Speakers:
Manuel Giffels
(Karlsruher Institut für Technologie (KIT)),
Matthias Schnepf
(CMS (CMS-Experiment)), Dr
Max Fischer
(KIT SCC),
Michael Boehler, Dr
Oliver Freyermuth
(University of Bonn (DE)),
Ralf Florian von Cube
(Karlsruhe Institute of Technology (KIT))
KIT:
- AUDITOR instance at KIT upgraded to version v0.0.7
- contains meta structure
- adjusted HTCondor collector accordingly
- collects data successfully
- creates APEL report - output needs to be validated - but seems to be ok
- KIT added HEPScore values to several workernodes -> HEPSpec and HEPScore can be collected simultaneously
- HTCondor plugin will be added to the plugin dir in the AUDITOR repo on github
- harvester IDs should be added to AUDITOR, for AUDITOR <-> panda validation checks
- rough schedule: ready until next meeting
FR:
- APEL-plugin
- adjusted to meta structure
- todo: report both HEPSpec and HEPScore
- setup APEL client as linux service
- enlarge coverage of automatic tests
- provide documentation
- slurm collector
- collect info stored as pseudo json in slurm comment field
- AUDITOR core components:
- provided RUST and PYTHON blocking clients
- will be released in version v0.0.8
- python client can add runtime to record for testing purposes
WUP:
- auditor vs panda validation:
- job in auditor longer, since they represent the pilots and not the payload (as expected)
- long running jobs discrepancy < 1%
- panda requests per harvester id > factor 50 faster, than by batch system ID -> request to add harvester ID into AUDITOR db
BONN:
- benchmarks worker nodes with HEPScore
- currently benchmarks for full node usage works fine for nodes with ncores < 100
AOB:
- KIT was asked to patch APEL client (no manpower at APEL team)
- would be most beneficial for AUDITOR project to have a pre-production version out a.s.a.p
- next Meeting 17th April
