September 19, 2014 Data Monitoring

From GlueXWiki
Revision as of 15:34, 19 September 2014 by Pmatt (Talk | contribs) (Action Items)

Jump to: navigation, search

Online Data Monitoring

What we want to know

  • Are the detectors working?
    • Do all of the channels have hits?
    • What are the hit counts/rates per channel?
    • Are the energies & times OK, garbage, or out of range? (by channel)
    • Is the reconstruction working?: Run full reconstruction over a small % of data, view tracking/shower-reconstruction results online (hdview2, monitoring_hists plugins).

Example

Action Items

  • David is updating RootSpy and the documentation.
    • When he's finished, he'll contact the different detector groups and remind them to write their online monitoring plugins.
  • We (Paul?, David?, Sean?) will stay in contact with the detector groups, and make sure that they:
    • Write the plugins
    • Are histogramming the data we want (see above).
    • Integrate them into RootSpy, and determine which are the primary plots, the ones that are the most important for shift-takers.
  • Paul will make sure that the monitoring_hists plugin plots (for reconstruction quality) are integrated and viewable in RootSpy.

Offline Data Monitoring

What we want to know

  • Can we reproduce the online histograms offline:
    • Can we read data from tape?
    • Are all the channels present?
    • Are all the recorded values within the expected ranges?
  • How do the online histograms change as a function of time (e.g. hits/event vs run number)?
  • Calibration status/quality histograms (individual runs & a function of time)
  • Reconstruction status/quality histograms
    • Num tracks, showers, hits, etc.
    • Reconstructed track kinematics
    • Track / detector-hit matching
    • PID Quality
  • Analysis histograms
    • Can we see pi0 peaks
    • Can we see simple final states
      • proton pi+ pi-
      • proton pi+ pi- pi0
      • proton pi+ pi- eta

Action Items

  • Make sure that the detector groups make plugins for doing the calibrations and viewing the quality/results.
  • Make cron-job scripts that automatically read data from disk and make histograms of the above.
  • Make cron-job scripts that
  • Make cron-job scripts for visualizing the data:
    • Make it easy to view today's & yesterday's online data (webpage?)
    • Make it easy to view data from any given run (webpage?)
    • Make it easy to view the status of things as a function of time (webpage?)

Links