sprockets.mixins.metrics 1.1.1

Record performance metrics about your application

Adjust counter and timer metrics in InfluxDB or Graphite using the same API.

from sprockets.mixins import mediatype, metrics
from tornado import gen, web
import queries

class MyHandler(metrics.StatsdMixin, mediatype.ContentMixin,
                web.RequestHandler):

    def initialize(self):
        super(MyHandler, self).initialize()
        self.db = queries.TornadoSession(os.environ['MY_PGSQL_DSN'])

    @gen.coroutine
    def get(self, obj_id):
        with self.execution_timer('dbquery', 'get'):
           result = yield self.db.query('SELECT * FROM foo WHERE id=%s',
                                        obj_id)
        self.send_response(result)

This simple handler will emit a timer metric that identifies each call to the get method as well as a separate metric for the database query. Switching from using statsd to InfluxDB is simply a matter of switch from the metrics.StatsdMixin to the metrics.InfluxDBMixin.

The mix-in is configured through the tornado.web.Application settings property using a key defined by the specific mix-in.

Statsd Mixin

The following snippet configures the StatsD mix-in from common environment variables:

import os

from sprockets.mixins import metrics
from tornado import web

def make_application():
    settings = {
        metrics.StatsdMixin.SETTINGS_KEY: {
            'namespace': 'my-application',
            'host': os.environ.get('STATSD_HOST', '127.0.0.1'),
            'port': os.environ.get('STATSD_PORT', '8125'),
        }
    }
    return web.Application([
        # insert handlers here
    ], **settings)

namespace:

The namespace for the measurements

host:

The Statsd host

port:

The Statsd port

InfluxDB Mixin

The following snippet configures the InfluxDB mix-in from common environment variables:

import os

from sprockets.mixins import metrics
from tornado import web

def make_application():
    settings = {
        metrics.InfluxDBMixin.SETTINGS_KEY: {
            'measurement': 'my-application',
            'database': 'services',
            'write_url': 'http://{}:{}/write'.format(
                 os.environ.get('INFLUX_HOST', '127.0.0.1'),
                 os.environ.get('INFLUX_PORT', 8086)),
             'max_buffer_time': 3,
             'max_buffer_length': 100
        }
    }
    return web.Application([
        # insert handlers here
    ], **settings)

measurement:

The InfluxDB measurement name

database:

The InfluxDB database to write measurements into

write_url:

the InfluxDB write URL to send HTTP requests to

max_buffer_time:

The maximum elasped time measurements should remain in buffer before writing to InfluxDB.

max_buffer_length:

The maximum number of measurements to buffer before writing to InfluxDB.

Development Quickstart

$ python3.4 -mvenv env
$ . ./env/bin/activate
(env)$ env/bin/pip install -r requires/development.txt
(env)$ nosetests
test_that_cached_socket_is_used (tests.StatsdMethodTimingTests) ... ok
test_that_counter_accepts_increment_value (tests.StatsdMethodTimingTests) ... ok
test_that_counter_increment_defaults_to_one (tests.StatsdMethodTimingTests) ... ok
test_that_default_prefix_is_stored (tests.StatsdMethodTimingTests) ... ok
test_that_execution_timer_records_time_spent (tests.StatsdMethodTimingTests) ... ok
test_that_http_method_call_is_recorded (tests.StatsdMethodTimingTests) ... ok

----------------------------------------------------------------------
Ran 6 tests in 1.089s

OK
(env)$ ./setup.py build_sphinx -q
running build_sphinx
(env)$ open build/sphinx/html/index.html