Skip to content

README.md

Latest commit

 

History

History
119 lines (91 loc) · 7.49 KB

README.md

File metadata and controls

119 lines (91 loc) · 7.49 KB
Raw

CSCEfficiency/scripts: Making Ntuples

CRAB Submissions

Before submitting any jobs, make sure to authenticate your grid certificate. If you need a new one, visit this page. To authenticate the grid certificate, run voms-proxy-init -voms cms -valid 192:00.

To submit CRAB jobs, move to the submit/ directory and run crabSubmit.sh. For a quick summary of the script, run crabSubmit.sh --help. To customize the job that is submitted (such as dataset, request name, etc.) edit the settings.cfg file appropriately. If you want to submit a job with a pre-existing section's settings, you can run the following:

crabSubmit.sh <SECTION_NAME> ...

Here, <SECTION_NAME> refers to the name of the section with the settings you want to apply. If the name is not found in the settings file, an error will print and it will cancel the script. You can provide as many pre-existing section names in the command (symbolized by the ... above) and, as long as it exists, jobs will be run. There is a helper script that adds tab-completing for this command to help name the sections. To enable this, run source CSCEfficiency/submit/autocomp.sh.

Unused sections (like a section configuring 2023 jobs) can be deleted freely, but note that the settings used for those jobs are not saved elsewhere. The following sections describe how to create new setting sections to configure your jobs.

Configuring the job

To configure the job, you must edit the settings.cfg file to add a new config section corresponding to the dataset you wish to process. For example, if you want to process the dataset /Muon0/Run2025B-ZMu-PromptReco-v1/RAW-RECO, then you would add the following section:

[Muon0/2025Bv1]
dataset: /Muon0/Run2025B-ZMu-PromptReco-v1/RAW-RECO
globalTag: 150X_dataRun3_Prompt_v1

The section name is merely a convention to keep things organized - you can name it whatever you'd like. Some sections may be fairly bare and only contain these options, but certain defaults may need to be changed for some jobs. This way, the configuration for each job is saved in this file and we don't need to keep multiple long files per job. The required options in these user-defined sections are defined as follows:

  • dataset: The input dataset to be processed.
  • globalTag: The global tag for processing this dataset.

The appropriate global tag to process the data can be found on the Data Aggregation System (DAS) with the query dataset=..., where the desired dataset is provided. (For example, 2025 data can be found here.) Note that, depending on the desired dataset, a different CMSSW release may be needed. These can both be found by selecting Config on the desired dataset on DAS.

NOTE: Once you've configured this new section, go to the section option in the DEFAULT section and change the value to the name of your newly created section. This option controls what jobs are submitted - make sure to change this when you submit a new job. Alternatively, you can provide this name when running crabSubmit.sh. Often, the values in the DEFAULT section need to be different for a job. The next section explains how to do so without disrupting other user-defined sections.

Overwriting default options

Other than the section option, no values should be modified within the DEFAULT section itself! If you want to change one of these options for your analyis, add the value to your new section instead. For example, if you want to change the default requestName, your new section could look something like the following:

[Muon0/2025Bv1]
dataset: /Muon0/Run2025B-ZMu-PromptReco-v1/RAW-RECO
globalTag: 150X_dataRun3_Prompt_v1 
requestName: myNewRequestName

Note that the request name will always have the job attempt appended to it automatically. For example, when you submit the above example the request name will be set to myNewRequestName_1. If you were to submit the request again, the script will notice you already have a CRAB directory for that name and will increment the attempt counter. Then, the new job would be named myNewRequestName_2, and so on.

NOTE: The psetName option can be changed, but note that the cmsDriver file you use instead may not be able to process the input config params from crabConfig.py appropriately.

The additional options in the DEFAULT section are defined as follows:

  • psetName: The cmsDriver config file to run for processing. In most cases, this should not be overwritten.
  • unitsPerJob: The number of files to include in each job.
  • appendDate: Whether to append the current date (YYYY_MMDD) to the output dataset tag.
  • lumiMask: A lumi-mask to apply to the input dataset before analysis.
  • runRange: The runs and/or run ranges to process.
  • maxMemoryMB: The maximum memory (in MB) a job is allowed to use. (Single-core max is 5000MB and multi-core is 2500 per core.)
  • numCores: The number of requested cores per job. Note that the maxMemoryMB is total and will be divided equally per core.
  • dataID: An identifier for the dataset being processed. Note that the default value uses values for stream, era, and version, which are parsed from the input dataset option.
    • stream: The identifier for the split Muon primary dataset stream.
    • era: The acquisition era name of the dataset.
    • version: The processing version of the dataset.
  • requestName: A name given to the request.
  • tag: A custom string used in both the LFN of the output files and the publication dataset name.
  • outputFile: The name of the output file to be passed to the cmsRun script.
  • section: The config section in the file to pull process-specific options from for analysis. You can change this manually or by providing it as the first argument when running crabSubmit.sh.

Monitoring jobs

Once a CRAB job is successfully submitted, some information will be provided to stdout. If you run the provided crab status command and visit the Grafana webpage provided in the printout, the progress on the job can be monitored through a nice interface. Note, though, that it'll sometimes take 10-20 minutes to appear on this page.

Local submissions

Local jobs can also be submitted using ntuplize_cfg.py. It can be run from any directory, but to avoid clutter in the main package directory it is recommended to use a subdirectory (or simply submit). To see the available options, run

# assuming you run the command within CSCEfficiency/
cmsRun scripts/ntuplize_cfg.py --help

The required options for this script are inputFiles and globalTag. The outputFile option can also be set if desired. Note that the CRAB option should only be set to 1 for CRAB submissions. By default, it is 0 for local submissions. An example use of the command is below.

cd CSCEfficiency/submit
cmsRun ../scripts/ntuplize_cfg.py inputFiles=root://cms-xrd-global.cern.ch//store/data/Run2025B/Muon0/RAW-RECO/ZMu-PromptReco-v1/000/391/668/00000/b7adfb3e-c0f3-4c04-9908-b4a5cbacad1a.root globalTag=150X_dataRun3_Prompt_v1 outputFile=test.root