Go to the Indico Home Page Oldest event Older event Up to category Newer event Newest event Printable version
  • Compact style
  • Indico style
  • Indico style - inline minutes
  • Indico style - numbered
  • Indico style - numbered + minutes
  • Indico Weeks View
    • Switch to the management area of this event
    Login

    ATLAS UK Cloud Support

    → Europe/London
    Vidyo

    Vidyo
    Tim Adye (Science and Technology Facilities Council STFC (GB)), James William Walder (Science and Technology Facilities Council STFC (GB))
    Hide

    GGUS

    - 146882 UKI-LT2-QMUL 
     Updated Singularity according to recommendations;
    Experiments will likely provide own Singularity version. Site would like to have their own Singularity , e.g. for smaller V0s, or for GPU

    - 146878 UKI-NORTHGRID-MAN-HEP 
    Networking issues; restricted by Covid-19;   Other VOs running with active FNAL connections also possible contributing factor

    - 146771 UKI-SCOTGRID-ECDF:
     Can ATLAS switch to using IPv4 http(s) to delete files against our DPM head node?
    Possible cause is issue between  centos6  and DPM signing;   
    Migration to  Centos 7 expected to fix this longterm to enable ipv6

    - 146651 RAL Singularity 
     # User namespace environment requested; 
    Question on which version of the singularity executable to be discussed with the site.

    - 146619 UKI-SOUTHGRID-RALPP
    To be followed up, if implemented and can be closed

    - 146588 Failovers from RAL-LCG2 to CERN CVMFS backup proxy
    * upgrade to 4.10 occurred.

    - 146525 UKI-NORTHGRID-SHEF-HEP: evicted jobs
    * Waiting for completion of rebuild

    - 146523 
    Networking issues believed to be the source/ can be Rucio mover bug ?
    Full network may cause the timeouts.

    BHAM only site to access MAN; could be moved ? 
    10 -> 40Gb/s (with potential 100Gb/s) planned (Covid-19) / stalled.
    Some mitigation might be possible  if remove storage from DOMA.

    - 146374 UKI-NORTHGRID-SHEF-HEP CE
    Awaiting response

    - 146159 UKI-SCOTGRID-GLASGOW_DATADISK
    Considered to be done now.
    Physical add new spares to the disk on 38  (Different disk than problematic 39)
    Prioritise to move away from DPM asap. 
     

    - 145688 Very old version of squids at UKI-NORTHGRID-MAN-HEP
    Specific squid 3 puppets module; needs testing before upgrading RPM
    Upgrading to 4.10  in planning

    - 145510 RAL-LCG2: timeouts on stage-in/outs
    Discussion on hyper-threading and performance monitoring.  Was noted that with Singularity, top can report much higher utilisation than expected.

    - 142329 Sussex:
    In progress

    CPU

    - Northgrid
    MAN - reassigned jobs affecting cpu.

    - Scotgrid
    DPM work ongoing in Durham

    Other new issues

    Durham:
     - Squid is deployed and running
     - ATLAS jobs now should be using non-SRM as default

    Ongoing issues

    - CentOS7 - Sussex: described above

    - Glasgow Ceph storage:
      Permissions problems writing to Ceph from Xroot and GridFTP believed due to the update to Nautilus and ceph plugin for Xrootd. 

    - Grand Unified queues
    Awaiting Sheffield; then will close

     

    News round-table

    - Vip:
       No decision on move to Storageless yet -  ATLAS could move to using other sites in between, e.g. trial period to switch across; to be followed-up

    Others NTR.

     

    There are minutes attached to this event. Show them.
      • 10:00 → 10:20
        Status 20m
        • Outstanding tickets 10m
          Open ATLAS UK GGUS tickets

          - 146882 UKI-LT2-QMUL 
           Updated Singularity according to recommendations;
          Experiments will likely provide own Singularity version. Site would like to have their own Singularity , e.g. for smaller V0s, or for GPU

          - 146878 UKI-NORTHGRID-MAN-HEP 
          Networking issues; restricted by Covid-19;   Other VOs running with active FNAL connections also possible contributing factor

          - 146771 UKI-SCOTGRID-ECDF:
           Can ATLAS switch to using IPv4 http(s) to delete files against our DPM head node?
          Possible cause is issue between  centos6  and DPM signing;   
          Migration to  Centos 7 expected to fix this longterm to enable ipv6

          - 146651 RAL Singularity 
           # User namespace environment requested; 
          Question on which version of the singularity executable to be discussed with the site.

          - 146619 UKI-SOUTHGRID-RALPP
          To be followed up, if implemented and can be closed

          - 146588 Failovers from RAL-LCG2 to CERN CVMFS backup proxy
          * upgrade to 4.10 occurred.

          - 146525 UKI-NORTHGRID-SHEF-HEP: evicted jobs
          * Waiting for completion of rebuild

          - 146523 
          Networking issues believed to be the source/ can be Rucio mover bug ?
          Full network may cause the timeouts.

          BHAM only site to access MAN; could be moved ? 
          10 -> 40Gb/s (with potential 100Gb/s) planned (Covid-19) / stalled.
          Some mitigation might be possible  if remove storage from DOMA.

          - 146374 UKI-NORTHGRID-SHEF-HEP CE
          Awaiting response

          - 146159 UKI-SCOTGRID-GLASGOW_DATADISK
          Considered to be done now.
          Physical add new spares to the disk on 38  (Different disk than problematic 39)
          Prioritise to move away from DPM asap. 
           

          - 145688 Very old version of squids at UKI-NORTHGRID-MAN-HEP
          Specific squid 3 puppets module; needs testing before upgrading RPM
          Upgrading to 4.10  in planning

          - 145510 RAL-LCG2: timeouts on stage-in/outs
          Discussion on hyper-threading and performance monitoring.  Was noted that with Singularity, top can report much higher utilisation than expected.

          - 142329 Sussex:
          In progress

        • CPU 5m
          UK Cloud jobs over last week

          - Northgrid
          MAN - reassigned jobs affecting cpu.

          - Scotgrid
          DPM work ongoing in Durham

        • Other new issues 5m

          Durham:
           - Squid is deployed and running
           - ATLAS jobs now should be using non-SRM as default

      • 10:20 → 10:40
        Ongoing issues 20m

        - CentOS7 - Sussex: described above

        - Glasgow Ceph storage:
          Permissions problems writing to Ceph from Xroot and GridFTP believed due to the update to Nautilus and ceph plugin for Xrootd. 

        - Grand Unified queues
        Awaiting Sheffield; then will close

      • 10:40 → 10:50
        News round-table 10m

        - Vip:
           No decision on move to Storageless yet -  ATLAS could move to using other sites in between, e.g. trial period to switch across; to be followed-up

        Others NTR.

      • 10:50 → 11:00
        AOB 10m

        NTR