SouthGrid Technical Board Meeting

Thursday 22 May 2008, 10:00 → 16:30 Europe/London

Birmingham Physics Department

Birmingham 1000-1630 Arrive 10.00 for coffee etc meeting start at 10:30

Minutes tbsupport_220508.doc tbsupport_220508.pdf

10:00 → 10:30 Arrive - Coffee

Instructions on how to get here.

In the West seminar room 115 on the first floor of the Physics
West Building located on the Edgbaston Campus and labelled with R8 on:

http://www.bham.ac.uk/documents/B1796_Edgbaston_all_zones1.pdf

10:30 → 11:45 TB Support Meeting

http://indico.cern.ch/conferenceDisplay.py?confId=34076

11:45 → 12:30 Recent Accounting Performance

11:45 Recent performance

SouthGrid wrt other T2s

11:52 Recent Performance - Our sites

SouthGrid sites performance

12:00 Any Accounting problems at sites

Correct SPEC values being published Multiple sub clusters being correctly accounted

12:30 → 13:15 lunch

13:30 → 14:00 CCRC08

Software versions required.
Problems with updates at sites?

14:00 → 15:00 Site Stautus update

15:00 → 15:30 Open Tickets and Experiment issues

Look at any Open tickets

Can we get more Atlas involvement at RAL PPD and Cambridge

Zeus/ hone/ LHCb etc

15:30 → 16:15 AOB

Minor agenda item: I need to find out from the rest of SouthGrid how they
claim RAM size in site-info.def. RAM per entire WN or RAM per CPU/Core?
Who in SouthGrid has WN with 2GB RAM/core?

I'm also interested in asking sites about stability (or not) of SL4 CE.

Find out from other T2 sites if they have gridftp servers separated out from
their SE; if they do, what kind of typical load is on them, & what are
their hardware specs.

Often sites (like Bristol's DPM & I think Oxford?) have all the SE bits
running on one machine, in Oxford's case "beefy" hardware.

Bigger sites have separate gridftp servers, & Jon W for StoRM thinks the
gridftp function should be separate. So we would like to find out some data
on what load other T2 sites see in their gridftp servers, & what kind of
hardware they have serving that load.