donnoman / health-plugin

Health / Rails Plugin (Requires Rails >= 3.1.0)

Health Plugin

This is a Rails (>= 3.1.0) plugin to enable functionality for monitoring of a Rails application.

No modification of the host Rails application should be required; add the gem to your bundler Gemfile and configure the health-plugin via an initializer in your host application.

Initializer

You will need an initializer in your mail Rails application to configure the health-plugin. Each callback maps to a health-plugin URI for the host application. Each callback Proc should return a Hash formatted as follows:

1. A header value; this is rendered as a header in the HTTP response for the corresponding health-plugin URI.

2. A body value; this is rendered as the body of the HTTP response for the corresponding health-plugin URI.

3. A status value; this is rendered as the HTTP response code for the corresponding health-plugin URI.

The following callbacks exists: state branch describe env ident ping timestamp

All headers are prefixed with the value from config.prefix. In the example below; the env callback, for example, would end up with X-Company-Env as it's header.

Here is an example initializer for the health-plugin:

module Health
  class << self

    def ping
      ping = Array.new
      ping << disabled
      ping << galaxy
      ping << redis
      ping.compact.join('|')
    end

    def disabled
      File.exists?(File.join(Rails.root, ".disabled")) and return "PANG DISABLED"
      nil
    end

    def redis
      Redis.current.info and return "PONG REDIS"
      "PANG REDIS"
    rescue Exception => e
      "PANG REDIS #{e.inspect}"
    end

    def galaxy
      site = URI(SomeApp::Base.site.to_s.dup)
      site.path = "/site/ping"
      Array(200..299).include?(Net::HTTP.get_response(site).code.to_i) and return "PONG SOMEAPP"
      "PANG SOMEAPP"
    rescue Exception => e
      "PANG SOMEAPP #{e.inspect}"
    end

  end
end

HealthPlugin.config do |config|
  config.prefix = "X-Company"
  config.mounts += %w( site )

  config.callbacks.ping = Proc.new do
    ping = Health.ping
    status = (ping =~ /PANG/ ? 501 : 200)
    {
      header: ping, d
      body: ping,
      status: status
    }
  end

  config.callbacks.state = Proc.new do
    ping = Health.ping
    state = (ping =~ /PANG/ ? 'OFF' : 'ON')
    {
      header: state,
      body: nil,
      status: 204
    }
  end

end

Your loadbalancer can then call these endpoints to make intelligent forwarding decisions. You can also call these directly with curl for diagnostic purposes.