WLCG Archival Storage Group

Thursday 5 Jul 2018, 16:00 → 18:00 Europe/Zurich

31/S-027 (CERN)

WLCG Archival Storage Group

• Experiment plans
  •   Atlas carousel R&D
  •   Other experiments
• Finalising "best practise" documentation
  • Final text
  • https://twiki.cern.ch/twiki/bin/view/HEPTape/Survey_Conclusions
• Actions for FTS
  • Easy (configuration) and, maybe, harder (requiring dev)
• Actions for other retrieval orchestration systems
• Update on metric reporting

atlas-data-carousel-wlcg-wg.pdf

FTSUpdate.pdf

# WLCG Archival Storage Group Meeting Minutes

Thu 5th July 2018

https://indico.cern.ch/event/739586/

# Present

Alessandro Di Girolamo
Bo Jayatilaka
Dorin Lobontu
Doris Ressmann
Enrico Fatti
Jens Jensen
Oliver Keeble
Piere-Emmanuel Brinette
Rob
Xin Zhao
Andrea Manzi
Michael Davis
Giuseppe Lo Presti
Vlado Bahyl

# Xin's slides

Please see the slides themselves for the content

Not planning tests at T0
  ...because it's critical

Vlado - if you want to test T0, do so outside the heavy ion run oct/nov
Ale - good to test sites under stress
Giuseppe - but this ion run will be a very atypical time

Oliver - any problems with FTS? Submission rate OK?
  - no problems noted
  - no failures noted
Let FTS team know when the next test happens, we'll check monitoring

# Contributions from other experiments

None.

Need to get an explicit message from cms, lhcb and alice about their role in the group

# Finalising "best practise" documentation

No further comments. Oliver to create final doc from the text

# FTS report

Configuration of fts3-pilot has been updated to submit bulks of 500 files. No problems observed. FTS team proposes moving to 1000. No objections.

Oliver - summarised thread on managing a dCache queue. Do we need to use this advanced queue management when it's only useful for handling priority requests at the FTS level?
Ale: ...premature to implement priorities in FTS for staging
  better check tape families, tags etc.

Discussion on backing off submissions with FTS. Conclusion - FTS team to check feasibility of making "SRM_INTERNAL_ERROR in response to srmBringOnline" a non-fatal error.

# Info from other submission systems

None

# Metric reporting

# Next steps

Oliver - should we use this group to follow up results of Atlas tests, making results generally accessible, or is this happening in Atlas anyway?
Xin - good to use the WG for this
Oliver - propose for next meeting to have a technical dialogue between the dCache team and Atlas/sites/FTS. Topic would be preparation/optimisation for the Atlas tests. Accepted.

Next meeting September