Awesome

Django InfluxDB Metrics

A reusable Django app that sends metrics about your project to InfluxDB.

IMPORTANT NOTE: This release only supports InfluxDB >= 0.9. We have also dropped a few measurements like CPU, memory and disk-space because Telegraf can collect these in a much much better way.

Prerequisites

This module has celery support but you don't have to use it, if you don't want to.

Installation

To get the latest stable release from PyPi

.. code-block:: bash

pip install django-influxdb-metrics

To get the latest commit from GitHub

.. code-block:: bash

pip install -e git+git://github.com/bitmazk/django-influxdb-metrics.git#egg=influxdb_metrics

Add influxdb_metrics to your INSTALLED_APPS

.. code-block:: python

INSTALLED_APPS = (
    ...,
    'influxdb_metrics',
)

Settings

You need to set the following settings::

INFLUXDB_HOST = 'your.influxdbhost.com'
INFLUXDB_PORT = '8086'
INFLUXDB_USER = 'youruser'
INFLUXDB_PASSWORD = 'yourpassword'
INFLUXDB_DATABASE = 'yourdatabase'

# This is for tagging the data sent to your influxdb instance so that you
# can query by host
INFLUXDB_TAGS_HOST = 'your_hostname'

# Seconds to wait for the request to the influxdb server before timing out
INFLUXDB_TIMEOUT = 5

# Set this to True if you are using Celery
INFLUXDB_USE_CELERY = True

# Set this to True if you are not using Celery
INFLUXDB_USE_THREADING = False

If you would like to disable sending of metrics (i.e. for local development), you can set::

INFLUXDB_DISABLED = True

If you are having trouble getting the postgresql database size, you might need to set::

INFLUXDB_POSTGRESQL_USE_LOCALHOST = True

Use ssl with INFLUXDB_HOST::

INFLUXDB_SSL = True # default is False

Optional with ssl::

INFLUXDB_VERIFY_SSL = True # default is False

Specify a prefix for metric measurement names (default is django_, E.g. django_request)

INFLUXDB_PREFIX = "my_app"     # measurement name == 'my_app_request'
INFLUXDB_PREFIX = ""           # measurement name == 'request'
INFLUXDB_PREFIX = None         # measurement name == 'request'

Usage

The app comes with several management commands which you should schedule via crontab.

influxdb_get_postgresql_size

Collects the total disk usage for the given database.

You can run it like this::

./manage.py influxdb_get_postgresql_size db_role db_name

You should provide role and name for the database you want to measure. Make sure that you have a .pgpass file in place so that you don't need to enter a password for this user.

You could schedule it like this::

0 */1 * * * cd /path/to/project/ && /path/to/venv/bin/python /path/to/project/manage.py influxdb_get_postgresql_size db_role db_name > $HOME/mylogs/cron/influxdb-get-postgresql-size.log 2>&1

The measurement created in your InfluxDB will be named postgresql_size and will have the following fields:

value: The total database size in bytes

InfluxDbEmailBackend

If you would like to track the number of emails sent, you can set your EMAIL_BACKEND::

EMAIL_BACKEND = 'influxdb_metrics.email.InfluxDbEmailBackend'

When the setting is set, metrics will be sent every time you run .manage.py send_mail.

The measurement created in your InfluxDB will be named django_email_sent and will have the following fields:

value: The number of emails sent

InfluxDBRequestMiddleware

If you would like to track the number and speed of all requests, you can add the InfluxDBRequestMiddleware at the top of your MIDDLEWARE_CLASSES::

MIDDLEWARE_CLASSES = [
    'influxdb_metrics.middleware.InfluxDBRequestMiddleware',
    ...
]

The measurement created in your InfluxDB will be named django.request and will have the following fields:

value: The request time in milliseconds.

Additionally, it will have the following tags:

is_ajax: true if it was an AJAX request, otherwise false
is_authenticated: true if user was authenticated, otherwise false
is_staff: true if user was a staff user, otherwise false
is_superuser: true user was a superuser, otherwise false
method: The request method (GET or POST)
module: The python module that handled the request
view: The view class or function that handled the request
referer: The full URL from request.META['HTTP_REFERER']
referer_tld: The top level domain of the referer. It tries to be smart and regards google.co.uk as a top level domain (instead of co.uk)
full_path: The full path that was requested
path: The path without GET params that was requested
campaign: A value that is extracted from the GET-parameter campaign, if present. You can change the name of this keyword from campaign to anything via the setting INFLUXDB_METRICS_CAMPAIGN_KEYWORD.

If you have a highly frequented site, this table could get big really quick. You should make sure to create a shard with a low retention time for this series (i.e. 7d) and add a continuous query to downsample the data into hourly/daily averages. When doing that, you will obviously lose the detailed information like referer and referer_tld but it might make sense to create a second continuous query to count and downsample at least the referer_tld values.

NOTE: I don't know what impact this has on overall request time or how much stress this would put on the InfluxDB server if you get thousands of requests. It would probably wise to consider something like statsd to aggregate the requests first and then send them to InfluxDB in bulk.

Tracking Users

This app's models.py contains a post_save and a post_delete handler which will detect when a user is created or deleted.

It will create three measurements in your InfluxDB:

The first one will be named django_auth_user_create and will have the following fields:

value: 1

The second one will be named django_auth_user_delete and will have the following fields:

value: 1

The third one will be named django_auth_user_count and will have the following fields:

value: The total number of users in the database

Tracking User Logins

This app's models.py contains a handler for the user_logged_in signal.

The measurement created in your InfluxDB will be named django_auth_user_login and will have the following fields:

value: 1

Making Queries

If you need to get data out of your InfluxDB instance, you can easily do it like so::

from influxdb_metrics.utils import query query('select * from series.name', time_precision='s', chunked=False)

The method declaration is the same as the one in InfluxDBClient.query(). This wrapper simply instanciates a client based on your settings.

Contribute

If you want to contribute to this project, please perform the following steps

.. code-block:: bash

# Fork this repository
# Clone your fork
mkvirtualenv -p python3.5 django-influxdb-metrics
make develop

git co -b feature_branch master
# Implement your feature and tests
git add . && git commit
git push -u origin feature_branch
# Send us a pull request for your feature branch

Runing tests

For running the tests Docker and Docker compose is required.

The test setup a Influxdb database for testing against real queries.

In order to run the tests just run the command::

./run_tests_with_docker.sh