Graphite consists of a storage backend and a web-based visualization frontend. Client applications send streams of numeric time-series data to the Graphite backend (called carbon), where it gets stored in fixed-size database files similar in design to RRD. The web frontend provides 2 distinct user interfaces for visualizing this data in graphs as well as a simple URL-based API for direct graph generation.
Graphite's design is focussed on providing simple interfaces (both to users and applications), real-time visualization, high-availability, and enterprise scalability.
This charm will install Graphite. You will also need something like gunicorn or
apache in front of it.
$ make deploy
If you would like to deploy with postgres instead of sqlite3 you will also need
to deploy postgres and add the relation:
$ juju add-relation graphite:pgsql postgres:db
This charm does not currently support relations with other services that would
generate statistics to be added to it. You will need to manually configure
services to connect to the graphite instance and pass it the statistics you
need. This is partially due to juju's current lack of cross-environment
relations (i.e. you may well want to set up one graphite instance that can
accept data from multiple juju environments), but could be solved at some point
by a proxy charm that know the location of a central graphite server and allows
multiple juju environments to connect to the same graphite server.
from txstatsd.client import UdpStatsDClient
from txstatsd.metrics.metrics import Metrics
conn = UdpStatsDClient(host='127.0.0.1', port=8125) # replace IP with graphite deployment IP
METRICS = Metrics(connection=conn, namespace='my-project') # replace namespace as needed
The makefile to run tests requires the following dependencies
$ sudo apt-get install python-nose python-mock python-flake8
Presently the graphite charm only works with one graphite instance deployed or
one graphite and one postgresql instance deployed. It is not presently able to
scale out further than that.