my.god

#
# God please,
#
# Generic Monitor
#
def generic_monitoring(w, options = {})
  w.interval = 30.seconds # default
  w.start_if do |start|
    start.condition(:process_running) do |c|
      c.interval = 10.seconds
      c.running = false
    end
  end

  w.restart_if do |restart|
    restart.condition(:memory_usage) do |c|
      c.notify = {:contacts => ["nofxx", "developers"], :priority => 1, :category => "mem"}
      c.above = options[:memory_limit]
      c.times = [3, 5] # 3 out of 5 intervals
    end

    restart.condition(:cpu_usage) do |c|
      c.notify = {:contacts => ["nofxx", "developers"], :priority => 1, :category => "cpu"}
      c.above = options[:cpu_limit]
      c.times = 5
    end
  end

  w.lifecycle do |on|
    on.condition(:flapping) do |c|
      c.to_state = [:start, :restart]
      c.times = 5
      c.within = 5.minute
      c.transition = :unmonitored
      c.retry_in = 10.minutes
      c.retry_times = 5
      c.retry_within = 2.hours
    end
  end
end


# Nginx
#
God.watch do |w|
  w.name = "nginx"
  w.uid = "root"
  w.gid = "root"
  w.group = "webapp"
  w.start = "/etc/rc.d/nginx start"
  w.stop = "/etc/rc.d/nginx stop" 
  w.restart = "/etc/rc.d/nginx restart"
  w.start_grace = 30.seconds
  w.restart_grace = 30.seconds
  w.pid_file = File.join("/var/run/nginx.pid")
  w.behavior(:clean_pid_file)
  generic_monitoring(w, :cpu_limit => 80.percent, :memory_limit => 800.megabytes)
end

# Nanites
#
for nanite in [:foo, :bar]
  God.watch do |w|
    w.name = nanite.to_s
    w.group = "nanites"
    w.uid = WEBAPP[:uid]
    w.gid = WEBAPP[:gid]
    w.pid_file = File.join(WEBAPP[:path], "tmp", "pids", "nanite.nanite-#{nanite}.pid")
    w.start = "nanite-agent --token #{nanite} --nanite #{WEBAPP[:path]}/nanite/#{nanite} --pid-dir #{WEBAPP[:path]}/tmp/pids --log-dir #{WEBAPP[:path]}/log  --daemonize"
    w.stop = "kill `cat #{w.pid_file}`"
    w.restart = "#{w.stop} && #{w.start}"
    w.start_grace = 20.seconds
    w.restart_grace = 20.seconds
    w.behavior :clean_pid_file
    generic_monitoring(w, :cpu_limit => 80.percent, :memory_limit => 800.megabytes)
  end
end

# PostgreSQL
#
God.watch do |w|
  w.name = "postgresql"
  w.uid = "root"
  w.gid = "root"
  w.group = "db"
  w.start = "/etc/rc.d/postgresql start" #w.start = "service postgresql start"
  w.stop = "/etc/rc.d/postgresql stop"   #w.stop = "service postgresql stop"
  w.restart = "/etc/rc.d/postgresql restart"
  w.start_grace = 30.seconds
  w.restart_grace = 30.seconds
  w.pid_file = File.join(WEBAPP[:db], "postmaster.pid")
  w.behavior(:clean_pid_file)
  generic_monitoring(w, :cpu_limit => 80.percent, :memory_limit => 800.megabytes)
end

# Tyrant
#
God.watch do |w|
  w.name = "tyrant"
  w.uid = "root"
  w.gid = "root"
  w.group = "db"
  w.start = "rake -f #{WEBAPP[:path]}/Rakefile tyrant:start"
  w.stop = "rake -f #{WEBAPP[:path]}/Rakefile tyrant:stop"
  w.restart = "rake -f #{WEBAPP[:path]}/Rakefile tyrant:restart"
  w.start_grace = 30.seconds
  w.restart_grace = 30.seconds
  w.pid_file = File.join(WEBAPP[:path], "tmp", "pids", "tyrant.pid")
  w.behavior(:clean_pid_file)
  generic_monitoring(w, :cpu_limit => 80.percent, :memory_limit => 800.megabytes)
end

# RabbitMQ
#
God.watch do |w|
  w.name = "rabbitmq"
  w.uid = "root"
  w.gid = "root"
  w.group = "db"
  w.start = "/etc/rc.d/rabbitmq start"
  w.stop = "/etc/rc.d/rabbitmq stop"
  w.restart = "/etc/rc.d/rabbitmq restart"
  w.start_grace = 30.seconds
  w.restart_grace = 30.seconds
  w.pid_file = File.join("/var", "run", "rabbitmq.pid")
  w.behavior(:clean_pid_file)
  generic_monitoring(w, :cpu_limit => 80.percent, :memory_limit => 800.megabytes)
end