multithreading - Python logger prints the same output several times in multithreaded environment -
this question has answer here:
in code, somehow logger prints out events twice, conventional print shows correct number of events.
select 1,2 select 1,2 2013-04-19 18:37:30,618:4561354752 - sqllogger - debug - select 1,2 2013-04-19 18:37:30,618:4561354752 - sqllogger - debug - select 1,2 2013-04-19 18:37:30,618:4565561344 - sqllogger - debug - select 1,2 2013-04-19 18:37:30,618:4565561344 - sqllogger - debug - select 1,2 it not clear me why same message being published mutliple times in multithreaded situations.
import logging threading import thread class sqlengine(object): def __init__(self, db_path): ch = logging.streamhandler() ch.setlevel(logging.debug) formatter = logging.formatter('%(asctime)s:%(thread)d - %(name)s - %(levelname)s - %(message)s') ch.setformatter(formatter) self.logger = logging.getlogger('sqllogger') self.logger.setlevel(logging.debug) self.logger.addhandler(ch) def execute(self,sql,): self.logger.debug(sql) print sql class dbworker(thread): def __init__(self, name): thread.__init__(self) self.name = name def run(self): db = sqlengine('') db.execute('select 1,2') if __name__ == '__main__': dbworker('thread 1').start() dbworker('thread 2').start()
looking @ logger documentation:
loggers never instantiated directly, through module-level function logging.getlogger(name). multiple calls getlogger() same name return reference same logger object.
now constructor calling following code:
ch = logging.streamhandler() self.logger = logging.getlogger('sqllogger') self.logger.addhandler(ch) note don't create new logger every sqlengine object, reference same logger instance. means you're adding handler 1 , same logger, hence after creating second object logger has 2 handlers, each of them printing screen.
you we'll need either register 1 handler (for example outside of sqlengine constructor) or call getlogger in __init__ different name every sqlengine instance.
Comments
Post a Comment