Summary of SHUTTLE operations during the FDR Alberto
Summary of SHUTTLE operations during the FDR Alberto, Jan Fiete CERN, January 11 th, 2008 Offline Commissioning mini-workshop Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 1
Data load: some numbers ● ● 6207 runs processed by Shuttle 1898 DA files created and copied to DAQ FXS Detector N. of files in DAQ FXS Total size (MB) % files processed Total size (processed) FMD 1 3. 560 100 3. 560 GRP 1730 1647 34 955. 0 MCH 113 325. 2 58 220. 4 TPC 54 10. 27 0 0 Total 1898 1986 34 1179 Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 2
Data load: some numbers (2) ● ● 45 MB of condition data stored into Grid (44 in OCDB + 1 in Reference) è Base folders: /alice/data/2007/LHC 07 w/[OCDB ¦ Reference] Detector Size of data in OCDB (MB) Size of data in Reference (MB) FMD 0. 14 - HMPID 3. 9 - MUON 37 - TOF 0. 03 TPC 2. 8 - ZDC - 1. 3 Total 43. 87 1. 33 892 MB of raw tags merged files stored by GRP into raw folders Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 3
Data load: some numbers (3) ● DCS datapoints used in the FDR Detector Datapoints declared Datapoints actually used Detector Datapoints declared ACO 60 not running SSD missing EMC missing TOF 10512 4 FMD missing TPC 411 HMP 429 TRD 3258 372 MCH 788 0 T 00 157 0 MTR not needed V 00 64 0 PHS 80 ZDC 28 28 CPV missing GRP 12 6 PMD 48 SPD not needed SDD 1820 Cern, January 11, 2008 not running Datapoints actually used not running 0 Offline Commissioning mini-workshop Alberto, Jan Fiete 4
Summary: SHUTTLE core ● Commissioning run gave the opportunity to test the SHUTTLE in production environment, fix a few “hidden” bugs and optimize its performance. ● Most remarkable actions taken during the two-weeks run: è Conflict using Mona. Lisa and Ali. En packages (which caused the early periodic crashes) fixed; è Self-triggering in case of no EOR signal (timeout settable from configuration) è SHUTTLE status monitored by Mona. Lisa; è Skip processing of runs with 0 events; è Raw data tags merged files written in /alice/data/…/lch. Period_DET/run. Nb/raw if partition is made of DET only; è User-friendly scripts to run and monitor the SHUTTLE in pcalishuttle 01: /home/shuttle {start¦stop¦status¦restart¦reconfig} ● Only 1 unrecoverable error è Night shift missed Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 5
Summary: ECS logbook ● Logbook worked smoothly ● ECS EOR signal not present (SHUTTLE ran in self-triggering mode) ● Open issue: SHUTTLE table should be filled at EOR (not SOR) è To be implemented by DAQ Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 6
Summary: FXS+DA ● DAQ FXS: è Frameworked smoothly; è Some DAs successfully integrated and run ― FMD, GRP, MCH, TPC è Some problems on the “run type”: most detectors did not specify the run type in the ECS framework correctly/in time. ― In these cases the preprocessors code had to be corrected in order to process the available data. ● DCS FXS: è Framework up and running, but it was not used ― CTP wrote counters data for a couple of runs but logbook was not filled correctly (wrong detector name: TRI → GRP) ● HLT FXS: è Framework up and running, but no activity detected Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 7
Summary: DCS archive + Amanda ● DCS → Amanda → SHUTTLE chain worked ● Still missing (and on the critical path): è “Bonus values”! ― To be implemented by DCS è Keep. Alive message! ― To be implemented on the SHUTTE client and tested ● Possible delays in data archival detected by TPC! è In a couple of runs (e. g. 14755) a fraction of the DPs were not yet archived when the SHUTTLE queried them (~20 mins after EOR)! ● On the detector side: è Only HMP, TPC and ZDC were ready with the requirements (all required data points è defined in the archive) Other detectors had to remove part or all the required DPs from SHUTTLE configuration Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 8
Summary: Detector preprocessors ● ACORDE ● EMCAL ● FMD ● HMPID ● MUON tracker ● MUON trigger ● PHOS (EMC+CPV) ● PMD è No preprocessor available yet! è No activity è 1 DA (Pedestal) over 2 run with success è Preprocessor (Pedestal algorithm) run with success è DCS data points retrieved and stored in OCDB è No activity from FXS (Pedestal DA did not run) è DCS archive part missing è 1 DA (Pedestal) over 3 run with success è No activity Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 9
Summary: Detector preprocessors ● ITS (SPD+SDD+SSD) ● TOF ● TPC è No runs processed successfully (work on DAs not finished in time) è Only a fraction of DCS DPs available in the archive è Physics DA not run è All required DCS data points available ― Possible delay problem detected! (see before) è Pulser DA run, but preprocessor did not take data due to wrong run type setting ● TRD è A fraction of DCS DPs available in the archive ― ● T 0 ● VZERO ● ZDC ● GRP However, due to unavailability of the “bonus values”, preprocessor always failed because it always found “empty” DPs! è No activity (work on DAs not finished in time) è No activity (DA not ready) è All required DCS data points available è No DA run with success è Not all DCS data points available è Raw data tags merged files produced and stored on the Grid è CTP data retrieval missing Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 10
SHUTTLE preprocessor status in the FDR GRP AC OR DE EMCAL HMP ID FMD ITS SPD ITS SDD ITS SSD MU TRG MU TRK PHS EMC PHS CPV PMD not part T 0 TOF TPC TRD V 0 ZDC O O Preprocessor not part Working DAQ FXS output files O DCS DPs DCS FXS output files HLT FXS output files 0 O O O O O Last Update: 09. 01. 08 = nothing to be done Cern, January 11, 2008 O Offline Commissioning mini-workshop Alberto, Jan Fiete 11
Open issues ● SHUTTLE core è Mark processed runs in the catalog, e. g. by touching a specific file (to trigger start of reconstruction): ― <some-path>/shuttle-done/<run-nr> ● ECS/DAQ è ECS EOR signal (DIM) to be implemented è Shuttle table should be filled at EOR (not SOR) ● DCS (Archive) è Implementation of Keep. Alive message in Offline client ― Actually an Offline task è Bonus values è Possible delays in data archival to be analyzed Cern, January 11, 2008 Offline Commissioning mini-workshop Alberto, Jan Fiete 12
- Slides: 12