Discussion of FCC Dataset Management System

Europe/Zurich
32/S-C22 (CERN)

32/S-C22

CERN

17
Show room on map
Zoom Meeting ID
66297467276
Host
Juraj Smiesko
Useful links
Join via phone
Zoom URL

Minutes from Discussion of FCC Dataset Management System

Present: Andre, Birgit, Brieuc, Giovanni, Jan, Juraj, Louis

  • Rucio + DIRAC are still not enough to cover all needs
    • In terms of metadata
    • ATLAS has additional AMI service to provide detailed metadata
  • No full metadata catalog in Rucio
    • Plugin can add this functionality
  • In case of Rucio we would need to come up with a dataset naming convention
    • DIRAC uses LFNs
  • Rucio provided as a CERN IT service for small experiments
    • No big hiccups encountered during the Pilot period
    • Not all actually started using it
  • Rucio containers can be nested
    • In newer versions there can "one line address"
  • Users are required to have x509 or token
    • This rises a bar for the common analyzer to join
    • Work under way to integrate with CERN Keykloak (Auth service)
  • Dataset relations are present in Rucio only as the file name/metadata
    • Dirac file catalog has it build in
    • We need provenance graphs when the data sample production will get more complex (FCC-hh)
  • FTS handles the Rucio data transfers
    • Dirac has plugin for it as well
  • We agreed to set up CNAF SE as Rucio SE and investigate the usage and needs
  • Validation capabilities boil down to metadata fields
  • Monitoring tools need to be set up separately and then brought to Rucio
  • Dirac DMS and Rucio are very similar capabilities wise
    • DIRAC can have diverse set of plugin
    • DIRAC can also be described as declarative
  • User production to be moved fully to DIRAC
    • No need for personal EventProducer
    • We need to have user production process as close to the real production as possible
    • Dirac user job can output to a common scratch disk and Rucio will pick the samples there (and turn them into official production)
  • Physics dataset validation is handled only by the MC production manager (Louis)
    • We need more people to join the effort
    • Can be handled in the FCC Physics website by metadata flag
    • The physics validation started to be developed at the PF workshop last week, not operational yet
There are minutes attached to this event. Show them.