Author: Gaspar, C.
Paper Title Page
THDPL01 Configuring and Automating an LHC Experiment for Faster and Better Physics Output 1233
  • C. Gaspar, R. Aaij, F. Alessio, J. Barbosa, L.G. Cardoso, M. Frank, B. Jost, N. Neufeld, R. Schwemmer
    CERN, Geneva, Switzerland
  LHCb has introduced a novel online detector alignment and calibration for LHC Run II. This strategy allows for better trigger efficiency, better data quality and direct physics analysis at the trigger output. This implies: running a first High Level Trigger (HLT) pass synchronously with data taking and buffering locally its output; use the data collected at the beginning of the fill, or on a run-by-run basis, to determine the new alignment and calibration constants; run a second HLT pass on the buffered data using the new constants. Operationally, it represented a challenge: it required running different activities concurrently in the farm, starting at different times and load balanced depending on the LHC state. However, these activities are now an integral part of LHCb's dataflow, seamlessly integrated in the Experiment Control System and completely automated under the supervision of LHCb's 'Big Brother'. In total, around 60000 tasks run in the ~1600 nodes of the farm. Load balancing of tasks between activities takes less than 1 second. The mechanisms for configuring, scheduling and synchronizing different activities on the farm and in the experiment in general will be discussed.  
video icon Talk as video stream:  
slides icon Slides THDPL01 [3.600 MB]  
DOI • reference for this paper ※  
Export • reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)