BlueJeans: https://bluejeans.com/426716450

attending: Eric BellmIan Sullivan Krzysztof Findeisen Unknown User (emorganson) Unknown User (cmorrison) John Swinbank Unknown User (gkovacs) John Parejko Hayden 

Topics for discussion

  • HiTS reprocessing (Meredith Rawls):
    • Meredith is on vacation.
    • DM-18687 found that the photometric calibration is probably fine, but bad subtractions make the forced photometry unreliable
    • jointly review apPipe.py configs?
      • Chris Morrison thinks going from bottom up would be painful–would be natural to start from parent configs (ISR, etc.) and work down from there.  Some ip_diffims overrides are effectively "magic", by lore from Yusra.  Schedule some of this in sprint planning?  Focus just on comparisons to obs_subaru, e.g. in processCcd?  Should this be a higher level project with DRP?  Have series of brownbags reviewing the component tasks, starting with higher-level tasks?
  • DES processing (Unknown User (emorganson)):
    • ops rehearsal next week, continuing to adjust how alert distribution works under Kubernetes, putting up Kafka for the EFD
  • ap_association (Unknown User (cmorrison)):
    • did timing tests for pandas in dax_ppdb, no degradation.  Colin may lead some benchmarking on postgres.
  • Metrics & ap_verify (Krzysztof Findeisen ):
    • no changes
  • image differencing (Unknown User (gkovacs) ):
    • show and tell for kernel solution visualization.  Seems like a small number of kernel basis functions seem to show particularly large swings in bad images, suggesting they may be particularly susceptible to errors
  • DCR (Ian Sullivan ):
    • going to send Krzysztof the review of the convergence hack
  • Review CI (https://chronograf-demo.lsst.codes/):
    • no changes, but "Total Unassociated DiaObjects" is returning no results
      • Krzysztof Findeisento investigate
        • When I converted "Total Unassociated DiaObjects" from a CCD-level to a dataset-level metric, I updated the plot query but forgot to make the same changes to the table query. – Krzysztof
    • no runtime changes
  • AOB
    • Chris M to be out the next two meetings