CRAB

Tips for CRAB Usage

When Copying Data to a T2 (copy_data = 1), Adjust so Each Job Produces 100MB-300MB of Output

TFileService Output is Automatically Added to the CRAB Outputfiles

EDM Output will be automatically returned if get_edm_ouput = 1

You Must Have An Account on the T2 You want to Output to--talk to Admin

If you output any other files, add their filenames to comma seperated list: ouput_file

To find your own crab output at UF T2:

lcg-ls -l srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=/cms/data/store/user/

Just take \"/cms/data\" off of the front of the filenames to use them in a CMSSW job.

Example CRAB cfg

[CRAB] jobtype = cmssw scheduler = glidein use_server = 1 [CMSSW] datasetpath=/TTJets_TuneZ2_7TeV-madgraph-tauola/Fall11-PU_Chamonix12_START44_V10-v2/AODSIM pset = copyData_cfg.py get_edm_output = 1 #total_number_of_lumis = -1 #number_of_jobs = 5 total_number_of_events = -1 events_per_job = 10000 [USER] eMail = youremail@blarg.com return_data = 0 copy_data = 1 storage_element = T2_US_Florida ## Won't matter if publishing, will be in <PrimaryDataset>/<publish_data_name>/<randomnumbers>/ user_remote_dir = dirname/subdirname #publish_data =1 #publish_data_name = mypublisheddatasetname #dbs_url_for_publication = https://cmsdbsprod.cern.ch:8443/cms_dbs_ph_analysis_02_writer/servlet/DBSServlet ## Do this if you get stupid error during crab -create, but be careful! #check_user_remote_dir = 0 [GRID] se_black_list =T2_US_Nebraska #My Arch Nemisis (Only the T2!) ;-) ## Will make your jobs exectue quicker at T2's once scheduled max_cpu_time =360 max_wall_clock_time =360

Script For Deleting Files From an SE

#!/bin/bash

CRABUSERNAME=myusername
BASEPATH=/cms/data/store/user/$CRABUSERNAME

#######################
### Define Functions
#######################

deleteFile ()
{
  if [ ! -n "$1" ] ; then 
      echo "deleteFile Error: Supply a file to be deleted"
      exit;
  fi;

  echo "Running srmrm srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=$1"
  srmrm srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=$1
}

deleteDir ()
{
  if [ ! -n "$1" ] ; then 
      echo "deleteDir Error: Supply a file to be deleted"
      exit;
  fi;

  echo "Running srmrmdir -recursive srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=$1"
  srmrmdir -recursive srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=$1
}

deleteAllFiles ()
{
  echo "Running deleteAllFiles($1)"
  for fn in `lcg-ls srm://srmb.ihepa.ufl.edu:8443/srm/v2/server?SFN=$1`
  do
    #echo "fn: $fn"
    if [[ $fn =~ ".*\.root" ]] ; then
        deleteFile $fn
    elif [[ "$fn" = $1 ]] ; then
        echo "Empty Dir: $1"
    else
        deleteAllFiles $fn
    fi
  done
}


################
## Run Stuff
################

deleteAllFiles $BASEPATH$1
deleteDir $BASEPATH$1

MultiCrab

Multi-Crab allows a user to submit the same crab job for multiple datasets. It is basically used in place of crab. A crab.cfg must be used similar to the one above, and then a multicrab.cfg file includes the parameters changed between datasets. Besides the [MULTICRAB]{=html} section that describes the crab.cfg file to use, each section defines another dataset to be processed, with parameters specific to it. If the data is staged out to a T2, the output will be put into a subdirectory of the directory listed in the crab.cfg called the section title in this file.

[MULTICRAB] cfg=crab.cfg #[QCDPt0to5] #CMSSW.datasetpath=/QCD_Pt-0to5_TuneZ2star_8TeV_pythia6/Summer12-PU_S7_START52_V9-v1/AODSIM #CMSSW.events_per_job=10000 # #[QCDPt5to15] #CMSSW.datasetpath=/QCD_Pt-5to15_TuneZ2star_8TeV_pythia6/Summer12-PU_S7_START52_V9-v1/AODSIM #CMSSW.events_per_job=10000 [QCDPt30to50] CMSSW.datasetpath=/QCD_Pt-30to50_TuneZ2star_8TeV_pythia6/Summer12-PU_S7_START52_V9-v1/AODSIM CMSSW.events_per_job=100000 [QCDPt50to80] CMSSW.datasetpath=/QCD_Pt-50to80_TuneZ2star_8TeV_pythia6/Summer12-PU_S7_START52_V9-v1/AODSIM CMSSW.events_per_job=100000

Hugonweb | CRAB

Tips for CRAB Usage

Example CRAB cfg

Script For Deleting Files From an SE

MultiCrab