Useful links

  • CMS Twiki page about 2015 collisions
  • Browsing information about the datasets ( DAS)
  • List of datasets present at LNL
  • Browsing and requesting data transfer ( Phedex)
  • Some useful documentation about [git (Chapter 1 and 2 is basically all you need to read)

Ntuple production

Installling Heppy (framework only)

The following recipe explains how to install our version of the Heppy framework on your local computer (either a t2-ui or a cloud virtual machine)

# Log in to a t2-ui or a cloud virtual machine

# Set the environment variable to get a 7_4_X CMSSW release
# on sh

# on csh
setenv SCRAM_ARCH slc6_amd64_gcc491
# Install CMSSW (preferably somewhere under /lustre/cmswork/YOURHOME/): 
cmsrel CMSSW_7_4_14
cd CMSSW_7_4_14/src

# Initialize git
git init

# Copy the spare-checkout (which contains the list of packages that have to be checked out from the remote repository) and enable it
cp /lustre/cmswork/mzanetti/cms/cmssw/git/sparse-checkout .git/info/
git config core.sparsecheckout true

# Set the remote repository and fetch the updates
git remote add origin
git fetch origin

# Checkout the current working branch (or more formally, create locally the a branch that tracks the remote branch)
git checkout --track origin/heppy-7_4_14

# Compile
scram b -j 8

Install DMPD (analyzers and producers)

The current repository can be found at link:

cd $CMSSW_BASE/src/
git clone


git checkout -b DMPD_$LOGNAME

scram b -j 8

Running Heppy

Once your code is settled, you are ready to run the ntuples production. The main files that needs to be edited is


. There you find -among the other things- the list of datasets you can run on. Select those that you care about by commenting/uncommenting the "selectedComponents" array.

To give it a try, you can select one sample only and run interactively:

cd $CMSSW_BASE/src/DMPD/Heppy/test/

otherwise to run the full production you need the following command: -o Batch -b 'bsub  < ./' 

before doing that you need to decide whether to run on DATA or on MC (Warning, important you need to launch the jobs separately for the two Warning, important). If you run on DATA, the command

 filterAnalyzer.processName = 'RECO' 

needs to be uncommented (and commented otherwise)


If you'd like to contribute to the DMPD framework (our flavor of Heppy), here are some useful instructions (or more precisely, how you can learn to stop worrying and love git).

# if you haven't done it before, track the reference remote repository and create your development branch from that
git checkout --track origin/heppy-7_4_12
git checkout -b heppy-7_4_12-DEV

# edit stuff (in this example "")in your development branch and commit it
git commit -a -m “comment to my commit”

# switch to the original (tracked) branch, fetch updates from the remote repository (in case it got updated by somebody else in the meanwhile) and merge your DEV branch 
git checkout heppy-7_4_12
git fetch origin
git merge heppy-7_4_12-DEV

# push it back to the server
git push origin heppy-7_4_12

# delete your local development branch
git commit -d heppy-7_4_12-DEV

-- MarcoZanetti - 2015-10-01

This topic: Main > WebPreferences > CMSPDAnalysis > 13TeVHowTo
Topic revision: r10 - 2015-12-10 - JacopoPazzini
This site is powered by the TWiki collaboration platform Powered by PerlCopyright &© 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback