CTA Dev Meeting

Europe/Zurich
31/S-028 (CERN)

31/S-028

CERN

30
Show room on map
Michael Davis (CERN)
    • 14:00 14:10
      CTA Catalogue Schema 10m

      Next CTA Schema Version

      • Version v10 backwards-incompatible changes #1151
      • Most schema changes are now in master
      • Need to add additional foreign key constraints to DISK_INSTANCE table (VO, ARCHIVE_FILE and the 3 mount rules tables)—will be done after Easter. We push back the next release until this is done.

      Schema and software upgrades

      • Last week's tape server crash is better understood : upgrading the schema before upgrading the software led to a NOT NULL constraint violation which caused the tape server to crash.
      • Schema change was not to blame per se, but it exposed a bug where getString was used to read a field which could contain NULL values instead of getOptionalString.
      • This schema change was "backwards compatible" but still broke the software!
      • Need a process to better evaluate dependencies between software changes and schema changes and order of software/schema upgrades.
    • 14:10 14:20
      Tape States/Failure Handling/Retry Logic 10m
      • Maintenance process should notify EOS when it cannot requeue a request #1176—fix bug where requests on a DISABLED tape were failed and lost.
      • Add new REPACKING tape state with same logic as DISABLED (see below)
      • Change logic of DISABLED #1170

      Repack

      • New drive state REPACKING has been added to the constraint in the DRIVE_STATE table.
      • Software support will be added by Joao later #1175

      Other tickets relating to retry logic

      We reviewed these last week, can be reviewed again after all of the above has been implemented.

      • #1023 Scheduler retry logic for archive and retrieve
      • #1099 Reasons for tape session failures
    • 14:20 14:30
      CTA Release Roadmap 10m

      See CTA Release Roadmap

      Roadmap Status

      • Schema changes
      • Disk system/Disk instance
      • Tape Server Drive/session status
      • Link with XRootD 5
      • Some unit tests not passing with RelWithDebugInfo
      • Clean up git repo #1182

      Deployment Schedule

      • Release 4.6.1-1 / DB schema v4.6 has been deployed in production
      • w/c 11/04/2022 is a short week with some people on holiday
      • w/c 18/04/2022 : freeze master for release 4.7 / DB schema v10.0
      • w/c 25/04/2022 : deploy v4.7 in PPS
      • w/c 2/05/2022 : deploy v4.7 in Production
    • 14:30 14:40
      CASTOR to CTA Migration 10m

      CASTOR clean-up tasks #677

      • 30 million 0-length files to be imported to EOSCTA public instances
      • CASTOR monitoring shows 87.5 million files (53.7 TB) left in CASTOR
      • Files migrated to CERNBox have been deleted.
      • Vlado to move supply tapes etc. to CTA
      • Giuseppe to eventually remove CERNBox files from Hadoop tapes in the r_backup pool so they can be repacked. (No deadline has been set).
    • 14:40 14:50
      AOB 10m
      • George's request that we grant access to tapeoperations-docs.
      • Richard's presentation of the reconciliation dashboards.