Present:
Edinburgh (Phil, Alasdair)
London (Owen M)
RAL-Tier 1 (Andrew, David C, Steve)
RAL-Storage (Jens, Owen)
Apologies:
Glasgow (Fraser)
RAL-SRB (Peter)
> 1. Site update (all)
Fraser reported (by email) no news from NorthGrid.
Edinburgh reported that their systems will be down for
(expected) two weeks; after that they will be interested
in testing various SE implementations. In general,
they're "almost" at LCG2.
London at LCG2.0. Intend to test dCache.
New EDG release coming up with improved internal metadata
handling by end of this week; also an SRM pre-release coming
up but currently requires experts to help install and configure.
ACTION Owen S: determine how long edg-rm will keep supporting
interface to EDG SE.
SRB migration status couldn't be determined as Bristol didn't
join.
>
>
> 2. dCache evaluation status (Andrew?)
Testing but no schedule determined. Steve T and Derek Ross
will be testing. Cluster of 11 machines allocated.
>
> 3. LCG Disk pool manager proposal
There is a disk pool manager proposal from J. Casey, J.-P. Baud,
and I. Bird. Andrew circulated the document. The document
sketches a DPM-slash-tactical-SE as aligned with EGEE architecture.
ACTION Jens: determine what LCG actually plan to build and
what is the timeline for this.
>
> 4. SRM->SRB interface status + Timur & Don visiting (Phil)
Building an SRM->SRB interface is feasible but currently a
low priority. As long as files are only read, no problem,
but when files are written, that will be a "major headache".
It would be useful to identify projects who will require such
an interface. Owen M mentioned that people from White Rose
Grid have files in SRB and may need to access them from LCG
middleware (i.e. via SRM).
Background: Timur Perelmutov and Don Petravick are both from
Fermilab; Timur wrote the dCache SRM, and Don used to work
more directly with SRMs but has now been promoted too much :-)
Both will be coming to Edinburgh for a week after CHEP (4-8 Oct
by my calendar). Networking people will also be invited to
attend. Don is also interested in networking between SEs,
ideally in the sense that they can negotiate resources between
them as required. IIRC, he has a paper to this effect at
the next CHEP (http://chep2004.web.cern.ch/chep2004/).
--jens
|