Attendees

Fritz Mueller Andy Salnikov Michelle Gower Unknown User (mbutler)Unknown User (cs2018) Unknown User (cpond) Jim Bosch Tim Jenness John Parejko Nate Lust

Status

  • Tim:
    • just getting back to it.  Doing reviews for Jim.
  • Jim:
    • Confluence page posted re. prototype work.  Feedback welcome.  Beginning to ticket implementation.
  • Nate:
    • Small tickets, doc merges etc.  Trying to wrap these up.
    • Soon to start working in support of jointcal
  • Chris W.:
    • just back from vacation, working on implementing review tickets
  • John P:
    • working on write-up re. ingesting calibrations
  • Andy:
    • Time literal ticket.  Discussion w/ K-T and RFC now posted ( RFC-648 - Getting issue details... STATUS ).  Continue discussion there.
    • Writing metadata from tasks: understand how to do this now, will work up this week.
  • Chris S.:
    • starting to look through Jim's new architectural changes doc
    • misc work on NCSA Oracle env.
  • MG:
    • mostly busy with NCSA planning week
    • trying to run new stuff from Chris Waters; iterating with Chris, working around hardcoded paths etc
  • Michelle Butler:
    • Dec 12 is maintenance day -- 6am to 10am project time.  Includes file systems this time.

Dec 18 Demo

  • Dedicate page here: Gen3 Milestone FY20xxx-y
  • Contents:
    • Nate to cover pipeline YAML
    • Chris W. to cover current state of calibration pipeline
    • MG to cover having run this with BPS on Oracle
    • Next steps
  • Blockers:
    • For Chris W: focus on cleanup rather than Chris running on Oracle.  Finish up reviews.
    • For MG: adapting to any late-breaking changes and finding time to work on it.
  • Outline slides for next week
  • Doodle poll for demo time

Misc

  • There was some discussion on DMS-REQ-0131.  This requirement tries to constrain calibration products production by stating that it should take no more than 20 minutes to process 25 calibration exposures.
  • AP team is digging in on middleware now – should they use same Slack channel?
    • Recommend: start using same channel; endeavor to thread responses
  • MG: empty directories?  Leave them around?
    • Tim: if random incoming "puts", probably easiest?
    • How much will this happen? (Should be rare.) Will job scratch repos make this less of an issue?
  • MG: raw IDs the same no matter who does ingest?
    • Jim: can do if it helps DBB sync etc.?  What was the motivating use case?
  • There was some discussion of Gen 3 OODS and ingestion. Firstly it was suggested that the "group ID" generated during observing could be treated as an exposure parameter and not a "visit" identifier.  This would prevent confusion over the term "visit" and allow OODS to ingest each exposure incrementally without having to know how many exposures are in the "visit".  Secondly there was an issue over whether OODS is required to ingest all 189 CCD files in one ingest or if it can do that ingest incrementally as files are picked up. It probably makes sense to ingest in one go given that an "ingest complete" event is required on completion.
  • No labels