Attendees: Daniel, Serge, Bill, Douglas, Jacek
integrating time series analysis into qserv etc
https://dev.lsstcorp.org/trac/wiki/dbTimeSeries
- need separate system alongside qserv to run
non-sql "queries", indexes will be in specialized
format
- consider using hadoop
- need to think through how to integrate w/qserv,
eg for queries that need specialized time series
analysis that are based on object ids or have
spatial constraints
- could be done in user code,
--> need dedicated meeting to discuss details
reworking scripts that partition csv input data
- want to throw arbitrary number of resources at
partitioning
- implemented such script, almost finished
- script is using local mysqld instance on each
batch machine as a "scratch" space
- almost done with (re-)partitioning full W13
- also packed the data using myisampack
--> prepare space for dumping qserv-ready
on lsst10 [Jacek]
concurrency test on yilis
- Daniel to email queries to Serge
Bill
- tried to simulate different error conditions
- ready to merge with master
- next step: fault tolerance (trigger error during
query execution, catch it, rerun query under the
hood, all without user knowing about the extra
activities)
parser
- fixing near neighbor
push metadata to master?
- it will be disruptive (now we need to run qms,
metadata must be loaded)
- warn the whole team first, especially the in2p3 group
no meeting next week
- Jacek @mysql conf, Daniel taking ~2 days off
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
|