Rucio Development Meeting

Thursday 11 Oct 2018, 15:00 → 16:30 Europe/Zurich

3196/R-023 (CERN)

Martin Barisits (CERN)

15:00 → 15:10 News¶

Rucio Community Workshop 2019 - Call for hosting.pdf

Rucio-logo-2018.zip

• Meeting format
  • Protected material
  • Pre-Filling of minutes and topics to discuss
• Meeting room
  • ATLAS P1 for 2018
  • 2019 meeting room closer to main building
• New Rucio project logo
  • Will be integrated in the website
  • Please use this one for presentations etc.
• 2nd Rucio community workshop
  • Call for Hosting until Oct 19
• Interest for a Rucio Coding camp in December 2018?
  • 2 days at CERN
  • Pre-select some specific developments and use two focused days to work on these

15:10 → 15:30 News from the experiments¶

• CMS developers at DUNE workshop at the moment
  • Currently working on Kubernetes setup
• DUNE workshop in Edinburgh - Cedric attended
  • Possible manpower to do some development in Rucio, possible objectstores;
  • Data Management session 
    • Cedric showed slides
    • Tape management different - More similar to STAGING area
• Ian at RAL migrating to PostgreSQL
  • Migration recipe - Should put it on ReadTheDocs
• Oracle schema from ATLAS DBA
  • Iterating through it to get it in sync with github/schema.sql
• News on CMS intance
  • Everything setup on Kubernetes cluster except authentication
  • No certificate to move data around
• LIGO / IceCube installation
  • Making progress: Moving files between RSEs

15:30 → 15:50 Hot topics¶

• Hanging Judge evaluator
  • Recently there was an issue of a hanging judge evaluator which lead to some unwanted replica deletion
  • As the application of replication rules (replica locks) to files is done asynchronously for performance reasons, a specific workflow lead to the deletion of replicas which should not have been deleted
    • 1: File₁ with Replica₁^A gets added to Dataset_X (which has a rule for RSE^A)
    • 2: Dataset_X gets deleted --> Rule gets removed --> Replica₁^A gets a tombstone
    • 3: File₁ gets added to Dataset_Y (which has a rule)
      • As the Judge evaluator was hanging, Replica₁^A never got the tombstone removed
    • 4: As RSE^A is full, Replica₁^A gets deleted within 3-4 hours
  • This was the first time the evaluator was hanging since 2014
  • The workflow was adapted to not remove Dataset_X immediately, but give it an expiration of 6h
  • Discuss if #1578 should be implemented, which would prevent the reaper from deleting data if it detects a backlog in the evaluator
    • --> Go ahead with the check

15:50 → 16:20 Developers roundtable¶

• Custom configuration / policy / permission files #533
  • Currently part of the repository & package
  • Should be removed from main repository and made simple for users to add
  • Load the module or path in the configuration (Can be path or module)
• Protection of sources by the reaper #1637
  • Currently the reaper is potentially over-protective of source replicas
  • Changes possible/necessary
    • If 1 source, do not allow deletion
    • If 2 sources or more, always keep the alphabetically first one
      • Issue for alphabetical selection?
        • Should not be, as replica becomes eligible after transfer finished
• Python 3 plan #1505 and #67
  • Via pyline --p3k
  • Hannes will prepare a script
• Rucio sometimes returns a 404 for a list-replicas call, even if the file exists #1568
  • Rerun arcls against integration service and try to find out based on the logfile
• Message payload #48
  • Add new column (CLOB) and only write messages in there
    • Only context switch when the non-clob column is empty
• ZIP Files #1091
  • Open ticket for list_replica inconsistencies
• Conveyor-consumer wrong handling of multi source jobs #704
  • In preparation
• Kubernetes
  • traefik released now, needs to be integrated at CERN
  • For WebUI minimal work, for authentication server more complex
• Database full for DOMA instance
  • message_history, rules_hist_recent, requests_history
• Nagios probes #1638
  • Need to evaluate what we want to use instead nagios?
  • Should be a separate repository for probes (maybe a common one)
  • Also probes which create network metrics

16:20 → 16:30 AOB¶

• Development meeting schedule for 2018
  • Development meeting every week except
    • 15. Nov 2018 - SC'18
    • 13. Dec 2018 - ATLAS Software & Computing week
      • Maybe CMS people can join ATLAS DDM Session?
    • 20. Dec 2018 - pre CERN-closure ???
      • People might be on vacation already - will check closer in time
    • 27. Dec 2018 - CERN closure
  • Any CMS constraints?
    • Computing week in 2 weeks, but CMS guys would like to join Rucio Dev Meeting in person
  • First meeting in 2019: 10. Jan 2019
• Dev meeting on the 25. Oct 2018:
  • Rucio 1.19.0 "Fantastic Donkey" release planning