs, the PM updates the DB by hand, for multiple runs, it is done via a wrapper script. A cron job periodically updates the list of jobs and files in the DB.

Site resources

This is a simple table to keep track of site resources (please update if changes occur): Site CPU WMS Support Storage FTS Channel to RAL FTS Channel to CERN NA62 S/W Glasgow r188 IC r188 RAL -- r188 Liverpool   Birmingham   UCL       Note on s/w labelling: both a NA62 release number (e.g. r170) and a tarball version (e.g. v5) are used. They are kept track of in a dedicated DB table. Site admins - persons to contact in case something goes wrong on one of the sites are: Glasgow - Dan Protpopescu (dan.protopopeascu#glasgow.ac.uk), Stuart Purdie (stuart.purdie#glasgow.ac.uk) IC - Janusz Martyniak (janusz.martyniak#imperial.ac.uk) RAL - Andrew Lahiff (andrew.lahiff#stfc.ac.uk) Liverpool - Stephen Jones (sjones#hep.ph.liv.ac.uk) Birmingham - Mark Slater (mslater#cern.ch) UCL - Sergey Podolsky (spodolsky#mail.ru), Pavel Demin (pavel.demin#uclouvain.be) Please replace the #-s.

Cron jobs

Cronjobs are employed to keep track of jobs and output files:

To do before next production round

• Install new tarball once it's available from Tonino (Dan & Janusz)
• Open remaining FTS channels (Janusz)
• Move web UI to a dedicated monitoring webserver (Dan)
• Set up a controller-side cron job to double-check the FTS transfers of job outputs to CERN (Janusz)

Troubleshooting

To log problems encountered (and solutions) use this wiki:

• Troubleshooting page

Links

• NA62 voms server: https://voms.gridpp.ac.uk:8443/voms/na62.vo.gridpp.ac.uk/
• VO Info https://www.gridpp.ac.uk/wiki/GridPP_approved_VOs