(*) Rough notes from last meeting, mainly courtesy mjn. some from azs

   Notes from UM Working Group Meeting, Tue 9th Aug 2012
   11am - 12:30 pm, Rm 10E.
   Attendees: rab, yix, Ilia, mjn, mdix, azs, Say, Scott,
             Joerg, tpugh,lwenming. Apologies: mrezny
   Discussion topics:
   1) UM components progress
   2) TIDG Technical Infrastructure Development Group status
   3) NCI issues
   4) NECTAR Updates
   5) non-ACCESS components such as OPSUI, use of accesscollab
     server etc.

  1. UM Components, vn8.2  and related issues.
   - Initial build and run jobs available both for solar and vayu.
   - Version to build Martin's (UKMO's) way to be done.
   - remove "-O2" compiler option, no longer required with updated
     intel compiler according to Ilia.
   - Send patch of local config to UKMO.
   - Current runs are with N48. Other resolution (eg. N512) not yet tried.
   - Interest in vn7.8 1.5km ?
   - Thinking abour hourly cycling / upgrade to later version.
   - Ilia's segmentation fault crash with vn8.2 endgame & Glenn's run job
     -- problem went away when problem routine compiled with "bounds check".
   - rmifdf (#ifdef removal) work by mjn & Ilia. Currently on ND/EG 
     (Normal dynamics and EndGame code unification)
   - JULES repository now set up in access-svn
   - Evaluation with with profiling tools - Joerg writing report.
  2. TIDG

   - Bom firewall issue: For access of UMUI jobs on accesscollab from solar
     -- still unresolved.
   - Shared repositories with UKMO?
   - HPC project for allocation process
   - CSIRO projects: 2 projects for 2 pots of money
         access vs cmip5 - ???
   - CoE project groups
   - Following up from UM Users workshop in May, Technical Infrastructure
     work package are being finalised. Version from UKMO have been received
     and CAWCR's update is awaiting management approval.
       FTE committment
          UKMO: 4   KMA: 4
          BoM: 4 (including 0.5 from CoE and 1.5 from CSIRO ??)
   -- UM code Related issues
       -- code improvements
       -- shared repositories
       -- Test Framework
       -- UKMO access to NCI supercomputer
  3. NCI issues

   - There has been meetings with David Singleton and Ben Evans.
   - Migration plan for CAWCR research to new Petaflop machine.
      -- projects + needs + key staff
      -- get input from areas on requirements
      -- wiki with migration info
           cawcr_wiki -or- access_collab wiki
           Include NECTAR work too.
   - Next meetings
      - Seminars
      - none next month
      - NCI transfer WG meeting
      - later discussions on NCI aspects
  4. NeCTAR

   - Contract still not signed. No payment yet.
   - steering, stakeholders, teams
   - Project plans and ?use? cases
   - Basic + Community
   - Staffing
   - Email list
   - Question on NECTAR cloud by azs:
      --  available to Australian researchers
      --  big bunch of virtual machines
      --  app setup for software stack
      --  could be useful for ROSE testing.
   - OPS works on accesscollab, access-svn
   - IRIS ?

  - Express queue for solar requested by rab.
  - use of "non-excl" in solar jobs
  - add time to e77 project at NCI
      - CSIRO under utilitise quota in last quarter
      - Rob Bell could add significant usage.
  - accesscollab (virtual) server at NCI
     -- where to put software
     -- access user password
           --- mdix: add staff from access group on solar.
     -- owned by nci
         --- Administered by ??
         --- access usage managed by access userid
     -- access-svn : subversion server
     -- Documentation in CoE wiki
     -- Diskspace on accesscollab (?limited?)
  - ROSE
     - need to install python etc. on accesscollab
  - data transfers between BoM and NCI
  - MARS at NCI?
  - Don't expect to have large data areas available until mid Dec.
  - Peter keen for ROSE/cylc here on solar with hourly assim this year.
  - no  on nci
    -- all yix scripts now running on cylc
    -- issues with long running interactive windows
       -- need to add "keep alive"
  - cylc infrastruxture at NCIU
    -- all cycl user require their own cylc server daemon (like SMS) ?
    -- Yix currently run cylc on flurry, not solar

 - How to get added to access userid?
 - Mailing list access_help @
     -- add more people to the list
     -- currently  CMS, Martin, Say,   Margaret,  DavidS,  Robin
 - Increase some user's $HOME on vayu to  5GB and 50GB on /short
 - Transfer data -- by users
    --> /data/projects/access
   It was proposed that the local UM Working group meeting
   be held once a month on alternate fornight to "Policy Group"

[azs, 2/10/2012] Transfer notes from email 29/8/2012

Last modified 7 years ago Last modified on Oct 10, 2014 2:28:38 PM