Page History

Turn Off History

New ClassAd Conversion TimeLine

Benchmarks

See ticket #227 for details on this work.

  1. A benchmark that instantiates N classads, adds A attributes into it, and then deletes them. Compare execution time and size b/w new and old classads.
    • Nick DONE

  2. A benchmark that instantiates N classads, then does a constraint evaluation on all of them. Compare execution time and size b/w new and old classads.
    • Nick DONE

  3. Grab the output of condor_status -l off one of the collectors (?) on a pool to get all the machine classads in the pool into a text file. Suck this up into a program which instantiates a classad for each of these machines. (Maybe make 10 instances of classads for each machine to magnify results). Check memory footprint; compare b/w new and old classads.
    • Nick Done

  4. Similarly with a job queue, grab the job classads, and suck them into a program which instantiates classads for each job in the text file. (Magnify as needed). Check memory footprint; compare b/w new and old classads.
    • Nick Done