WLCG DOMA BDT Meeting

Wednesday 7 Sept 2022, 16:30 → 17:30 Europe/Zurich

Brian Paul Bockelman (University of Wisconsin Madison (US)), Maria Arsuaga Rios (CERN), Petr Vokac (Czech Technical University in Prague (CZ))

Topic: WLCG DOMA BDT Meeting (twiki)

16:30 → 16:35 News

16:35 → 16:50 Packet marking

Speakers: Marian Babik (CERN), Shawn Mc Kee (University of Michigan (US))

Next meeting in 30 minutes starting at 17:00.

Is there an intersection between activities of this group and SENSE project?

• not really, different requirements
• scalability issues with SENSE approach based on source IPv6 address
  • complicated to build / keep up to date complete database of source address mapping

 

We have opened three tickets (so far) with Rucio and FTS to try to implement some of the needed components for packet marking.

 

https://github.com/rucio/rucio/issues/5856
https://its.cern.ch/jira/browse/FTS-1829
https://its.cern.ch/jira/browse/DMC-1344

 

Have started a discussion with dCache on how best to proceed.

16:50 → 17:05 Transfers with tokens

Speaker: Francesco Giacomini (INFN CNAF)

XRootD 5.5 released

• Already available also in the EPEL testing
• Pass all our current compliance tests (e.g. simple se-prague-xrootd with this configuration)
• Please update EOS and Echo testbed

Storage scope path matching

• discussion that started during last BDT meeting, lead to followup in the WLCG JWT profile issue#21
• dCache treat storage.*:/foo scope path section with logic how to deal with files and directories (impossible to create /foobar with given scope, even without trailing slash in the scope path it is possible to create only something like /foo/bar)
• XRootD and StoRM just compare strings (currently it is possible to create /foobar)
• dCache implementation is considered the correct one and also preferred in related WLCG JWT profile discussion
• Action point:
  • create compliance test issue with details about additional test that verify behavior of tokens with storage scope path that's not terminated by "/" character

WLCG storage configuration for X.509 + tokens

• Experiment requirements collected in WLCG AuthZ github pages
• Asked developers/experts and wlcg-doma-tpc mailing list:
  • check requirements
  • provide minimalist configuration examples (documentation) for each storage technology and experiment
  • provide endpoints configured according this recommendation
• CMS is planning to reach developers/experts in next one/two months and come with configuration for a few sites that can be later generalized to all CMS sites
  • easier to discuss all concrete requirements and site specific tweaks directly with experts
  • ATLAS and LHCb requirements are very similar, so recommendation for CMS can be useful for other experiments
  • we should not forget there are other experiment, e.g. BelleII, DUNE
    • they often rely on same software (Rucio, Dirac)
• we should not forget about TAPE & tokens
  • right now our focus is TAPE REST with X.509 (Antares would like to use it in production already in Q2 2023)
  • tape recall prioritization is important feature, because TAPE transfer queue can be very long
    • longer queue => reduced tape mounts => more efficient data recall
    • it is not clear how to pass priority requirements
      • historically (CASTOR) priority was associated with user identity
      • in the capability mode with storage.stage scope it is not clear how to prioritize tape recalls

17:05 → 17:20 Tape REST access

Speaker: Mihai PATRASCOIU (CERN)

17:20 → 17:30 AOB