Discussion items

  • AAIM: federated A&A will be rolled out next week
    • still waiting for audit component (performs annual check on all identities)
    • web app for registering identities all set
    • expecting documentation at rollout
    • Kian-Tat Lim : there are two parts, right? Github linkage
  • All 2018 acquisitions are complete
    • 2nd QServ master hardware handed over a week or two ago
    • Fritz Mueller & Igor Gaponenko : no problems with the node so far, have already installed QServ and tested that the 2nd node successfully functions as a master. Everything seems to be configured correctly.
      • Current version of QServ doesn't support actually running multiple masters simultaneously. Primarily a matter of configuration.
      • The plan is to run multiple masters as a load-sharing cluster, not just as a hot spare. Working out the details of how to manage any shared state, such as query history.
  • 2019 acquisition list is now in Wil O'Mullane's hands; some cost-limitation exercises are in progress. Fritz Mueller : it may not be possible to do a full refresh of the PDAC QServ cluster. Will provide Wil O'Mullane the needed details later today.

SLACFritz Mueller
  • Google proof-of-concept work (Vaikunth Thukral) progressing well. Part of the WISE dataset on QServ in Google Cloud. Their "quasi-local" storage seems to be working quite well. Would like Gregory Dubois-Felsmann to do some behavior/performance evaluation.
  • Christine Banek's CADC TAP service evaluation: up and running, will demo at workshop, apparently have Gaia data loaded. Concerns about how we would integrate LSST A&A, handle multiple database backends. Their original authorization was based on X.509, which we can't directly adopt.
  • Short discussion of image service.
  • Kenny Lo working on TAP_SCHEMA.
  • WebDAV situation: blocked by A&A concerns. Could set up a standard Apache WebDAV. Discussion of the priorities. Probably not worth it, should focus on a proper A&A design.
    • Brian has an idea for how to set up an nginx-based WebDAV, may be easier than the xrootd approach we were previously considering.
  • Discussion of shared workspace

  • Would like to give demos of recent JupyterLab extension work, as well as the (detection) Footprint viewer. Discussion of reclaiming a bit of the next-to-DB time in the workshop for this.

  • Continuing experiments with Dask. "Promising but confusing."
  • Discussion of the role of and the need for test and development platforms.

Action items