UKI Monthly Operations Meeting (TB-SUPPORT)

Thursday 17 Jan 2008, 10:30 → 11:30 Europe/Zurich

EVO - GridPP Deployment team meeting (under GridPP community)

Monthly review and discussion meeting for those involved with UKI EGEE/WCG deployment and operations. - The phone bridge number is +41 22 76 71400,. Th phone bridge ID is 189020 with code: 4880.

10:30 → 10:50 Site status review and issues

- Accounting -- Publishing stopped for many sites after Dec 31st due to a certificate expiry (APEL archiver). Should now be fixed but sites are encouraged to check -- This link may be useful: http://www3.egee.cesga.es/acctenfor/ -- Status on morning of 17th Jan --- Ireland: cpDIASie; csTCDie; giHECie; mpUCDie - slight --- ScotGrid: ScotGRID-Edinburgh; UKI-SCOTGRID-DURHAM - large (known) --- SouthGrid: UKI-SOUTHGRID-OX-HEP - large (known) --- NorthGrid: UKI-NORTHGRID-LIV-HEP - start Jan --- London: UKI-LT2-IC-HEP; UKI-LT2-IC-LeSC; UKI-LT2-RHUL (start Jan & known) - Monitoring status -- Q4 2007 SAM overviews are now available http://hepwww.ph.qmul.ac.uk/~lloyd/gridpp/samplots.html - comments invited shortly in the wiki on site trends (previous version was up until September http://www.gridpp.ac.uk/wiki/SAM_availability:_July-September_2007) -- Recent SAM problems (24hr/weekly) --- RHUL; UCL-CENTRAL --- Birmingham - Site concerns/issues discussion -- Edinburgh ECDF is now working. Lessons learnt about working with central facility. Major problem around job requirements from forked jobs under SGE.

10:50 → 10:55 Current experiment/VO activities and issues

- ATLAS -- Parallel meeting with this UKI one taking place in Birmingham today -- T1 problems over the Christmas period led to reduction in UK cloud activity - sites full and developed backlogs. Situation now clearing. Problems with the ATLAS CASTOR instance were traced to queries overflowing the Oracle query cache. This was increased and ATLAS production restarted last Wednesday. - CMS -- Addressing the main problem highlighted at recent T1/T2 reviews concerning software installation - LHCb -- Efforts restarted to address sites failing the LHCb SAM tests -- Migration of data from T1 dCache to CASTOR is almost complete - Regional VOs -- camont are running a new batch of jobs. They have been testing the gLite-WMS and run into the need to request longer lived proxies. Ideally they and others will soon have a mechanism to automatically renew proxies. -- No reports of major T2 VO activity. ScotGrid are testing the gLite-WMS - Other such as biomed -- UKI still provides most to biomed (approx. 27% for last month), but other regions are now scaling up and are now providing more used CPU cycles than UKI (biggest impact from ALICE).

10:55 → 11:05 Recent GDB/MB discussions

- There was a GDB last week: http://indico.cern.ch/conferenceDisplay.py?confId=20225 - Introduction: concerns about file sizes and impact on tape efficiency. Noted concern about maintaining consistency between expt. catalogues and SE content. - Benchmarking: Report from HEPiX working group. New benchmark needed for consistent description of experiment requirements, lab commitments and existing resources. Natural candidate SPEC CPU 2006. Want to cross-check how standards relate to expt. code. Machines come from volunteer sites. M Bly is UK person on team. - Data Management: Discussion on how to use storage tokens consistently (handling of T1D0 storage class). There was also a review of the GSSD mandate - develop migration from SRMv1 to SRMv2. - WN working group: Group looking at hard limits and efficient allocation within Grid. Better integrating heterogeneous clusters. https://twiki.cern.ch/twiki/bin/view/EGEE/WNWorkingGroup. Alessandra and Stephen in group. Short-term fixes proposed including splitting the CE node type and extending the ClusterPublisher. Managing VO tags was also covered in talk. - Security policies: -- Final call for comments on VO operations policy: https://edms.cern.ch/document/853968/1. -- Policy on Grid Multi-User pilot jobs: https://edmscernch/document/855383/1. -- Grid security logging and traceability policy: https://edmscernch/document/428037https://edms.cern.ch/document/428037 - Pilot jobs: Main change is setting up a team of people to review the experiment job submission frameworks. - CCRC (also had a F2F meeting on the Thursday http://indico.cern.ch/conferenceDisplay.py?confId=24844) MB meetings: - F2F Meeting on 8th covered things already mentioned under GDB http://indico.cern.ch/conferenceDisplay.py?confId=22194 - 15th: http://indico.cern.ch/conferenceDisplay.py?confId=22195. Focus was on SRM and CCRC planning. Some discussion of the December site reliability figures.

11:05 → 11:15 Preparing for CCRC'08 (GridPP sites)

- F2F meeting at CERN covered the main WLCG wide issues (see link above) - CCRC dates are 4th February to Friday 29th February and 5th May to Friday 30th May - Still need to clarify the exact services & their versions that are required at sites in production for February run. - Main UK T2 effort is in deploying latest SRM versions - Experiments have further developed their requirements (still mainly T0 and T1 for LHCb). CMS mention verifying Data Transfer and SE's at Tier-2 centers. ATLAS mention more specifics for T2s including spacetokens and suggested space/subscription (slide 13 - http://tinyurl.com/2pbldb). - Plans for monitoring and daily WLCG coordination meeting.

11:15 → 11:25 General discussion

- Deployment for WLCG in 2008 - Increasing the number of NGS affiliates - New sites/user communities - Current concerns about the way we work! - Footprints/GGUS problems to be followed up - ...

11:25 → 11:30 AOB

- Testing needed of the VO configurator tool - Register for GridPP20 http://www.gridpp.ac.uk/gridpp20/ - GridPP Tier-2 quarterly reports due this week for Q4 2007. Please help the coordinators complete these on time. - Preparations for the next WLCG workshop 21st-25th April at CERN will start sooon: http://indico.cern.ch/conferenceDisplay.py?confId=6552