StatsD

A very simple client to format and send metrics to a StatsD server.

gem install statsd

In your client code:

require 'rubygems'
require 'lookout/statsd'
STATSD = Lookout::StatsdClient.new(:host => 'localhost', :port => 8125)

STATSD.increment('some_counter') # basic incrementing
STATSD.increment('system.nested_counter', 0.1) # incrementing with sampling (10%)

STATSD.decrement(:some_other_counter) # basic decrememting using a symbol
STATSD.decrement('system.nested_counter', 0.1) # decrementing with sampling (10%)

STATSD.timing('some_job_time', 20) # reporting job that took 20ms
STATSD.timing('some_job_time', 20, 0.05) # reporting job that took 20ms with sampling (5% sampling)

There is an option for reduced DNS lookups, you can specify an additional constructor option :resolve_always and set it to false. By default, the client will always resolve the address unless host is set to 'localhost' or '127.0.0.1'.

require 'rubygems'
require 'lookout/statsd'

STATSD = Lookout::StatsdClient.new(:host => 'specialstats.host.example',
                                   :port => '8125',
                                   :resolve_always => false)

STATSD.increment('some_counter') # basic incrementing

Since class variables and instance variables are not thread-safe on initialization, there is a potential for multiple UDP sockets being opened upon if you are using a truly multithreaded ruby, i.e. JRuby. Make sure to take that in to account when initializing this library.

• UDP Client libraries use UDP to send information to the StatsD daemon.

• Graphite

Graphite uses "schemas" to define the different round robin datasets it houses (analogous to RRAs in rrdtool):

[stats]
priority = 110
pattern = ^stats\..*
retentions = 10:2160,60:10080,600:262974

That translates to:

• 6 hours of 10 second data (what we consider "near-realtime")
• 1 week of 1 minute data
• 5 years of 10 minute data

This has been a good tradeoff so far between size-of-file (round robin databases are fixed size) and data we care about. Each "stats" database is about 3.2 megs with these retentions.

Etsy's blog post.

StatsD was inspired (heavily) by the project (of the same name) at Flickr. Here's a post where Cal Henderson described it in depth: Counting and timing. Cal re-released the code recently: Perl StatsD