Python Persist Log File Stream with PyInotify

Question

I am experiencing an issue persisting a log file write stream through pyinotify and it's threads. I am using pyinotify to monitor a directory for CLOSE_WRITE file events. Before I initialize pyinotify I create a log stream using the built-in logging module like so:

import os, logging
from logging import handlers
from logging.config import dictConfig


log_dir = './var/log'
name = 'com.sadmicrowave.tesseract'
LOG_SETTINGS = { 'version' : 1
                ,'handlers': { 'core': {
                                    # make the logger a rotating file handler so the file automatically gets archived and a new one gets created, preventing files from becoming too large they are unmaintainable. 
                                    'class'     : 'logging.handlers.RotatingFileHandler'
                                    # by setting our logger to the DEBUG level (lowest level) we will include all other levels by default
                                    ,'level'        : 'DEBUG'
                                    # this references the 'core' handler located in the 'formatters' dict element below
                                    ,'formatter'    : 'core'
                                    # the path and file name of the output log file
                                    ,'filename'     : os.path.join(log_dir, "%s.log" % name)
                                    ,'mode'         : 'a'
                                    # the max size we want to log file to reach before it gets archived and a new file gets created
                                    ,'maxBytes'     : 100000
                                    # the max number of files we want to keep in archive
                                    ,'backupCount'  : 5 }
                            }
                             # create the formatters which are referenced in the handlers section above
                            ,'formatters': {'core': {'format': '%(levelname)s %(asctime)s %(module)s|%(funcName)s %(lineno)d: %(message)s' 
                                            }
                            }
                            ,'loggers'   : {'root': {
                                                        'level'     : 'DEBUG' # The most granular level of logging available in the log module
                                                        ,'handlers' : ['core']
                                            }
                            }
                        }

# use the built-in logger dict configuration tool to convert the dict to a logger config
dictConfig(LOG_SETTINGS)

# get the logger created in the config and named root in the 'loggers' section of the config
__log = logging.getLogger('root')

So, after my __log variable get initialized it works immediately, allowing for log writes. I want to start the pyinotify instance next and would like to pass __log using the following class definitions:

import asyncore, pyinotify

class Notify (object):
    def __init__ (self, log=None, verbose=True):
        wm = pyinotify.WatchManager()
        wm.add_watch( '/path/to/folder/to/monitor/', pyinotify.IN_CLOSE_WRITE, proc_fun=processEvent(log, verbose) )

        notifier = pyinotify.AsyncNotifier(wm, None)
        asyncore.loop()

class processEvent (pyinotify.ProcessEvent):
    def __init__ (self, log=None, verbose=True):
        log.info('logging some cool stuff')

        self.__log              = log
        self.__verbose          = verbose

    def process_IN_CLOSE_WRITE (self, event):
        print event

In the above implementation, my process_IN_CLOSE_WRITE method gets triggered exactly as expected from the pyinotify.AsyncNotifier ; however, the log line for logging some cool stuff never writes to the log file.

I feel like it has something to do with persisting the file stream through the pyinotify threading process; however, I'm not sure how to resolve this.

Any ideas?

Answer 1

I might have found a resolution that seems to work. Not sure if it is the best approach, so I will leave the OP open for now to see if any other ideas are posted.

I think I was handling my pyinotify.AsyncNotifier setup wrong. I changed the implementation to:

class Notify (object):
    def __init__ (self, log=None, verbose=True):
        notifiers = []
        descriptors = []
        wm = pyinotify.WatchManager()
        notifiers.append ( pyinotify.AsyncNotifier(wm, processEvent(log, verbose)) )
        descriptors.append( wm.add_watch( '/path/to/folder/to/monitor/', pyinotify.IN_CLOSE_WRITE, proc_fun=processEvent(log, verbose), auto_add=True )

        asyncore.loop()

Now, when my wrapper class processEvents gets triggered on instantiation of the listener, and when an event is triggered from a CLOSE_WRITE event, the log object is maintained and passed appropriately and can receive write events.