cmssw + condor guide

No big job is done in just single running on single machine. First, one should know how to use the condor batch farm and dcache. These make your big job done N times fast and give you huge amount of disk space.

MC generation

make your own account

ssh connect on gate.sscc.uos.ac.kr

• Linux

  river99@tier2:~$ ssh -X garam@gate.sscc.uos.ac.kr
  garam@gate.sscc.uos.ac.kr's password:
  Last login: Tue Feb 26 05:02:58 2008 from aannecy-257-1-8-133.w90-14.abo.wanadoo.fr
  [garam@gate ~]$ _
        

  using -X option, X11 Forwarding's enabled.
• Windows
  • Analysis on Windows XP

cmssw environment

• make your work space

  scramv1 project -n my_work_space CMSSW CMSSW_1_7_5

• get ready for analysis

  cd my_work_space/src/
  eval `scramv1 runtime -sh`

  through this, all the environment variables in bash shell related to certain cmssw version on this work space are set.

• kerberos initialization

  kinit

  the purpose of this command is to get a ticket which will be used when you access the CMS CVS repository at CERN. kinit : check your ticket (if you don't have an account at CERN, having an ticket is impossible. kdestroy : throw your ticket away

• set cvs project area

  project CMSSW

• get the recent analysis code

  cvs co -r CMSSW_1_7_5 GeneratorInterface/Pythia6Interface

  check out command

• modify it

• build(compile and link) it

  scramv1 build

• run it
  • with condor -> see below
  • without condor -> just run 'cmsRun your_config.cfg'

Data analysis

-- GaramHahn - 25 Feb 2008v