AsynchronousLogstashHandler is a custom logging handler which sends Logstash messages using UDP and TCP. For example:

import logging
import sys
from logstash_async.handler import AsynchronousLogstashHandler

host = 'localhost'
port = 5959

test_logger = logging.getLogger('python-logstash-logger')
    host, port, database_path='logstash.db'))

# If you don't want to write to a SQLite database, then you do
# not have to specify a database_path.
# NOTE: Without a database, messages are lost between process restarts.
# test_logger.addHandler(AsynchronousLogstashHandler(host, port))

test_logger.error('python-logstash-async: test logstash error message.')'python-logstash-async: test logstash info message.')
test_logger.warning('python-logstash-async: test logstash warning message.')

# add extra field to logstash message
extra = {
    'test_string': 'python version: ' + repr(sys.version_info),
    'test_boolean': True,
    'test_dict': {'a': 1, 'b': 'c'},
    'test_float': 1.23,
    'test_integer': 123,
    'test_list': [1, 2, '3'],
}'python-logstash: test extra fields', extra=extra)

When using the extra field make sure you don’t use reserved names. From Python documentation:

The keys in the dictionary passed in extra should not clash
with the keys used by the logging system.
(See the `Formatter <>`_ documentation
for more information on which keys are used by the logging system.)

You can also specify an additional extra dictionary in the logging configuration (e.g. via FileConfig, DictConfig or logging configuration in the code) with static values like the application name, environment, etc. These values will be merged with any extra dictionary items passed in the logging call into the configured extra prefix, e.g.:

from logstash_async.formatter import LogstashFormatter
from logstash_async.handler import AsynchronousLogstashHandler

logstash_formatter = LogstashFormatter(
    extra=dict(application='example-app', environment='production'))

Usage with Django

Modify your to integrate python-logstash-async with Django’s logging:

  'formatters': {
      'logstash': {
          '()': 'logstash_async.formatter.DjangoLogstashFormatter',
          'message_type': 'python-logstash',
          'fqdn': False, # Fully qualified domain name. Default value: false.
          'extra_prefix': 'dev', #
          'extra': {
              'application': PROJECT_APP,
              'project_path': PROJECT_APP_PATH,
              'environment': 'production'
  'handlers': {
      'logstash': {
          'level': 'DEBUG',
          'class': 'logstash_async.handler.AsynchronousLogstashHandler',
          'formatter': 'logstash',
          'transport': 'logstash_async.transport.TcpTransport',
          'host': '',
          'port': 5959,
          'ssl_enable': True,
          'ssl_verify': True,
          'ca_certs': 'etc/ssl/certs/logstash_ca.crt',
          'certfile': '/etc/ssl/certs/logstash.crt',
          'keyfile': '/etc/ssl/private/logstash.key',
          'database_path': '{}/logstash.db'.format(PROJECT_ROOT),
  'loggers': {
      'django.request': {
          'handlers': ['logstash'],
          'level': 'DEBUG',
          'propagate': True,

This would result in a Logstash event like the following (note: to some extend dependent of your Logstash configuration):

    "@timestamp": "2016-10-23T15:11:16.853Z",
    "@version": "1",
    "extra": {
        "application": "django_example",
        "django_version": "1.10.2",
        "environment": "production",
        "func_name": "get_response",
        "interpreter": "/home/enrico/example/venv/bin/python",
        "interpreter_version": "2.7.12",
        "line": 152,
        "logger_name": "django.request",
        "path": "/home/enrico/example/venv/lib/python2.7/site-packages/django/core/handlers/",
        "process_name": "MainProcess",
        "project_path": "/home/enrico/example/app",
        "req_host": "localhost",
        "req_method": "GET",
        "req_referer": "",
        "req_remote_address": "",
        "req_uri": "http://localhost/hosts/nonexistent/",
        "req_user": "enrico",
        "req_useragent": "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:40.0) Gecko/20100101 Firefox/40.1",
        "request": "<WSGIRequest: GET '/hosts/nonexistent/'>",
        "status_code": 404,
        "thread_name": "Thread-1"
    "host": "",
    "level": "WARNING",
    "logsource": "",
    "message": "Not Found: /hosts/nonexistent/",
    "pid": 23605,
    "port": 56170,
    "program": "",
    "type": "python-logstash"

Usage with Logging File Config

Example code for Python’s logging.config.fileConfig:

import logging
from logging.config import fileConfig

fileConfig('logging.conf', disable_existing_loggers=True)
logger = logging.getLogger()'python-logstash-async: test logstash info message.')

Example config for Python’s logging.config.fileConfig:

# loggers
keys = root

name = python-app
level = DEBUG
handlers = console,logstash
propagate = 1
qualname = root

# handlers
keys = console,logstash

class = StreamHandler
level = NOTSET
formatter = console
args = (sys.stdout,)

class = logstash_async.handler.AsynchronousLogstashHandler
level = DEBUG
formatter = logstash
args = ('%(host)s', %(port)s, '%(database_path)s', '%(transport)s', %(ssl_enable)s, %(ssl_verify)s, '%(keyfile)s', '%(certfile)s', '%(ca_certs)s', %(enable)s)
transport = logstash_async.transport.TcpTransport
host = localhost
port = 5959
enable = True
ssl_enable = True
ssl_verify = True
ca_certs = /etc/ssl/certs/ca.crt
certfile = /etc/ssl/certs/logstash.crt
keyfile = /etc/ssl/private/logstash.key
database_path = /var/lib/logstash.db

keys = console,logstash

format = %(asctime)s %(name)-12s %(levelname)+8s %(message)s

class = logstash_async.formatter.LogstashFormatter
# format, datefmt and style are a hack: we cannot specify "args" for formatters
# (see _create_formatters()), so we pass our values as format parameters
# "format" corresponds to LogstashFormatter's "message_type" argument,
# "datefmt" to "tags" and "style" to "fqdn" ("style" is Python3 only).
# However, the "tags" argument expects a list and "fqdn" expects a boolean but Python's
# logging framework passes strings for both, so this is of limited use.
format = format
datefmt = custom-tag
style = True


As also stated in the comment in the example configuration above, Python’s fileConfig format does not allow to pass arbitary arguments to a formatter class in the config file in the same way as for handlers. It supports only three arguments: format, datefmt and style (where style is Python3 only) and passes those as positional arguments to the formatter class.

You can either use the hack shown in the example by setting at least the message_type argument of LogstashFormatter which is its first position argument and so corresponds to format in the logging configuration.

A better and more clean solution is to create a subclass of LogstashFormatter and set the various configuration values there or use a different formatter like This is a limitation of Python’s logging file config format.

Another example using Python logging file config in combination with Gunicorn can be found on

Trigger event flushing

In case you need to trigger flushing queued events (as if it is important for your application to try to send events as fast as possible or similar), the AsynchronousLogstashHandler class provides a method flush which will trigger flushing of queued events in the asynchronous worker thread.

There is no guarantee that the flush will succeed but so you can bypass the next constants.QUEUED_EVENTS_FLUSH_INTERVAL resp. constants.QUEUED_EVENTS_FLUSH_COUNT (see Options for the asynchronous processing and formatting for details.).

In case sending the queued events to Logstash failed, the events will be requeued as usual and the flush signal is reset. That is, until the next attempt to send queued events, constants.QUEUED_EVENTS_FLUSH_INTERVAL and constants.QUEUED_EVENTS_FLUSH_COUNT will be taken into account again.