Attendees: Daniel, AndyS, AndyH, Bill, Jacek
Jacek
- playing with jira
- notifications: email notifications can be
set up, Tony J. at slac knows how to set it up
- would be useful to have a demo of how to
use stash
- reviewed 2939
- design/hackathon week
- 1 set of rooms reserved
--> perhaps get second room for afternoons
parallel sessions
- planning to look at other open source systems
~ week before (to learn how we can improve)
- all of us should do it
AndyH
- reworked all flow control for client and server,
based on a long discussion with Daniel
- everything is async
- will be ready tomorrow/Monday
Daniel
- 2939 (lat/long theta/phi)
- more cleanup: mysqlExportMgr --> chunkInventory,
improving interface, adding tests, proper error
handling
- looking into mock test frameworks
- dependency?
- both google and turtle bring dependency,
- hippoMocks super simple, one header
AndyS
- installing/building qserv on NCSA machine
Bill
[demo and discussion about logging]
- do we care about internal numbering of logging levels?
- log4cxx and python logging use different numbers
- people should not deal with the numbers
- also, names don't map perfectly (critical vs fatal)
- the layer written by Bill
- covers what was discussed and agreed 3 months ago
- no state in the layer
- nicer/friendlier api
- additional metedata (thread id, function name, line number etc)
- hierarchical naming?
- it is part of log4cxx
--> will play with that in the context of qserv (core modules)
- default appenders are synchronous
- will slow down qserv (!)
--> someone needs to research non-blocking asynchronous appenders
for log4cxx (yes, they do exist)
- observation: we can define our own set of macros for qserv
to shorten code for logging
- for some macros (but not all) everything is shut off when
given logging level is not active
- python side: still doing formatting in current implementation,
even if debugging level turned of.
- will fix it (separate all formatting into one argument
that is not processed unless needed)
- new string substitutions
(http://legacy.python.org/dev/peps/pep-3101/) vs "old one"
- need to decide which one we want to use
- threadId - better to have compact representation (to look
at things visually) - AndyH run into that in xrootd.
- solution: associate a simpler number with each thread.
- but don't maintain a dictionary, just pass a number
to each thread
- we might want to add key/value pairs in a way that is
machine readable (and human readable)
- misconfiguration might result in duplicate messages
- perhaps define special python handler, intercept and redirect
to our log4cxx-based logging?
- possible to compile all logging out?
- "log_info", "log_error" type macros are easy, generic "log"
is harder
- next steps
- improve documentation and then discuss on lsst-data
to get ball rolling (very soon)
- next iteration covering issues raised - no later than
end of March
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
|