Introduction

django-zesty-metrics is a middleware suite that uses statsd to report important metrics to a StatD server.

Out of the box, it will track:

• response times individually by view, and in aggregate

• aggregate user activity data

With some minor configuration, it will track:

• individual user activity data

• arbitrary metrics

Requirements

• Django >= 1.4

• statsd == 1.0 (StatsD client; GitHub <https://github.com/jsocol/pystatsd>)

• Graphite server (docs)

• Statsd server (blog post)

Installation

Install with pip install django-zesty-metrics.

Alternately, clone the git repository and execute python setup.py install within.

Configuration

In your Django settings:

• Add zesty_metrics to the INSTALLED_APPS

• Add zesty_metrics.middleware.MetricsMiddleware to MIDDLEWARE_CLASSES

• Set the following, as needed: - STATSD_HOST, default localhost - STATSD_PORT, default 8125 - STATSD_PREFIX, default None - ZESTY_TRACKING_CLASSES, default ('zesty_metrics.tracking.UserAccounts',)

• Run syncdb (or migrate if you use South).

Set up a cron job to run the report_metrics django-admin.py command regularly. At least once a day, but you can update it as often as you want. This command reports metrics from the trackers that you configure in ZESTY_TRACKING_CLASSES.

If you want to send metrics from the client-side, hook up the default URLs in your urls.py:

urlpatterns = patterns(
    '',
    url(r'^metrics/', include('zesty_metrics.urls')),
    )

Acknowledgements

Lots of ideas were taken from django-statsd and django-munin.

CHANGELOG

• 0.3.1: - Added missing migrations and management commands.

• 0.3: - Added tests! Many things that didn’t work now work. - Added DailyActivityRecord for tracking daily activity.

• 0.2: Added latency tracking, myriad bug-fixes.

• 0.1.1: Fixed “NO VALUE” error in report_metrics command when values were pulled from the cache.

• 0.1: Initial release