1 / 19

WLCG LHCC mini-review LHCb Summary

WLCG LHCC mini-review LHCb Summary. Outline. Activities in 2008: summary Status of DIRAC Activities in 2009: outlook Resources in 2009-10. Tier1s (re-)configuration. LFC mirror, ConditionsDB replication DB replication using 3D from CERN to all Tier1s In place for the whole year

buffy
Download Presentation

WLCG LHCC mini-review LHCb Summary

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. WLCG LHCC mini-reviewLHCb Summary

  2. Outline • Activities in 2008: summary • Status of DIRAC • Activities in 2009: outlook • Resources in 2009-10 PhC

  3. Tier1s (re-)configuration • LFC mirror, ConditionsDB replication • DB replication using 3D from CERN to all Tier1s • In place for the whole year • LFC service for scalability and redundancy • Some problems at GridKa • Site SE migration (winter 08-09) • RAL (dCache to Castor2) • T0D1 migration went rather smoothly (FTS copy of files) • T1D0 migration extremely painful (staging tape by tape) • Took several months • PIC (Castor1 to dCache for T1D0) • Went very smoothly without file copy (file migration to Enstore) • CNAF (Castor2 to StoRM for TxD1) • Performed in May • SURLs migrated to SRM v2.2 end-points (October) • Needs as datasets in use for 2.5 years now (DC06) • No dependency on SRM v1 WLCG-MB report, February 08

  4. Pit 8 to Tier0-Castor transfers • First weeks in February: continuous transfers at low rate • As of 18 Feb: nominal rate (70 MB/s) with ~50% duty cycle • A few longer stops for SW upgrades Tier1 Transfers Migration WLCG-MB report, February 08

  5. Castor migration WLCG-MB report, February 08

  6. May CCRC: reconstruction 41.2k reconstruction jobs submitted 27.6k jobs proceeded to done state Done/created ~67%

  7. Cosmics and first beam data • Cosmics used for detector commissioning • Of course very few detectors are hit! • Allows internal time alignment of subdetectors • Using readout of consecutive 25ns slots • Partially time alignment between detectors • Shifting timing by 12.5 ns and equalising population in consecutive bins… • All subdetectors included in global runs as of end August • TED data • LHCb was first to see tracks coming from the injection line! • Single shots with ~2 muons/cm2, but once every 48s only! • First tracks in the VeLo • Allowed rough global detector time alignment (~2ns) • 10th September • Only muons, calorimeters and for short time OT PhC

  8. First data PhC

  9. DIRAC3 put in production • Production activities • Started in July • Simulation, reconstruction, stripping • Includes file distribution strategy, failover mechanism • File access using local access protocol (rootd, rfio, (gsi)dcap, xrootd) • Commissioned alternative method: copy to local disk • Drawback: non-guaranteed space, less CPU efficiency, additional network traffic (possibly copied from remote site) • Failover using VOBOXes • File transfers (delegated to FTS) • LFC registration • Internal DIRAC operations (bookkeeping, job monitoring…) • Analysis • Started in September • Ganga available for DIRAC3 in November • DIRAC2 de-commissioned on January 12th PhC

  10. DIRAC3 jobs PhC

  11. Issues in 2008 • Data Management • Site configuration (non-scaling) • SRM v2.2 still not fully mature (e.g. pinning) • Many issues with StorageWare (mainly dCache) • Workload Management • Moved to gLite WMS, but still many issues with it (e.g. mix-up of identities). Better scaling behavior though than LCG-RB • LHCb moved to using “generic pilot jobs” (i.e. can execute workload from any user or production) • Not switching identity yet (gLexec / SCAS not available) • Not a show-stopper as not required by LHCb but by sites • Middleware deployment • LHCb distributes the client middleware • From distribution in the LCG-AA • Necessary to ensure bug fixes to be available • Allows multiple platform (OS, architecture, python version) PhC

  12. LHCb Computing Operations • Production manager • Schedules production work, sets up and checks workflows, reports to LHCb operations • Computing shifters • Computing Operations shifter (pool of ~12 shifters) • Covers 14h/day, 7 days / week • Computing Control room (2-R-014) • Data Quality shifter • Covers 8h/day, 7 days / week • Both are in the LHCb Computing Control room (2-R-014) • Daily DQ and Operations meetings • Week days (twice a week during shutdowns) • Grid Expert on-call • On duty for a week • Runs the operations meetings • Grid Team (~6 FTEs needed, ~2 missing) • Shared responsibilities (WMS, DMS, SAM, Bookkeeping…) PhC

  13. Plans for 2009 • Commissioning for 2009-10 data taking (FEST’09) • See next slides • Simulation • Replacing DC06 datasets • Signal and background samples (~300 Mevts) • Minimum bias for L0 and HLT commissioning (~100 Mevts) • Used for CP-violation performance studies • Nominal LHC settings (7 TeV, 25 ns, 2 1032 cm-2s-1) • Tuning stripping and HLT for 2010 • 4/5 TeV, 50 ns (no spillover), 1032 cm-1s-1 • Benchmark channels for first physics studies • Bµµ, Γs, BDh, BsJ/ψϕ, BK*µµ … • Large minimum bias samples (~ 1mn of LHC running) • Stripping performance required: ~ 50 Hz for benchmark channels • Tune HLT: efficiency vs retention, optimisation • Preparation for very first physics • 2 TeV, low luminosity • Large minimum bias sample (part used for FEST’09) PhC

  14. FEST’09 • Aim • Replace the non-existing 2008 beam data with MC • Points to be tested • L0 (Hardware trigger) strategy • Emulated in software • HLT strategy • First data (loose trigger) • Higher lumi/energy data (b-physics trigger) • Online detector monitoring • Based on event selection from HLT e.g. J/Psi events • Automatic detector problems detection • Data streaming • Physics stream (all triggers) and calibration stream (subset of triggers, typically 5 Hz) • Alignment and calibration loop • Trigger re-alignment • Run alignment processes • Validate new alignment (based on calibration stream) PhC

  15. FEST’09 runs • FEST activity • Define running conditions (rate, HLT version + config) • Start runs from the Control System • Events are injected and follow the normal path • Files export to Tier0 and distribution to Tier1s • Automatic reconstruction jobs at CERN and Tier1s • Commission Data Quality green-light • Short test periods • Typically full week, ½ to 1 day every week for tests • Depending on results, take a few weeks interval for fixing problems • Vary conditions • L0 parameters • Event rates • HLT parameters • Trigger calibration and alignment loop PhC

  16. Resources (very preliminary) • Consider 2009-10 as a whole (new LHC schedule) • Real data • Split year in two parts: • 0.5 106s at low lumi– LHC-phase1 • 3106s at higher lumi (1 1032) – LHC phase2 • Trigger rate independent on lumi and energy: 2 kHz • Simulation: 2 109events (nominal year) over 2 years • New assumptions for (re-)processing and analysis • More re-processings during LHC-phase1 • Add calibration checks (doneat CERN) • Envision more analysis at CERN with first data • Increase from 25% (TDR) to 50% (phase1) and 35% (phase2) • Include SW development and testing (LXBATCH) • Adjust event sizes and CPU needs to current estimates • Important effort to reduce data size (packed format for rDST, DST, µDST…) • Use new HEP-SPEC06 benchmarking PhC

  17. Resources (cont’d) • CERN usage • Tier0: • Real data recording, export to Tier1s • First pass reconstruction of ~85% of raw data • Reprocessing (in future foresee to use also the Online HLT farm) • CAF (“Calibration and Alignment Facility”) • Dedicated LXBATCH resources • Detector studies, alignment and calibration • CAF (“CERN Analysis Facility”) • Part of Grid distributed analysis facilities (estimate 40% in 2009-10) • Histograms and interactive analysis (lxplus, desk/lap-tops) • Tier1 usage • Reconstruction • First pass during data taking, reprocessing • Analysis facilities • Grid distributed analysis • Local storage for users’ data (LHCb_USER SRM space) PhC

  18. Conclusions • 2008 • CCRC very useful for LHCb (although irrelevant to be simultaneous due to low throughput) • DIRAC3 fully commissioned • Production in July • Analysis in November • As of now, called DIRAC • Last processing on DC06 • Analysis will continue in 2009 • Commission simulation and reconstruction for real data • 2009-10 • Large simulation requests for replacing DC06, preparing 2009-10 • FEST’09: ~1 week a month and 1 day a week • Resource requirements being prepared for WLCG workshop in March and C-RRB in April PhC

More Related