Logger does not find handler

[tampakrap]

For unknown reason, the logger is NOT able to find a handler unless a settings.VARIABLE is called! In some cronjobs, this was done through imports of other modules that reside inside django/, most commonly the import of django.db which contains (in the __init__.py file) the variable settings.DATABASES. In cronos, in order to work around it, I added settings.DEBUG in apps/__init__.py Steps to reproduce: 1) Comment out settings.DEBUG from apps/__init__.py 2) Put the following script under apps/teilar/handlerbug.py:

import os
import sys
from proj_root import PROJECT_ROOT
sys.path.append(PROJECT_ROOT)
os.environ['DJANGO_SETTINGS_MODULE'] = 'apps.settings'
from apps import log_extra_data
import logging

logger = logging.getLogger('cronos')
print logger.__dict__

try:
    import sldkfjsfj
except Exception as error:
    logger.error(error, extra = log_extra_data())

3) Run the script: python apps/teilar/handlerbug.py

• Result:

{'name': 'cronos', 'parent': <logging.RootLogger object at 0x9e17d0>, 'handlers': [], 'level': 0, 'disabled': 0, 'manager': <logging.Manager object at 0x9e1890>, 'propagate': 1, 'filters': []}
No handlers could be found for logger "cronos"

• Expected Result:

{'name': 'cronos', 'parent': <logging.RootLogger object at 0x9eb090>, 'handlers': [<logging.StreamHandler object at 0xe5c1d0>], 'level': 10, 'disabled': 0, 'manager': <logging.Manager object at 0x9eb150>, 'propagate': 1, 'filters': []}
cronos: ERROR    Message: No module named sldkfjsfj                     
cronos: File: test Function: <module> Line: 15

[tampakrap]

This could be a bug (or feature) in django itself, but the proper solution is to get rid of the cronjobs and move to celery

[Drakevr]

I'd go with Celery being more "Django-ish" in choice rather than dirty hacking stuff around with cronjobs to circumvent said problem.

However Celery requires a broker of its own for communication which increases the set of deps by 2 (Celery + your broker of choice)

if deps are not a problem though then yeah think of a move to that.

NOTE: if you are handling critical data.. you better avoid redis here , even if its feature complete it tends to barf with power failures

[tampakrap]

the migration to celery is still needed though, the wiki entry will stay there