1 / 14

Introduction

Introduction. Torre Wenaus BNL May ‘99 STAR Computing Meeting BNL May 24, 1999. Recent & Expected Arrivals. New member of the core BNL computing team: Mei-Li Chen , online developer From U Maryland (Super-K, Milagro) Very experienced online/DAQ expert

raziya
Télécharger la présentation

Introduction

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. Introduction Torre Wenaus BNL May ‘99 STAR Computing Meeting BNL May 24, 1999

  2. Recent & Expected Arrivals • New member of the core BNL computing team: • Mei-Li Chen, online developer • From U Maryland (Super-K, Milagro) • Very experienced online/DAQ expert • Has online event pool interface to DAQ working for full events over socket connection • Systems support hire (online systems; general computing infrastructure) expected soon (…?) STAR Computing Meeting 5/99

  3. Status • Infrastructure stabilized! • The train has pulled into the station, so people can catch up! • Core group now in ‘consolidation’ mode and focusing on commissioning run essentials • Bug fixing, documentation, examples, help with online integration, online monitoring needs, developing and responding to QA • Online system pre-production release • Run control; subsystem control,configuration, slow control • Essential components of the system now in place • Near term production plan • StMcEvent tests, peripheral signal and backgd, strangeness AuAu steady production, spectra AuAu • Ongoing production with simu data to • meet PWG needs; exercise and debug codes; perform QA; keep CRS well oiled (CRS became functional again Friday) • of course second priority to real data when it comes! STAR Computing Meeting 5/99

  4. Working with StEvent: doEvents.C • doEvents.C • Standard means of reading DST files and feeding them to StEvent based analysis • Has handled XDF files OK, but hasn’t handled ROOT files properly up to now • Needs to transparently handle all filetypes; shield distinctions from user. Victor now has this implemented. • On Linux, all filetypes (MDC1 XDF, MDC2 XDF, MDC2 ROOT, post-MDC2 ROOT) work transparently. Still a problem on Solaris that Victor is working on • Multiple files handled STAR Computing Meeting 5/99

  5. Using Persistent StEvent • StAnalysisMaker changes: • Remove ‘title’ (second parameter) from constructor • #include StEvent/XXX.hh -> XXX.h • Change method of event pointer retrieval: • Former: • StEventReaderMaker* evMaker = (StEventReaderMaker*) gStChain->Maker("events"); • if (! evMaker->event()) return kStOK; // If no event, we're done • StEvent& ev = *(evMaker->event()); • New: • StEvent* mEvent = (StEvent *) GetInputDS("StEvent"); • if (! mEvent) return kStOK; // If no event, we're done • StEvent& ev = *mEvent; STAR Computing Meeting 5/99

  6. This Meeting • Joint online/offline focusing on real data readiness and the immediate needs of the commissioning run • Data flow from DAQ to online, offline • Online system status • Processing of DAQ data in offline • Event I/O, data cataloguing • Conditions database • Online monitoring • Subsystem and global reconstruction for year 1 detectors • RCF and production • Current offline software in production, with people doing analysis QA on the results • In this and subsequent weeks try to reproduce some of the very productive environment of MDC2 STAR Computing Meeting 5/99

  7. Goals for Commissioning run environment • - stable offline environment as seen by users • - Maker structure and interface stable. Changes affecting the user • beyond those already made post-MDC2 are not foreseen. • - StEvent as used by analysis codes has been stable; some functionality • additions • - adding persistency to StEvent while keeping it unchanged as seen • by analysis codes has (subject to wider testing) been successful • and so can serve as a basis for uDST development and StEvent • extension to reconstruction (something we hoped to achieve but • did not promise) • - needs discussion at this meeting • - macros/Makers provided to transparently handle all data file • types (XDF, MDC2 root files, new root files) STAR Computing Meeting 5/99

  8. Commissioning environment (2) • - capability to process raw DAQ data • - stably functional data storage • - post-MDC2 ROOT I/O improvements implemented in STAR address • deficiencies in standard ROOT I/O: robustness against partially • completed jobs, forward and backward read compatibility for • data files, robust management of multiple event components on • different files • - schema evolution approach of Victor has to be (if at all possible) • fitted into overall ROOT I/O; long discussions hopefully converging • - capabilities satisfy commissioning and year 1 needs • - populated conditions database in use via stable interface STAR Computing Meeting 5/99

  9. RCF Issues • CRS working again, as of yesterday, after long absence • Manpower: supposed to plateau at 34 in ‘99. Actual: 23 (1 offer is out) • CCD: 4 of 8; Experiments: 4 of 8 • MDC2 experience • HPSS: similar level of problems to MDC1 • Stabler after fixes towards end of MDC, but still we see instability • Latest version (4.1) installed; stress testing today. • Disk: major access problems. NFS, AFS translator (Linux) • Move to real AFS 3.5; supported on current Linux • New CRS mgmt software: Major rewrite • Better monitoring and user interaction, improved HPSS interaction; worked well. Adequate for year 1. LSF not needed in CRS. • STAR program size! • Linux 2.2 kernel installed; swap increased to ~2GB • STAR CRS machines will all be 512MB (all CAS machines are 512) STAR Computing Meeting 5/99

  10. RCF Issues (2) • STAR Executable size 512MB with all makers (tfs), 587MB (trs) • Security issues • RCF switched off STAR machine access to RCF disks for new machines • Switching increasingly to AFS • Have partial agreement with RCF to allow us access to data disks; needs to be made official by RCF and implemented • Procurements • New Linux (CRS,CAS), Sun (CAS), disk (all RAID), and MDS (drives, processors) procurements underway • ADSM batch system June 8 • Validate migrated OSM data by then • Will need people with data in OSM to do spot checks • Details due from RCF STAR Computing Meeting 5/99

  11. Post Commissioning • - extending data management tools (mysql DB + disk file management • + HPSS file management + multi-component ROOT files) • - complete schema evolution, in collaboration with ROOT team • - management of CAS processing and data distribution both for mining • and individual physicist level analysis • - completion of the DB: integration of slow control as data source, • completion of online integration, implementation of DB as Geant • geometry data source, populating the DB for detectors other than • the TPC • - extend and apply OO data model (StEvent) to reconstruction, so • new reconstruction development can work with StEvent • - completion of online and the online processing farm for monitoring • - improving QA tools and procedures • - improving and better integrating visualization tools STAR Computing Meeting 5/99

  12. Post Commissioning (2) • - reconstruction and analysis code development • directed at addressing QA results and year 1 code completeness • and quality • following the priorities emerging from this meeting • responding to the real-data outcomes of the commissioning run STAR Computing Meeting 5/99

  13. New Package Creation • It is important to check in your code; submit your Makers as packages in the repository • When bugs or changes affect existing code, we can identify the affected code • Code sharing, collaboration, example for others, integration into QA, etc. etc. • When HyperNews email is working again, will set up a forum for requesting and creating new packages • Awareness and consultation on new package proposals STAR Computing Meeting 5/99

  14. Tutorials • Wed (and Thurs?) am: presentation/discussion sessions as a group • Followed by individuals or workgroups trying things out, with core computing people and experienced users sitting in and roaming the halls to help • Wed am: • Kathy on bfcread.C, histogram integration, general usage • Writing analysis code and using StEvent • Gene (and others..? Torre? Victor? Thomas? Craig Ogilvie?) • Wed or Thurs am: • Pavel on using GSTAR • Fri am: • Thomas on using UML and Rational Rose • Fri pm: • Issues that come up in the week? STAR Computing Meeting 5/99

More Related