statsd 0.4.0

Installing

The easiest way to install statsd is with pip!

You can install from PyPI:

$ pip install statsd

Or GitHub:

$ pip install -e git+https://github.com/jsocol/pystatsd#egg=statsd

Or from source:

$ git clone https://github.com/jsocol/pystatsd
$ cd statsd
$ python setup.py install

In Django

If you’re lucky enough to be using statsd in Django, you can configure a default client in your settings module with two values. The defaults are:

STATSD_HOST = 'localhost'
STATSD_PORT = 8125

Then instead of instantiating a new client every time, you can just grab:

>>> from statsd import statsd
>>> statsd.incr('foo')

You can even set a prefix (optionally):

STATSD_PREFIX = 'foo'

This can help differentiate between environments, like dev, staging, and production.

Context Manager

You can use a StatsClient instance as a context manager to easily time sections of code with the timer() method:

>>> from statsd import statsd
>>> with statsd.timer('bar'):
...     func()
...     func()

When the managed block exits, the client will automatically send the time it took to statsd.

If you’d like to catpure the elapsed time, add a variable to the with block:

>>> from statsd import statsd
>>> with statsd.timer('bar') as timer:
...     func()
>>> print timer.ms  # Elapsed time in milliseconds.

Decorator

You can also use a StatsClient instance as a decorator, also with the timer() method:

>>> from statsd import statsd
>>> @statsd.timer('bar')
... def foo():
...     pass

Every time foo() is called, timing information will be sent to the stat bar.

Sample Rates

All methods support an optional rate (kw)arg. This is a float between 0 and 1 that specifies what fraction of data to send through (for a specific call). Sample rates are recorded by statsd.

For example, here foo will be incremented approximately 50% of the time:

>>> from statsd import statsd
>>> statsd.incr('foo', 1, rate=0.5)

Statsd understands that this is a 50% sample rate and will adjust accordingly.

Similarly with decr() and timings:

>>> from statsd import statsd
>>> statsd.decr('foo', 1, rate=0.5)
>>> statsd.timing('foo', 320, rate=0.25)
>>> with statsd.timer('bar', rate=0.1):
...    pass
>>> @statsd.timer('bar', rate=0.5)
... def foo():
...     pass