bc-prometheus-ruby - Drop-in Prometheus metrics

Installation

gem 'bc-prometheus-ruby'

Then in your application.rb, prior to extending Rails::Application or any initializers:

require 'bigcommerce/prometheus'

Then in your web server config file (e.g. puma.rb)

before_fork do
  Rails.application.config.before_fork_callbacks.each(&:call)
end

You can then view your metrics at: http://0.0.0.0:9394/metrics

Puma

For extra Puma metrics, add this to config/puma.rb:

after_worker_fork do
  Rails.application.config.after_fork_callbacks.each(&:call)
end

Resque

In your task 'resque:setup' rake task, do:

require 'bigcommerce/prometheus'
Bigcommerce::Prometheus::Instrumentors::Resque.new(app: Rails.application).start

Per-job metrics (opt-in)

Set PROMETHEUS_RESQUE_PER_JOB_METRICS_ENABLED=1 on Resque worker pods to enable two additional histograms recorded from the parent worker process.

• resque_job_queue_latency_seconds{job_class} — time from scheduled_at (falling back to enqueued_at) until a worker picks the job up. Per attempt; retries-with-backoff anchor on scheduled_at so the intentional backoff doesn't show as latency.
• resque_job_perform_duration_seconds{job_class} — total Resque child lifetime (fork → Process.waitpid return). Includes fork overhead, Redis reconnect, after_fork hooks, perform, and exit.

These are off by default because they emit one histogram observation per job per worker pod, which adds cardinality. Opt in per service.

resque_job_queue_latency_seconds is supported for jobs enqueued via ActiveJob (.perform_later) — the enqueue timestamps come from ActiveJob's serialized payload, and the job_class label is the user's job class name, not ActiveJob::QueueAdapters::ResqueAdapter::JobWrapper. scheduled_at is serialized by ActiveJob from Rails 7.1; on older Rails the payload carries only enqueued_at, so intentional scheduling/backoff delay counts toward queue latency. Vanilla Resque jobs (Resque.enqueue) carry no enqueue timestamps, so queue_latency silently no-ops for them; perform_duration works for every job regardless.

Requires resque >= 1.27 workers running in the default fork-per-job mode — the hooks instrument Resque::Worker#perform_with_fork. Non-forking workers (FORK_PER_JOB=false, or platforms without fork) are not instrumented; a warning is logged at worker boot when per-job metrics are enabled but cannot record.

Configuration

After requiring the main file, you can further configure with:

Option	Description	Default	Environment Variable
client_custom_labels	A hash of custom labels to send with each client request	{}	None
client_max_queue_size	The max amount of metrics to send before flushing	10000	ENV['PROMETHEUS_CLIENT_MAX_QUEUE_SIZE']
client_thread_sleep	How often to sleep the worker thread that manages the client buffer (seconds)	0.5	ENV['PROMETHEUS_CLIENT_THREAD_SLEEP']
puma_collection_frequency	How often to poll puma collection metrics (seconds)	30	ENV['PROMETHEUS_PUMA_COLLECTION_FREQUENCY']
server_host	The host to run the exporter on	"0.0.0.0"	ENV['PROMETHEUS_SERVER_HOST']
server_port	The port to run the exporter on	9394	ENV['PROMETHEUS_SERVER_PORT']
server_thread_pool_size	The number of threads used for the exporter server	3	ENV['PROMETHEUS_SERVER_THREAD_POOL_SIZE']
process_name	What the current process name is (used in logging)	"unknown"	ENV['PROCESS']
railtie_disabled	Opt out flag for Railtie; use Bigcommerce::Prometheus::Instrumentors::Web.new(app: Rails.application).start in your app's code to start it up yourself	0	ENV['PROMETHEUS_DISABLE_RAILTIE']
resque_per_job_metrics_enabled	Enable per-job queue-latency and perform-duration histograms (parent-side, no synchronous flush)	0	ENV['PROMETHEUS_RESQUE_PER_JOB_METRICS_ENABLED']

Custom Collectors

To create custom metrics and collectors, simply create two files: a collector (the class that runs and collects metrics), and the type collector, which runs on the threaded prometheus server and

Type Collector

First, create a type collector. Note that the "type" of this will be the full name of the class, with TypeCollector stripped. This is important later. Our example here will have a "type" of "app".

class AppTypeCollector < ::Bigcommerce::Prometheus::TypeCollectors::Base
  def build_metrics
    {
      honks: PrometheusExporter::Metric::Counter.new('honks', 'Running counter of honks'),
      points: PrometheusExporter::Metric::Gauge.new('points', 'Current amount of points')
    }
  end

  def collect_metrics(data:, labels: {})
    metric(:points).observe(data.fetch('points', 0))
    metric(:honks).observe(1, labels) if data.fetch('honks', 0).to_i.positive?
  end
end

There are two important methods here: build_metrics, which registers the different metrics you want to measure, and collect_metrics, which actually takes in the metrics and prepares them to be rendered so that Prometheus can scrape them.

Note also in the example the different ways of observing Gauges vs Counters.

Collector

Next, create a collector. Your "type" of the Collector must match the type collector above, so that bc-prometheus-ruby knows how to map the metrics to the right TypeCollector. This is inferred from the class name. Here, it is "app":

class AppCollector < ::Bigcommerce::Prometheus::Collectors::Base
  def honk!
    push(
      honks: 1,
      custom_labels: {
        volume: 'loud'
      }
    )
  end

  def collect(metrics)
    metrics[:points] = rand(1..100)
    metrics
  end
end

There are two types of metrics here: on-demand, and polled. Let's look at the first:

On-Demand Metrics

To issue an on-demand metric (usually a counter) that then automatically updates, in your application code, you would then run:

app_collector = AppCollector.new
app_collector.honk!

This will "push" the metrics to our AppTypeCollector instance, which will render them as:

# HELP ruby_honks Running counter of honks
# TYPE ruby_honks counter
ruby_honks{volume="loud"} 2

As you can see this will respect any custom labels we push in as well.

Polling Metrics

Using our same AppCollector, if you note the collect method: this method will run on a 15 second polled basis (the frequency of which is configurable in the initializer of the AppCollector). Here we're just spitting out random points, so it'll look something like this:

# HELP ruby_points Current amount of points
# TYPE ruby_points gauge
ruby_points 42

Registering Our Collectors

Each different type of integration will need to have the collectors passed into them, where appropriate. For example, if we want these collectors to run on our web, resque, and hutch processes, we'll need to:

::Bigcommerce::Prometheus.configure do |c|
  c.web_collectors = [AppCollector]
  c.web_type_collectors = [AppTypeCollector.new]
  c.resque_collectors = [AppCollector]
  c.resque_type_collectors = [AppTypeCollector.new]
  c.hutch_collectors = [AppCollector]
  c.hutch_type_collectors = [AppTypeCollector.new]
end

Custom Server Integrations

For custom integrations that initialize their own server, you'll need to pass your TypeCollector instance via the .add_type_collector method on the prometheus server instance before starting it:

server = ::Bigcommerce::Prometheus::Server.new
Bigcommerce::Prometheus.web_type_collectors.each do |tc|
  server.add_type_collector(tc)
end

# and for polling:

AppCollector.start

License

Copyright (c) 2019-present, BigCommerce Pty. Ltd. All rights reserved

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.