The US-based qserv team keeps meeting every Thursday these days,
and we thought it might be useful if we make the meeting notes
available (there is an additional early morning phone call with
the IN2P3 qserv team on Thursdays too)


Attendees: Serge, Daniel, Douglas, Jacek

trac milestones
  - now we have:
    1) qserv-beta1 (beta1 means here "ready for user testing")
       --> maybe rename to qserv-USR2013
    2) qserv-fdr (things we have to do to be ready for FDR)
  --> add qserv-LS2013  (things we have to do before large
      scale test we plan to run this summer)


status from Serge
  - working on duplicator
    - figured method to calculate region intersections.
    - can be generalized for regions with holes etc.
    - for now, document in code, make more official if
      needed more widely in more places
  - partitioning will be ready end of next week
  - need to make csv loader somewhat parametrizable,
    eg, now end of line hardcoded as "\n"
     - but don't want to re-implement all that mysql
       does ("optionally enclosed by" etc...)

status from Douglas
  - partitioned W13, almost done loading
  - loading W13_1K data set
  - setting things on yilis cluster for concurrently
    testing
     --> revisit, we need ~10K chunks to reproduce
         the problem
  --> send W13 data set to in2p3 (csv files)
    they said "Could you please put it here:
    /sps/lsst/Qserv/data/Winter2013/"
  --> document hassles with W13/S12 data sets


status from Daniel
  - working on export paths
  - helping Bill, building some framework for his ticket
    (error passing)
- would be useful to have account for Daniel at in2p3
- need to sync with AndyH re xrd client
   --> Jacek should ping Daniel later this week


status from Jacek
  - starting to work on integrating metadata with qserv
  - was tied with xldb (org committee, agenda, docs for
    consortium etc)


Would be nice to load data from pipelines directly
to qserv
  - qserv could take it, does not need to see all
    data for given table
  - might need to buffer, don't want 3K rows every second,
    better to get larger set every minute or so, that is easy
  - bigger issue: apps code does not produce all columns
    we need (eg ra/decl for ForcedSources not there), so
    we need to post-process data they produce
  - would be nice if pipelines code would do that
     --> talk to KT about it

Large qserv meetings
  - say every 1st Thursday of each month, with KT,
    (maybe others too like Mario if they want to attend)
  - perhaps invite IN2P3 team eg every 2 months, monthly
    could be a burden because of time different
    - we think noon pacific / 8 or 9 pm French time
      would be acceptable for them, but need to check

Jacek

########################################################################
Use REPLY-ALL to reply to list

To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1