Heroku Reporting Dynos

1-README.md

Setup

• Add a new worker to your Procfile. In this gist I used reporter
• Add the Heroku Platform gem to your Gemfile
• Create the HerokuReporterDyno module in lib or any other auto-loadable path
• Create two jobs. A caller (Reporting::GeneratorJob) and a runner (Reporting::RunnerJob)
• Deploy
• Once deployed, log into Heroku and change the newly added reporter dyno to a Performance M or Performance M dyno

How this all works

Simply run the caller job Reporting::GeneratorJob.perform_later and it will scale the reporter dyno up to 1 and run the runner job. Once the runner job is complete, it will scale the reporter dynoe back to 0.

Why do this?

Simple... To run very heavy reporting aggregation jobs using high spec dynos at affordable costs

Gemfile

gem 'platform-api'

generator_job.rb

# app/jobs/reporting/generator_job.rb

class Reporting::GeneratorJob < ActiveJob::Base

  queue_as :default

  def perform
    HerokuReporterDyno.add
    Reporting::RunnerJob.perform_later
  end

end

heroku_reporter_dyno.rb

# lib/heroku_reporter_dyno.rb

module HerokuReporterDyno
  
  def self.add
    heroku = PlatformAPI.connect_oauth(ENV['HEROKU_API_KEY'])
    heroku.formation.update(ENV['HEROKU_APP_NAME'], 'reporter', { quantity: 1 })
  end

  def self.remove
    heroku = PlatformAPI.connect_oauth(ENV['HEROKU_API_KEY'])
    heroku.formation.update(ENV['HEROKU_APP_NAME'], 'reporter', { quantity: 0 })
  end
  
end

Procfile

web:      ...
worker:   bundle exec sidekiq -C config/sidekiq.yml
reporter: bundle exec sidekiq -c 25 -q reports

runner_job.rb

# app/jobs/reporting/runner_job.rb

class Reporting::RunnerJob < ActiveJob::Base

  queue_as :reports

  def perform
    do_some_heavy_reporting_stuff
    HerokuReporterDyno.remove
  end
  
end