DUNE UK Computing Meeting, 11 Nov 2019
========================================
https://indico.fnal.gov/event/22400/

Andrew McNab (AM, chair)
James Perry (JP)
Fergus Wilson (FW)
Matt Doidge (MD)
Raja Nandakumar (RN)
Peter Clarke (PC)
Chris Brew (CB)
Teng Li (TL)
Wenlong Yuan (WY)

Introduction and News (AM)
=====================

AM: Site renaming proposal: eg "London" -> "UK-Imperial" to show UK identity 
and clearer broadly agreed but not yet implemented.

PC: Possibilities of more capital for IRIS (and GridPP). 
    Excalibur call from EPSRC. Expect particle physics bid being put together,
     with LHC expts + DUNE approached in the UK, for software development. Need
     to be visible in HSFto justify requirements.
    SoI to STFC covering similar ground.

Roundtable
==========

Edinburgh
---------

JP: Working on RUCIO policy frameworks, factoring out experiment specific
    code in the main tree. Will allow external policies in parallel with
    expt specifics to start with.
TL: Working on data management monitoring

RAL-PPD
-------

RN: Resources allocation meeting at RAL Tier-1 (1PB disk confirmed)
    No plans to use RAL tape so no need to push for it.
    Current 250TB used, out of 1PB IRIS allocation (largest IRIS user at Tier1)
    SAM updated by Steve Timm to aid use of data in UK (@RAL) by jobs.
    Web page with monitoring plots: created and circulated URL:
      https://hepwww.pp.rl.ac.uk/users/nraja/DUNE/duneMonitoring-UK.html
PC: Can we monitor data access? To see if UK storage actually used?
RN: We can ask this afternoon at the DUNE computing sites meeting.
PC: Need to be able to demonstrate jobs are run and storage is accessed.

CB: 200TB for DUNE at RAL-PPD. Not yet added CI Logon Basic CA. Not needed?

FW: Dave Newbold suggested UK looks at online/offline interface

Manchester
----------

AM: Running random people's jobs in VMs. This week will now extend them to 
    other IRIS OpenStack sites. 
    (Multiprocessor production jobs should be fixed: but individual users
     may still do bad things.)
  
Lancaster
---------

MD: Not much running
PC: We need to keep pressure on DUNE centrally to understand if we are getting
    all the jobs we can

AOB
---

PC: Need to add Terry Froy to this list. QMUL agreed to join software
    group. Will set up PerfSonar mesh for DUNE sites around the world
    using existing boxes at sites. 

FW: Now GridPP6 approved, 15% of liaison to non-LHC experiments? What to do
    with that? 
PC: Some discussions within GridPP about this. Will raise it at the PMB.
AM: Next meeting Monday 2nd December as normal