VHcc

Search for VH(cc) process with CMS data, using coffea processor

Software setup

• First follow the Readme instructions from CoffeaRunner repo.
• Then, checkout this repo into ./src/ directory:

git clone [email protected]:cms-rwth/VHcc src/VHcc

• Re-compile: pip install -e .

Example job submission for VHcc

• Run the code using a config file:

python runner_wconfig.py --cfg src/VHcc/cfg_VHcc.py

• The processor where the selection is implemented is found at src/VHcc/workflows/Zll_process.py
• The config file - src/VHcc/cfg_VHcc.py - governs the submission parameters. For example:
  • dataset - the names of the datasets to run over, including the path to .json file obtained with fetcher.
    • Using samples and samples_exclude parameters one can run over specific samples.
  • workflow - the worflow to run
  • run_options - jobs submission parameters
  • userconfig - one can also create dedicated parameters for specific workflos.
  • More details on the config can be found in CoffeRunner's Readme

Useful commands to remove jobs from condor

Why?

To not fill the queue with more than the allowed number of individual jobs. (5000 I think)

(Because parsl does not delete finished jobs from the queue!)

How?

Remove successfully finished job (changes status to removed)

condor_rm --constraint 'JobStatus==4'

Actually force deletion of removed jobs from queue after being removed

condor_rm -forcex --constraint 'JobStatus==3'

If there are other jobs (unrelated) that should not be touched by this, there is also the option to remove a range of jobs.

condor_rm -forcex --constraint 'ClusterId > 21849531 && ClusterId < 21849551'