Tweaking Zope Event Log

Sometimes, you encounter a strange bug on a running server and you wish to change the log level.

C:\Zope275>bin\python lib/python/ZServer/...
  medusa/monitor_client_win32.py localhost 7099
>>> import logging
>>> e = logging.getLogger('event')
>>> print e.level
20
>>> print [h.level for h in e.handlers]  # what are the handlers logging at?
[5, 25]
>>> # handler 0 effectively logs at level 20, because e.getEffectiveLevel() == 20
>>> e.setLevel(0)
>>> from zLOG import LOG, TRACE
>>> LOG('Test Module', TRACE, 'Test Message') # 'Test message'
... should appear in the log file

There are two kinds of levels in the zope.conf. One is the logger level, and the other is the log-handlers level. These levels mean different things. The logger level actually specifies a filter for the child log-handlers, and would have been more appropriate to name it minimum-level. Confused? Here is an example:

<eventlog>
  # eventlog.level defaults to info.
  # uncomment below for expected behaviour
  # level all
  <timedrotatinglogfile>
    level all
    path $INSTANCE/log/event.log
    when D
  </timedrotatinglogfile>
  <email-notifier>
    level error
    from XXX
    to XXX
    smtp-server XXX
  </email-notifier>
</eventlog>

In the case above, event log has a logger-level of info, while timedrotatinglogfile has a loghandler-level of all. Effectively, timedrotatinglogfile will not log at level=all at all, as the event log level specifies that only logs of level info or greater will be recorded. Setting timedrotatinglogfile level to “all” does not override eventlog setting.