Using CRAB at UW-HEP CMS
This is not a complete guide to CRAB. It is intended to be used in
combination with
Fermilab CRAB Instructions. This document provides UW-HEP
specific environment setup.
Quick Recipe
First, get a DOE grid certificate. The instructions for doing this
are here.
Verify that your grid certificate is installed correctly in .globus:
<![CDATA[
grid-proxy-init
]]>
If all goes well, the above should ask you for the password of your
private key and will then generate a temporary grid proxy credential
for you. A grid proxy is simply a temporary identity that is signed
by your DOE certificate. Various grid commands use this proxy to
identify you to grid services.
Once you have a grid certificate, request to be added to the CMS
Virtual Organization (VO). Instructions are here.
Notice that it says you must already be registered with the CMS
Secretariat. I am not familiar with the procedure for that, but I
will find out and fill in the details here later.
Once you have been added to the CMS VO (actually one hour afterwards),
you should be able to start using CRAB to submit analysis jobs to any
CMS grid site. The following example shows how to submit jobs to
UWMadisonCMS.
<![CDATA[
# login to a machine capable of submitting grid jobs
# example: login01, login02, login03
# CRAB Version
export CRAB_VER=CRAB_1_5_3
# CMSSW Version
export CMSSW_VER=CMSSW_1_5_2
#The following command only needs to be done once.
scramv1 project CMSSW $CMSSW_VER
#Load CMSSW environment settings
cd ${CMSSW_VER}/src
eval `scramv1 runtime -sh`
#Load CRAB environment
source /cms/sw/${CRAB_VER}/crab.sh
#The first time you use CRAB, you will need to initialize BOSS
/cms/sw/${CRAB_VER}/configureBoss
]]>
Now you have your environment all set up for using CRAB. You next
need to configure the analysis jobs.
<![CDATA[
# Choose a working directory for your CRAB activities.
# If you choose one in AFS (not highly recommended, but possible)
# then be sure it has ACLs set up to be writable by condor:
# fs setacl -dir /insert/path/here -acl condor-hosts rlidkw
# Otherwise, consider making a directory for yourself in /scratch
# and working in there. Just be aware that this is not backed up!
cd /path/to/your/analysis/workdir
#(e.g. /scratch/username/analysisname)
cp /cms/sw/${CRAB_VER}/examples/crab_wisc.cfg .
cp /cms/sw/${CRAB_VER}/examples/cmssw.cfg .
# Edit these two files to your liking.
# For anything other than a test, you will want to change
# total_number_of_events events_per_job
# Now create CRAB jobs.
crab -cfg ./crab_wisc.cfg -create 1
# Now submit CRAB jobs to Condor.
crab -submit
# You should be able to see your jobs in the condor queue now:
condor_q
]]>
For additional information on how to use CRAB, please read the Fermilab
CRAB Instructions.
References
CRAB Tutorial@FNAL
CRAB Webpage
CRAB TWiki (Oliver's Page)
CRAB FAQs