"Auto" Scale Heroku Cedar instances with rake tasks

gistfile1.txt

I think this is a nice option for apps that are mostly used during particular hours and aren't really ready 
for a more intense auto-scaling thing. Hirefire (https://github.com/meskyanichi/hirefire) likely does a 
lot more, but I didn't really understand how it worked and don't really need something that fancy right now. 
I suspect it requires at least one worker itself, but I don't really know.

Add an API key to your app's Heroku config
heroku config:add HEROKU_API_KEY=your_key (from https://api.heroku.com/account )

Add the above rake task. Change 'YOUR-APPS-NAME' to indigo-butterfly-77 or whatever your app's name is.

Put some scale instructions in Heroku scheduler:
rake scale[web=2,worker=1] at 6:00 am
rake scale[web=1,worker=1] at 6:00 pm
rake scale[web=0,worker=0] at 1:00 am
rake scale[web=1]          at 5:00 am

This was an easy little thing to reduce processor hours. 
I'm thinking about using it to set web=0 on my free apps at night, 
so I have some free hours to burn on workers/etc during the day.

scale.rake

desc "Scale up/down dynos" 
task :scale, :arg1, :arg2, :arg3, :arg4, :arg5, :arg6, :arg7, :arg8 do |t, args|
  args.entries.collect(&:last).each do |scale_instruction|
    process_type, number_of_instances = scale_instruction.split('=')
    scale(process_type, number_of_instances) if number_of_instances =~ /\A\d+\Z/
  end
end

def scale(process_type,number_of_instances)
  curl_command = <<-eos
    curl -H "Accept: application/json" \
      -u :#{ENV['HEROKU_API_KEY']} \
      -d "type=#{process_type}" \
      -d "qty=#{number_of_instances}" \
      -X POST https://api.heroku.com/apps/YOUR-APPS-NAME/ps/scale
  eos
  system(curl_command.gsub(/\s+/,' '))
end

To allow this rake task to be deployed to both 'staging' and production apps on Heroku, we changed the line:

-X POST https://api.heroku.com/apps/YOUR-APPS-NAME/ps/scale

to read

-X POST https://api.heroku.com/apps/#{ENV['HEROKU_APP_NAME']}/ps/scale