Attendance: Stuart Kenny (Trinity), Graeme Stewart (Glasgow, Minutes), Jeremy Coles (RAL, Chair), Stephen Burke (RAL), Fraser Speirs (Glasgow), Jens Jensen (RAL), Pete Gronbech (Oxford), Peter Love (Lancaster), Rosario Esposito (Oxford), Simon George (RHUL), Matt Doige (Lancaster), Yves Coppens (Birmingham), Chris Brew (RAL PPD), John Grodon (RAL), Alessandra Forti (Manchester), Barney Garrett (Edinburgh), Gianfranco & William (UCL HEP/CCC), Greig Cowan (Edinburgh), Duncan Rand (RHUL/Brunel), Peter Clarke (Edinburgh), Steve Traylen (RAL), Giuseppe Mazza (QMUL), Grigori Rybkine (RHUL) Appologies: Olivier van der Aa, Mona Aggarwal (IC-HEP) SC4/Transfer Tests ------------------ Dcache-DPM interaction resolved. 8 sites were able to transfer TB. Sites involved have had a lot of success in finding out things on the network. (Liv, Dur 100Mb bottlenecks.) Other sites at known 100Mb (RHUL, Durham) Optimisation of T2 write is tricky - Glasgow have done work on DPM - use CERN kernel and xfs is the message. Need to get more sites involved to start this. Using 2.1 SRM? No - not yet. Tools not avaliable. gLite 1.5 release? GridPP wiki works well. Documentation is scattered - knowing the right person to ask is tricky. Planning of tests & releases are too separated. Specific problem with 2.6 UI did cause problems with transfer tests. LCG 2.7.0 Release ----------------- Yves - problems with DPM databases. Essential to take a db backup and practice a restore! Information publishing is still problematic - needs to be improved. SFT results don't seem to be availiable, which doesn't help debug problems. BDII crashes and needs restarted. DPM and LFC YAIM has improved. dCache YAIM in 2.7.0 much the same as before. Positive experience with YAIM VO tool. Very easy to use, but watch capitalisation. Sometimes information is not complete. VOs need to provide more information. Tarball install for WN awaited. New VO accounts. Have "VOprd" accounts for production use. Idea is that these users will have greater priority in batch system. Nikhev have used distinct mappings for sgm accounts. Upgrade Schedule: Scotgrid: Glasgow 22-24 Feb, Durham this week, Ed when power comes back. NorthGrid: Lancaster tomorrow. Manchester end of month. Sheffield/Liv? SouthGrid: Birmingham done. Oxford next week. Cam , RAL PPD tomorrow. LT2: RHUL this week. Brunel ?. UCL-HEP 20 Feb (tbc). UCL-CCC merging clusters right now. Supporting New VOs ------------------ T2B have requested extra VOs to be enabled. Usage across GridPP is quite low. Enabling more VOs helps. Some information is missing from the wiki. Use the YAIM configurator or look on CIC portal. Sites just need the "vital statistics". RAL T1 support virtually all these VOs, so information is known. "The truth is out there." Number of VO accounts has increased. Status of VOMS in UKI Region ---------------------------- VOMS is in production at Manchester. New VOs are being created. Chris Brew wanted a local VO supported. GridPP will host these VOs (0.5 FTE funding at Manchester). Improving Interaction with Experiments -------------------------------------- User Board has been allocating resources at T1. Need to get more use of T2s - particularly disk. IC-HEP have been moving CMS data on to their SE. Lancaster - Atlas. Edinburgh - LHCb (but do they want to use T2s at all?). Storage group have been looking as publishing "quality" of storage into the BDII to help VOs decide on how to use T2 disk. AOB --- LHC OPN meeting last Monday (Amsterdam). Mostly discuss T0-T1. No specific plans to connect T2s to OPN. In principle plenty of bandwidth on SJ5. On a case by base basis any T2 can investigate connection to OPN - discuss with Pete Clarke and Robin Tasker to help make request to UKERNA. But central planning is needed! Will ask Mark Leese or Robin Tasker to attend next meeting for network monitoring box status. Ad Hoc discussions about 100Mb links.e.g., Brunel have problem because of MAN connection. UKENRA are responsible to the universities. Need to get university and network people on your side! UK BDII: new DNS alias for GridPP BDII, lcg-bdii.gridpp.ac.uk. All sites need to reconfigure to use this as RAL have a scheduled power outage. APEL Accounting: Seems to be fixed now.