Hi,

Please send me privetly any corrections in particular
if you were there and I missed your name apologies.

 Steve


Minutes GridPP Testbed/Deployment Meeting
Thursday Apr 15th 2004 14:00

SouthGrid
  Oxford        Pete Gronbech
  Birmingham    Laurie Lowe
  RAL-PPD       Barry Saunders, Chris Brew
  Oxford        Rhys Neuman, Pete Gronbech
  Cambridge     Santanu Das.

ScotGrid
  Durham        Mark Nelson
  Edinburgh     Steve Thorn
  Glasgow       ?

NorthGrid
  Manchester    Alessandro Forti.
  Liverpool     Barry King, Michael George
  Sheffield     Matt Robinson.

London
  Imperial       Dave Colling, Owen Maroney
  Royal Holloway ?
  Brunel         Akram Khan
  QMUL           Chris Williams

UCL              Peter Clarke

RAL Tier1        Steve Traylen

RAL GOC          Dave Kant

RAL, loose cannon?   Steve Burke

(I have not included email address since they are available here
 http://www.gridpp.ac.uk/members/, please add your self in if you are
 not there http://www.gridpp.ac.uk/website/gridpp-contacts.html )


+ Current LCG is LCG2_0_0 released Wed Apr 7, there are no changes
  at all other than the version number and the
  architecture (tier 1 change only) from the previous release
  lcg2_20040225_1700.

+ Is it worth having a LCG Getting Started Page on GridPP.
  This was thought to be useful to have a single page with a few links
  and nothing else.

  http://lcgdeploy.cvs.cern.ch/cgi-bin/lcgdeploy.cgi/lcg2/
  and docs/lcg2-install-notes.txt. Is the place to start.

+ Site Status and Reports.

  o Imperial and London (Dave C, Owen M)
    Last pre-LCG2.0.0 installed and in reality this is the same thing so
    nothing to do at this time.

    Owen Maroney is currently visiting other sites in london,
    helping and hoping to have the 5 london sites within LCG.

  o Oxford and South (Rhys Newman)

    20 brand new machines have arrived and will be installed now within
    LCG at Oxford. Experiance gained to be spread then across South Grid.

    Cambridge is already within LCG.

  o Birmingham
     Want to convert current EDG machines to  LCG2 machines.

  o RAL-PPD
     Starting to install LCG at previous equivalent tag. 4 nodes installed.
     No problems installing. (Just another LCFG install).

  o Bristol
     From Chris Brew 20 nodes for a test cluster in EDG now could go to LCG
     but no one obvious to do it. Possibly other babar nodes could join in the
     longer future.

  o Edinburgh
     Trying to get front ends together, front ends arriving in a week or two.

  o Glasgow
     LCG 2.0.0 installed including lcfg,  ce, se, wn, ui all at 2.0.0. Now following
     the paper work to join LCG2.

  o Durham
     8 front ends available. Plan a manual install of LCG2 in upcoming weeks. The
     LRM of choice is sun grid engine for which job managers exist. Information
     providers needed though for this. Dave Colling expresed intrest for LESC who
     also use SGE.

     Some 40 nodes running batch would idealy be joined to LCG in time.

  o Manchester
     LCFG and CE at LCG2 since yesterday will continue.

  o Liverpool, 500 nodes possible.

AOB

+ General questions about number of boxes. Are all these service nodes required
  CE, SE, ... ? Comments made that the CE and SE can easily fill up their CPU
  anyway.

+ Dave Colling said that frontend boxes form Gridpp were forthcoming for
  grid interfaces.

+ Question from Rhys. Where should I looks to see overall architecture document?
  Lots of blank faces but suggestions of EDG architecture documents and
  possibly EDG papers in the future. (Perhaps someone could provide links)


+ GridPP Map. GOC

  There should be a UK view of Grid resources.
  Dave Kant and GOC plan to add a UK view of LCG resources.
  Sites following the LCG install procedure will be added to the
  map automatically.
  http://goc.grid-support.ac.uk/,  follow the monitoring link from there.

+ Matt, Which testbed for HEP production work?

  LCG2 is the definitive answer and to some point at least LCG2
  is the same as EGEE production anyway but LCG2 is what exists today and
  has the longest future.

+ Santanu. A question about different hardware within the batch workers and how
  this should be represented in information system.

  .. You can't , GLUE assumes all hardware is the same.
  Select some middle ground perhaps with the exception of
  memory where you should publish the smallest memory available.


Next meeting will take place in 3 weeks on the 6th May 2004 at 14:00, I'll
post an agenda.























--
Steve Traylen
[log in to unmask]
http://www.gridpp.ac.uk/