LT2 Management Board Meeting II: 9th March 2005 Present D. Colling (Chair) O. Maroney (Secretary) S. McGough (Imperial College London) A. Martin (QMUL) M. Green (RHUL) B. Waugh (UCL) Apologies Paul Kyberd (Brunel) 1. Minutes and actions from previous meeting. Actions. 1.1 (AM) to ensure DC receives signed copy of MoU from QMUL. Completed. 1.2 (DC) to ensure future meetings have a dial-in option. Ongoing. 1.3 (All) compile list of VOs to be supported by LT2. All institutes have agreed to support all 4 LHC VOs. Still need to complete list of other VOs (HEP and non-HEP). 1.4 (OM) to provide CPU and disk utilisation information from LCG monitoring software, when it becomes available. Ongoing. See (4). 1.5 (OM) to provide figures on funded and unfunded effort to future meetings. Ongoing. See (3). 1.6 (DC) to ask LeSC to create an LT2MB mailing list. Done, but no-one subscribed. *Action* DC to subscribe members of MB to mailing list. 2. Administrative issues. DC has suggested experiment representatives, within LT2, be invited to future LT2 MB meetings (in a non-voting capacity) to discuss expected use of grid resources. 1 representative per LHC VO (excluding Alice) and other as appropriate. If no representative is available in LT2 sites, the Technical Co-ordinator to represent that VO. MG asked if representatives could submit a report, possibly as part of item (4). This item could be at a higher position in the agenda so experiment representatives could leave early. OM will be leaving his post as technical co-ordinator in July. IC to recruit replacement. 3. Technical Report. OM distributed tables on level of funded and unfunded effort to support LT2 sites in the last two quarters. Approximately 2.5 FTE seems required (not including the technical co-ordinator) just to maintain a running service at all sites. During periods where an upgrade is required this rises by 1 FTE. It is expected that the 2.5 FTE GridPP funded posts would largely replace existing unfunded effort. OM noted that LeSC resources were still not online and that this implied extra manpower requirements. OM reported there had been criticisms raised of GridPP in general, but LT2 in particular, over the slow speed of sites upgrades to the latest LCG version. There had been some discussion in the UK Deployment Team as to how that could be improved, and had led to a proposal, adopted by LCG, that releases should in future be scheduled, and take place on a quarterly. Sites would have 3 weeks to upgrade after a fixed release date. BW questioned whether LT2 sites should attempt to stagger upgrades. OM said there were no plans across LCG as a whole to stagger upgrades and that the experiments had requested a fast upgrade, so accepted there would be a period during which the grid as a whole was not in a stable state. SM noted dedicated clusters would be likely to upgrading first. The next upgrade was expected to be simple. Two major changes would be included: VOMS and SRM. However, VOMS middleware would not be configured by default and SRM SEs would be released but sites would not be required to immediately migrate to these. There is a new ticketing system ("CIC-on-duty") developed by RAL and the ROCs for tracking problems at sites. *Action* OM to report on tickets raised on LT2 sites at future meetings. 4. Experiment Support. OM distributed figures from APEL accounting on usage of LT2 resources, but noted that only UCL-CCC and Imperial farms were actually reporting data. This should change in future. The figures showed a high usage by LHCb. This was in part due to the aggresive scheduling policy used by LHCbs jobs submission framework, but mostly due to the absence of use of LCG by other VOs in that time period. It was requested that in future these figures could also be presented as a % utilisation of the total available resources. *Action* OM to provide % utilisation figures. DC noted it had also been suggested there may be a role for experiment representatives in LT2 Technical Support Group meetings. Experiment shares for the next quarter were agreed to be the same as the last quarter. 5. Service Challenge. DC outlined the developments of the LCG Service Challenges. So far this has involved only T0 and T1s. SC3 will start to involve T2s. LCG have asked for 2 T2 sites to volunteer to be the initial sites. These may both be in the UK and Imperial have volunteered to be one. Within 1 year all T2 sites are expected to be participating in the service challenges. SC3 involves a data transfer test at half the expected rate for a single experiment. (CMS for Imperial, Atlas for Lancaster). Under the MoU sites have committed to provide a level of network bandwidth. CMS requires 1Gb/s: this is currently met by all LT2 sites except Brunel and RHUL, who only have 100Mb/s connections to the London MAN. AM noted LMAN only promises 100Mb/s as a baselink and itself only has 2.5Gb/s capacity. QMUL paid for Gb/s connections speeds from SRIF funding. MB wished for clarification of what exactly is required in the course of the next year. DC to forwared experiments requirements. *Action* DC to circulate experiments requirements. SM suggested LT2 sites could start by stress testing internal T2 to T2 network connections. DC said SC involvement must be a high priority for EGEE and GridPP funded posts. OM added that there was a UK specialist support post in networking that would also be involved through the SC Taskforce. 6. GridPP Support Posts Current status: UCL: Mark Dixon 0.25 FTE IC: Readvertised and made an offer this afternoon for 1FTE post. Brunel/RHUL: Have 0.5 FTE from GridPP and 0.5FTE matching funding for joint post. MG said advert is in HR department for approval. QMUL: 0.75 FTE from GridPP with 0.25 FTE matching funding . Advert is with HR for approval. OM noted that IC's experience of needed to readvertise was similar to other sites in the UK. It is a good idea to interview quickly and appoint quickly or good candidates are likely to accept other posts. 7. AOB MG SRIF funding has a clause where joint bids would avoid needing to find 10% matching funds internally. LT2 could provide a good basis for such co-operation. Brunel and RHUL were considering a joint bid. BW noted UCL had managed to find a local solution to avoid this. Actions. 1.2 (DC) to ensure future meetings have a dial-in option. 1.3 (All) compile list of non-LHC VOs to be supported by LT2. 1.4 (OM) to provide CPU and disk utilisation information from LCG monitoring software, when it becomes available, including as a % of potential usage. 1.5 (OM) to provide figures on funded and unfunded effort to future meetings. 2.1 (DC) to subscribe members of MB to mailing list. 2.2 (OM) to report on CIC-on-duty tickets raised on LT2 sites. 2.3 (DC )to circulate experiments requirements.