Tape Challenge '22

Thursday 10 Mar 2022, 17:00 → 18:00 Europe/Zurich

Alessandra Forti (University of Manchester (GB)), Luca Mascetti (CERN), Maria Arsuaga Rios (CERN), Xin Zhao (Brookhaven National Laboratory (US))

• Attendees :  Maria Arsuaga, Lisa Paspalaki, Alastair Dewhurst, Artur Darovic Gottmann, David Mason, Alessandra Forti, Benedikt Maier, Christophe Haen, Dorin Lobontu, Enrico Fattibene, George Patargias, Haykuhi, James Walder, Joao Pedro Lopes, Julien Leduc, Katy Ellis, Luca Mascetti, Mario Lassnig, Mihai Patrascoiu, Oliver Keeble, Shigeki, Steven Murray, Vladimir Sapunenko, Xin Zhao

 

• ATLAS : not much changes to the plan   

  • DDM has created the subscriptions of data export to T1s

    • Will double check with SFO experts on the exact names of the RAW data streams

  • Relevant experts notified for tuning the FTS throughput limits for all T1s 

    • Default is 3.5GB/s per site.  If sites want to set it higher, please let ATLAS DDM know before next week. 

  • Will skip NDGF in the peak export rate test, as the new tape library at one of the NDGF tape sites is delayed. 

 

• CMS plan :  

  • A-DT test ongoing this week

    • In contact with sites to better understand various performance. Too soon to draw any conclusion. 

    • Bigger test volume observed from sites, double counting somewhere ? 

  • DT test next week

    • P5 will used part of the DT test for 1.5 days. Exact rate targets at T0 to be confirmed.

    • The exact time to start is to be decided. Time to write to T1s may vary due to time zone differences.

• LHCb plan : 

  • ggdoc updated with expected tape write/read rates in 2022. 

    • Sites may want to double check 

  • Start time for DT test will be on 16th (Wed.)

    • Unfortunately, this will not overlap with DAQ days of ATLAS and CMS.  

  • Data cleanup from disk buffer : after the DT week, will remove data from T1 disks centrally. For data on tape disk buffer, will coordinate with sites to clean up (list of files to be shared with sites)

 

• Monitoring

  • NCBJ tape endpoint added to monitoring plot on the central dashboard. 

  • Horizontal line on the T1 tape throughput plots are removed to avoid misleading, as they represent peak rates for a certain operation mode, not all the time. 

 

• Sites 

  • RAL : Antares Status after recent migration. 

    • Ready to go for the tape challenge

    • Contingency plan from VOs in case of issues ? FTS can be used to control the real traffic. FTS team will keep an eye.  

  • KIT 

    • Will use the new tape system for CMS/LHCb and old tape system for ATLAS in this challenge. 

1 plan from experiments and sites

-- ATLAS
-- CMS
-- LHCb
-- Monitoring
-- T0/T1s (RAL Antares, ...)

Speakers: Alastair Dewhurst (Science and Technology Facilities Council STFC (GB)), Alessandra Forti (University of Manchester (GB)), Benedikt Maier (CERN), Christophe Haen (CERN), Garyfallia Paspalaki (Purdue University (US)), James William Walder (Science and Technology Facilities Council STFC (GB)), Luca Mascetti (CERN), Maria Arsuaga Rios (CERN), Xin Zhao (Brookhaven National Laboratory (US))

Antares20220310.pdf

2 AOB