Crabpage

How to submit jobs with crab

To submit jobs with Crab you have to do following steps:

• Make sure that the CMS/Crab environment is set up

Setup the Crab environment

• login to a machine where the Crab software is installed

 e.g: 

ssh -X <tt>whoami</tt>@master2.iihe.ac.be

• setup the environment
  • source the CMS environment

  setenv VO_CMS_SW_DIR /msa3/cmssoft
  source $VO_CMS_SW_DIR/cmsset_default.csh ## initialize the cms env

• • source the Crab environment

  source /msa3/crab/latest/crab.csh

• next go into your project directory and run eg (maybe replace scramv1 by scram and -sh by -csh)

  eval <tt>scramv1 runtime -csh</tt>

• Note that you can writ this commands in your .cshrc file

Adjust the crab.cfg to your specific needs

• Go to your working directory and copy the example crab.cfg from the working directory:

  cp $CRABDIR/python/crab.cfg .

• This file is the config for Crab
  • This file has different sections indicated with '[]'
  • [CMSSW]: CMSSW related: (number of events, cfg file, ...)
  • datasetpath = [string that describes dataset]. You can find this on DBS discovery page.
  • pset = [the name of the Pset which fits your code]. This is the config file which you use to a 'cmsRun' in an interactive run. Make sure it works before submitting to the grid
  • output_file = [The name for your output file]. Important: be consistent with the name inside your pset. Of course you can give more then one (comma separated)
  • total_number_of_events = [number of events you want to acces, -1 for all data]
  • Crab will work out the right number of evts/jobs according to the user requests: evts/job or nomber of jobs
    • events_per_job = [number of events accessed by a single job]
    • numbers_of_jobs = [number of jobs you want to run]
  • [USER]: user related: (where to store the output, ...)
  • return_data = [0|1]: Send output_file with stdout, stderr to User interface
  • copy_data = [0|1]: Copies output_file. If copy_data = 1, you have to specify storage_element = [name of the storage element to copy output files to] and storage_path = [path on the SE]. Important: crab makes the directory for you on the storage (if you have the permission), but the output file allready exists, the output file won't be written to storage.
  • [EDG]: grid related: ...

Create and submit a crab job

• Before you can submit your job you have to prepare everything that is needed

crab -create

• This will create a directory crab_0_<date>_ which contains:
  • job: contains CMSSW.cfg (config file for executable) and CMSSW.sh
  • res: output from 'crab -getoutput' comes her (CMSSW_000001.stderr and CMSSW_000001.stderr)
  • share: contains jobsubmission files and info
  • log: contains log files with crab output

• Once everything is created you can submit the job. By default it will submit all the jobs, specify N: number of jobs you want to submit

crab -submit <N> -c

Template:TracNotice