Skip to content

A puma plugin that sends key metrics to statsd

License

Notifications You must be signed in to change notification settings

GetDutchie/puma-plugin-statsd

 
 

Repository files navigation

Puma Statsd Plugin

Puma integration with statsd for easy tracking of key metrics that puma can provide:

  • puma.workers
  • puma.booted_workers
  • puma.running
  • puma.backlog
  • puma.pool_capacity
  • puma.max_threads

Installation

Add this gem to your Gemfile with puma and then bundle:

gem "puma"
gem "puma-plugin-statsd"

Add it to your puma config:

# config/puma.rb

bind "http://127.0.0.1:9292"

workers 1
threads 8, 16

plugin :statsd

Usage

Ensure you have an environment variable set that points to a statsd host, then boot your puma app as usual. Optionally you may specify a port (default is 8125).

STATSD_HOST=127.0.0.1 bundle exec puma
STATSD_HOST=127.0.0.1 STATSD_PORT=9125 bundle exec puma

Datadog Integration

metric tags are a non-standard addition to the statsd protocol, supported by the datadog "dogstatsd" server.

Should you be reporting the puma metrics to a dogstatsd server, you can set tags via the following three environment variables.

DD_TAGS

DD_TAGS: Set this to a space-separated list of tags, using the datadog agent standard format.

For example, you could set this environment variable to set three datadog tags, and then you can filter by in the datadog interface:

export DD_TAGS="env:test simple-tag-0 tag-key-1:tag-value-1"
bundle exec rails server

MY_POD_NAME

MY_POD_NAME: Set a pod_name tag to the metrics. The MY_POD_NAME environment variable is recommended in the datadog kubernetes setup documentation, and for puma apps deployed to kubernetes it's very helpful to have the option to report on specific pods.

You can set it on your pods like this:

env:
  - name: MY_POD_NAME
    valueFrom:
      fieldRef:
        fieldPath: metadata.name

STATSD_GROUPING

STATSD_GROUPING: add a grouping tag to the metrics, with a value equal to the environment variable value. This is particularly helpful in a kubernetes deployment where each pod has a unique name but you want the option to group metrics across all pods in a deployment. Setting this on the pods in a deployment might look something like:

env:
  - name: STATSD_GROUPING
    value: deployment-foo

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/yob/puma-plugin-statsd.

Testing the data being sent to statsd

Start a pretend statsd server that listens for UDP packets on port 8125:

ruby devtools/statsd-to-stdout.rb

Start puma:

STATSD_HOST=127.0.0.1 bundle exec puma devtools/config.ru --config devtools/puma-config.rb

Throw some traffic at it, either with curl or a tool like ab:

curl http://127.0.0.1:9292/
ab -n 10000 -c 20 http://127.0.0.1:9292/

Watch the output of the UDP server process - you should see statsd data printed to stdout.

Acknowledgements

This gem is a fork of the excellent puma-plugin-systemd by Samuel Cochran.

Other puma plugins that were helpful references:

The puma docs were also helpful.

License

The gem is available as open source under the terms of the MIT License.

About

A puma plugin that sends key metrics to statsd

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Ruby 100.0%