heroku.rake for CrowdTrail https://www.slideshare.net/shikendon/crawler-from-zero

heroku.rake

require "heroku-api"

# heroku = Heroku::API.new(:api_key => API_KEY)                           # use API Key
# heroku = Heroku::API.new(:username => USERNAME, :password => PASSWORD)  # use username and password
# heroku = Heroku::API.new(:headers => {'User-Agent' => 'custom'})        # use custom header

# NOTE: You can leave out the :api_key if ENV['HEROKU_API_KEY'] is set instead.
# https://github.com/heroku/heroku.rb

namespace :heroku do
  namespace :webs do
    desc "Restarts all Heroku webs"
    task :restart do
      heroku = Heroku::API.new
      response = heroku.get_ps(ENV["HEROKU_APP_NAME"])
      webs = response.body.select { |item| item["process"] =~ /web/ }

      webs.each do |web|
        process = web["process"]
        pretty_state = web["pretty_state"]
        puts "Restarting #{process} (#{pretty_state}) . . ."
        heroku.post_ps_restart(ENV["HEROKU_APP_NAME"], "ps" => process)
      end
    end
  end

  namespace :workers do
    desc "Restarts all Heroku workers"
    task :restart do
      heroku = Heroku::API.new
      response = heroku.get_ps(ENV["HEROKU_APP_NAME"])
      workers = response.body.select { |item| item["process"] =~ /worker/ }

      workers.each do |worker|
        process = worker["process"]
        pretty_state = worker["pretty_state"]
        puts "Restarting #{process} (#{pretty_state}) . . ."
        heroku.post_ps_restart(ENV["HEROKU_APP_NAME"], "ps" => process)
      end
    end

    desc "Restarts the Heroku worker that matches the rules"
    task :try_restart, [:max_running_time] do |_t, args|
      args.with_defaults(max_running_time: 600)
      heroku = Heroku::API.new
      response = heroku.get_ps(ENV["HEROKU_APP_NAME"])
      workers = response.body.select { |item| item["process"] =~ /worker/ }

      workers.each do |worker|
        process = worker["process"]
        elapsed = worker["elapsed"] # Seconds
        pretty_state = worker["pretty_state"]
        next if elapsed < args[:max_running_time]
        puts "Restarting #{process} (#{pretty_state}) . . ."
        heroku.post_ps_restart(ENV["HEROKU_APP_NAME"], "ps" => process)
        break
      end
    end

    desc "Auto scaling the Heroku workers count according to Sidekiq jobs count"
    task :auto_scaling, [:WORKER_NAME] => :environment do |_t, args|
      args.with_defaults(WORKER_NAME: "worker")
      APP_NAME = ENV["HEROKU_APP_NAME"]
      WORKER_NAME = args[:WORKER_NAME]

      heroku = Heroku::API.new
      queues = Sidekiq::Queue.all # 所有 Sidekiq 佇列名稱物件
      queues_size = queues.map { |queue| Sidekiq::Queue.new(queue.name).size }.inject(0, :+) # 剩餘任務數量

      # 平均一個 2X dyno 每分鐘可以處理 600 個 jobs
      # 目前設定每個小時 50 分的時候會重新檢查一次有無尚未 parse 的 project_log
      # 因此所有 jobs 都必須在 45 分以前完成否則會影響到下一個小時的任務效能
      now_minutes = Time.now.strftime("%M").to_i # 目前分鐘
      left_minutes = now_minutes.between?(0, 45) ? 45 - now_minutes : 0 # 剩餘分鐘
      workers_size = queues_size / 500 / [left_minutes, 1].max # 剩餘任務數量 / 處理速度 / 剩餘分鐘
      workers_size = 1 if workers_size < 1 # 最少開一個 worker
      workers_size = 10 if workers_size > 10 # 最多開 10 個 worker
      puts "Scaling #{WORKER_NAME} dyno count to #{workers_size}"
      heroku.post_ps_scale(APP_NAME, WORKER_NAME, workers_size)
    end
  end
end