Minutes of the UKI monthly meeting, 24th May 2006
=================================================

Present: Stephen Burke (minutes), Greig Cowan, Mark Nelson, Matt Hodges, Matt Doidge,
Winnie Lacesso, Pete Gronbech, Duncan Rand, Chris Brew, Steve Traylen, Brian Davies,
Jens Jensen, Olivier van der Aa, Jeremy Coles, Derek Ross, Stephen Childs, Yves Coppens,
Alessandra Forti, William Hay, Gianfranco Sciacca, Santanu Das, Grigori Rybkine,
Dave Colling, Simon George


1) Transfer tests - status is on wiki, sites with red numbers needs to improve.
GC explained some of the reasons site have problems - many possible problems, but
mostly known now. Storage group will advise sites. Tests so far only with single files
which may be a bit unrealistic.

2) glite 3 deployment - Steve T discussed what RAL has done so far. LCG CE fairly easy
(3.0 has both LCG and glite CEs, which are quite different). Also new client tools
(LCG + glite) on WNs. UI install changes root profile, considered unnacceptable
(may also affect WNs, tarball is OK). R-GMA also upgraded, but a small change. glite CE
under way but difficult. SE install is somewhat decoupled from release, but DPM upgrade
seems smooth (YC).

  Timescale was originally for T2 upgrades to start on ~ May 15th, but delayed - T1
deployment is slow. Open to upgrade for T2 today, but too late for June 1st. AF
volunteers Mancester as a pilot T2 to upgrade, starting next week. PG says Oxford.
UCL central as well. Scotgrid maybe no time just yet. Final deadline unclear - what
is really needed for SC4? Will clash with T2 workshop anyway. Maybe aim to finish
by mid-July? glite CE not needed yet, but need to check that multiple CEs continue to
be supported.

3) Operational issues - operational procedures need to get tighter as LHC startup approaches. 24 hour cover at T1? Not clear what's needed at T2. Need to
meet MOU commitments. Scheduled downtime must be treated more strictly, e.g.
giving advance notice. Ticket response times need to improve. May be some changes to
weekly report to allow easier updating. May need more representation in the weekly
operations meeting - via T2Cs? Need better accounting, especially for storage - T1
now but will be needed at T2. No disk quotas yet, maybe by ~ end of year.

  Comment: don't enable minos yet, the VOMS server may move to FNAL.

4) AOB - Mention of the Gilda testbed if any Uk site wants to join. May be useful for
UK companies interested in investigating use of the grid. May attract PPARC tech
transfer funding. However, could divert effort from other things. Maybe some interest
at Cambridge.

  Some sites have gaps in APEL logs - sites should check that they are publishing.

  Sites should not mark SFT failures as "not relevant" without good reason.

  Footprints is changing to match site names automatically - sites should check
that tickets they get are correct.

  Some discussion about the use of scheduled downtime. What about unscheduled?
What about if specific services go down? JC will try to clarify.