Below the minutes I took. Please feel free to correct and add.

cheers
alessandra

###############################################################################
Minutes of LCG GD Phone Conference , 2nd July 2004

Present

RAL: Steve T. Steven B. Dave K. Chris B.
Lancaster: Peter Love
SouthGrid: Rhys
UCL: Ben
LondonGrid: Owen
IC: Dave Collins
ScotGrid: Steve for Fraser
NorthGrid: Alessandra
Dublin: Dave

ScotGrid
========

Glasgow: frontend setup with one worker node. Also working on IBM cluster
Durham: stalled for lack of man power Fraser was there last week to help.
Edinburgh: Lcfg configured frontend but not worker node.

NorthGrid
=========

Liverpool: no progress
Manchester: Installed a second small farm 1CE 1 SE and 10 WN
for atlas DC. Keep on working to the lcfg installation of a babar farm.
Problems so far: frontend boots only from floppy had to recompile the kernel
WN boot from pxe but didn't like the defaul lcfg kernel had to recompile
that as well. Easy to reinstall a farm if babar software is on software
server and not local to the farm.
Sheffield: not present
Lancaster: network congestion problems. Looking at ganglia the network graphs
look chaotic. Might be ftp but it is difficult to understand who's provoking
the problem and how. Will keep on looking. Seen also a couple of error
messages in ganglia. lhcb and atlas have installed their software but the
system is still idle

SouthGrid
=========

Not much changed

Cambridge: ok
Birmingham: 3 weeks ago entered the test zone
RAL-PPD: ok
Warwick: out of the map but there are neither man power nor computing
resources.
Bristol: no man power no hope for a while. HP first commercial lcg2 but
no progress on it.

London
======

IC: increased cpu to 66 running in core zone lots lchb restarted mds server
a number of times for unknown reason.
UCL: (Hep) 20 nodes since a month. Second cluster with 60 nodes has been setup.
There are contradictory numbers from CC and HEP on the number of slots.
HEP counts 66 slots, CC 120 slots. Half of the slots have very low priority anyway.
QMW: ported the lcg software to the fedora. Next weel expected to rolling this
64 dual processor nodes which will become 320 cpus when airconditioning is
turned on. Fedora published in information system as OS.
RHUL: 140 nodes farm next week manual installation of the worker nodes.
London e-science center: manual installation of WN to use with
sungrid engine

David (Dublin)

Setup SE CE manually
Working on porting lcg/egee/edg to IBM and fedora and user mode linux.
There are might issues with pxe booting. Fedora already done at QMW maybe
would be useful to contact them.

It would be useful also to setup a documentation page.
SouthGrid has one where people documents can be uploaded this makes easier
maintainance same as TB documentation pages.

Tier1
=====

160 cpu online
Lhcb job stopped because of lack of space on the worker nodes
Difficult to schedule jobs: they run for a couple of days and
then have to clean up.
New kit 512 processors 2/3 added eventually to lcg and 1/3 for babar
won't be online for few weeks.

Gridpp provided front ends
=========================

Delivery in august but more possibly in september. Need of site contacts and
addresses for delivery.
Tier2 managers will provide the contacts (haven't they done it already?).
To confirm the delivery an email should be enough and paper signature
shouldn't be required.

Tender cannot be circulated for evaluation to the tier2 coordinator for
legal reason. By the 14/07/04 the evaluation material should be there.

Supporing the Data Challenges
=============================

Lhcb runnig but with problems.
Atlas installed everywhere but not yet running still in evaluation phase.

Next LCG upgrade
================

New features of the next lcg release important for the UK are:
RGMA and atlas datastore support are in the next release
SRM is still with bug fixes
BDII should be more stable
But upgrade hasn't been decided yet.

Is it a testbed zone for development foreseen in which sites can commit
just one node? Nothing like that exist. The most common case is
as the CIS testbed which goes back and forward between the two releases
using a mixed mode on production testbed.

On Mon, 28 Jun 2004, Steve Traylen wrote:

> Hi,
>
>  It is time for another a phone conference across the UK to
>  report on progress. This time members from Ireland Grid and
>  Trinity, Dublin in particular will be invited as we move into an
>  EGEE framework with the UK/I ROC.
>
>   Friday 2nd July  , 16:00
>   Minutes - Alesandra
>
>   Agenda
>
>   + New changes, developments and issues from sites.
>
>     Scotgrid (Apologies from  Fraser, will povide written account)
>       Edinburgh, Glasgow, Durham, ..
>
>     London Grid (Owen)
>       Imperial, RHUL, UCL, ...
>
>     Southgrid (Rhys)
>       Oxford, RAL-PPD, Cambridge, Bristol
>
>     NorthGrid (Alesandra)
>       Manchester, Sheffield, Liverpool,..
>
>     Ireland (John?)
>       Trinity, Cork,...
>
>     Tier1  (Steve T)
>       RAL
>
>   + Installation Problems.
>
>   + GridPP Provided Frontends. (Barry S, Dave C)
>       Status of purchase.
>       Projected arrival date.
>
>   + Supporing the Data Challenges. (Frederic, Ian ?)
>        Atlas and LHCb are curretly most active.
>
>   + Next meeting.
>   + AOB.
>
> To join the conference please dial 0871-711-1533 or
> ( 0044 870 088 5704 if outside the UK ). Listen to the instructions
> and use access code 102852.
>
> This meeting is imediatly after the UK HEP Sysman meeting at RAL
> ends, anyone there can join this meeting with me in John's old office.
>
> This agenda is posted at
> http://www.gridpp.ac.uk/tb-support/phoneconf/
>
>
>    Steve
>
>
>
>
>
>
> --
> Steve Traylen
> [log in to unmask]
> http://www.gridpp.ac.uk/
>