LHCONE R&D and RNT-WG call

Europe/Zurich
Edoardo Martelli (CERN), Joe Mambretti (International Center for Advanced Internet Research Northwestern University), Marian Babik (CERN), Shawn Mc Kee (University of Michigan (US))
Zoom Meeting ID
97610360540
Host
Edoardo Martelli
Useful links
Join via phone
Zoom URL
    • 17:00 17:20
      Jumbo frame survey 20m
      Presents: Edoardo Martelli, Joe Mambretti, Bill Jhonston, Shawn McKee, Dale Carder, Phil Demar, Garham Atterbury, Andy Lake, Laurent Duflot,, Jim Chen, Marian Babik, Tristan Sullivan, Thomas Araneta, Hiro, Vitialiy Kondratenko, Asif Shan, Carmen Misa, Bruno Hoeft, Tim Chown

      Discussion about the Jumbo Frame survey.
      - What is the audience? WLCG sites connected to LHCONE and not
      - Shawn cam mention the Survey on the ATLAS site Jamboree beginning of February
      - Recommended to use a proper survey tool
      - Invitation to answer could be sent to the GDB mailing list
      - Suggested to add a short paragraph with the motivation
      - In the past it was recommended to implement Jumbo on the LHCONE networks, but not on the end hosts https://indico.cern.ch/event/725706/contributions/3120030/attachments/1743507/2821722/LHCONE-MTU-recommendation.pdf
       


    • 17:20 17:40
      RNTWG 20m

      Packet marking
      - Firefly (R&E) dashboard now at:
        - https://public.stardust.es.net/d/b8dddac0-5b24-4739-9c8d-e88a05c1344f/scientific-network-tags3a-rande-dashboard?orgId=1
         - Waiting for feedback on this
         - This is ready to be deployed/tested on other collectors 
         - Potential improvements to consider:
              - Translate exp/act IDs to strings so they can be shown in the graphs 
              - Calculate flow durations (from start and end times)
              - Calculate transfer rates (from start/end times and bytes sent/recv) - will be approx.
         - Validation - we have benefit of having just one site, so we could compare UNL internal stats, ESnet UNL and fireflies to see what is our coverage 

      Packet pacing
      - Regular meetings moved to this meeting
      - Testbed Google sheet ready - thanks for volunteering nodes
        - Missing ring buffers information for some nodes
        - https://docs.google.com/spreadsheets/d/1U0VXIfWHfpK7bX7k2ucFep4Xo_4KQ5x-7rKD7e4az7Y/edit?usp=sharing
      - Test report (Carmen)
        - Google sheet was updated with results from recent testing (mostly focused on commissioning testbed and testing functionality)
        - 

      - Tests coordination - we have around 4-6 working weeks until DC24 
         - We will need to align testbed nodes - 1/2 weeks
              - Install BBRv3 kernel (instructions were sent to the net-wg)
              - Sysctl/ring buffers tuning 
              - On DTNs: install docker and perfsonar testpoint ?
         - This leaves us with ~ 4 weeks of testing before DC24 starts


      THANKS for your help in making 2023 a successful year for RNTWG Packet marking & pacing !
      Best wishes for the New Year !

    • 17:40 17:55
      AOB 15m
    • 17:55 18:00
      Next meeting 5m

      The meeting on the the 2nd of January is cancelled

      Next meeting: Tuesday 16th of January