GangaCK

Improving Ganga for better productivity.

Features:

• Jobtree: improved visualization of jobtree for better jobs organization. This can be called both inside/outside ganga interactive session.

  

• IOUtils: Misc operations to convert to/from (collection of) PFN, LFN, Bookkeeping uri (evt+std://, sim+std://), PPL, xml, lfns, eos, ... There is a caching algorithm to help where it's usefully applicable. One particular application is LHCbDataset.new where it can accept arbitary argument from the list of support inputs above. For example:

  LHCbDataset.new(
  
      'some/local/file.dst', # LOCAL
  
      'root://some-remote-file.dst',  # REMOTE
  
      'file:///another-remote-file.dst',  # REMOTE
  
      '/lhcb/MC/Dev/LDST/00041927/0000/00041927_00000002_1.ldst', # LFN
  
      'evt+std://MC/2012/42100000/Beam4000GeV-2012-MagDown-Nu2.5-Pythia8/...', # BKQ
  
      'sim+std://LHCb/Collision12/Beam4000GeV-VeloClosed-MagDown/...',  # BKQ
  
      '$EOS_HOME/ganga/4083/000.dst', # EOS
  
      '/cvmfs/lhcb.cern.ch/.../pool_xml_catalog_Reco14_Run125113.xml', # XML
  
      open('text_file_with_url_per_line.txt'), # local list
  
      jobs(123),  # output from another Ganga job.
  
      LHCbDataset(['foo', 'bar']),  # another ds.
  
  ) # accept heterogenous input appropriately,
  

• Magics: because ganga is embedded inside IPython, why not more magics?

  • jv : show status of subjobs from all running jobs. Extremely useful for monitoring.
  • jt : for improved jobtree operation.
  • peek: based on Job.peek, but look deeper when possible.
  • jsh : provide shell-like syntax to operate Job with less (no-shift) typing, for example, jsh 197.12 remove True instead of jobs("197.12").remove(True). Less typing saves your life's time...
  • grun: similar to the built-in magic ganga, but it'll pick the local ganga*.py immediately or ask in case of ambiguity.
  • resubmit: Smartly handle resubmission/backend.reset of failed Dirac jobs based on its failing status (e.g., "Pending Requests", "Job has reached the CPU limit of the queue", "Stalling for more than ...", etc.)

• Additional instance methods:

  • Job: lfn_list, lfn_size, lfn_purge, pfn_size, ppl_list, eos_list, humansize, is_final.
  • Gauss: nickname, to retrieve nickname from $DECFILESROOT.

Scripts:

• ganga_cache_viewer: display the list of cache made by this package.

• ganga_cleaner: Complete all-in-one script for tidying your ganga environment.

• offline_ganga_reader: Quick script to read the content in Ganga's JobTree offline.

• xmlgensum: Report summary of GeneratorLog.xml from all subjobs of Ganga-Gauss-Job

• xmlmerge: Merge summary.xml files from Ganga's subjobs and neatly archive the dir.

Installation

It's available on pip: pip install gangack

Disclaimer

This package was written and used during my PhD in 2013-2017 at EPFL (Lausanne) and LHCb collaboration (CERN), for the work in Z->tau tau cross-section measurement and H->mu tau searches at LHCb (8TeV).

As such, it's developped during the period of Ganga 5.34 -- 6.0.44. Because of the fast-pace development and non-backward compat nature of Ganga, this package can be obsoleted against newer version of Ganga.