Skip to main content

Nonblocking Logging Handler for Python Logging

Project description

Nonblocking Log Handler

Provides a NonblockingLogHandler class consistent with the Python logging

This handler acts as a proxy for the another log handler that may be slow
to execute: e.g. the SMTPHandler, SocketHandler, SysLogHandler - especially
when they are talking to remote servers.

If you have a real-time system (i.e. one where a late response is a wrong
response) and you are sending log messages via email, http, syslog, etc.,
you should consider using this module to protect against unexpected delays.

It is intended to be a drop-in replacement (see provisos below) for your
proxied handler. Unlike the proxied handler, this handler which returns
quickly, and executes the actually logging in the background, in a separate


* If you pass an objects to the logging functions, be aware that their
__str__() functions should be fast. I/O bound str() calls are outside the
scope of this module.

Execution of str() functions on message parameters and string formatting
is done immediately, in the calling thread. This is to guarantee both
the thread-safety of the logged objects and also that the log shows the
object's values at the time of the call, not the time of the emit.

* If you subclass from Formatter, note that it will not have the user
parameters from the logging call passed to it. It will only receive the
resulting message string formatted according to the client's wishes. It
can still format dates, threadnames, levels, etc.

* If you subclass from Filter, be aware that they should be fast. I/O-bound
filters are outside of the scope of this module.

* Clients should not directly call format() or emit() on the
NonblockingLogHandler class. Consider them private.

* Once the NonblockingLogHandler handler is initialised, any further calls
to setLevel(), addFilter and removeFilter() made on the delegated handler
are ignored by the proxy handler. Conversely, the delegated handler is
not informed of calls made on the NonblockingLogHandler instance.

Best practice would be to only update the NonblockingLogHandler handler,
and not use the values stored in the delegated handler.

* If the close() method is called on the NonblockingLogHandler handler,
it does *not* call close on the delegated handler, in case you are using
it elsewhere in the application. This makes it difficult to ever call
close() on the delegated handler, becausse it may still be in use.

Example usage:
nonblocking_email_handler = nonblockingloghandler.NonblockingLogHandler(
"localhost", "",
"", "Log message")
db_logger = logging.getLogger("database")

# This operation will return immediately, before email is sent.
db_logger.critical("Database corrupted")



* Only tested on Python 2.7.2, Windows and Ubuntu
* Let me know if there is demand for other versions.
* Known to have a rare intermittent shut-down race condition on Windows
* Results in exception report being displayed to screen:
"Exception in thread NonblockingLogHandler.[...] (most likely raised
during interpreter shutdown)"
* Suspected (unconfirmed) cause Python run-time bug, fixed in Python 3.2.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution (37.2 kB view hashes)

Uploaded Source

Built Distribution

nonblockingloghandler-1.1.2-py2.7.egg (13.0 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page