Application Heartbeat

Many a times developers use simple monitoring solutions like Uptime Robot or Pingdom to ping servers to check if they are working fine. But when supplying urls to monitoring services, we need to make sure that they are indeed the right url to for the job.

Most developers supply a generic root url to check if things are fine.

Expect to hit the url every n seconds

If the url does very heavy work and has lots of database calls, there are chances that the monitoring system can burden the server. Performing expensive operations is not really required for monitoring systems. They can even bring down the server.

Does this check every system in the stack?

May a times the front page might just be served from the cache without connecting to the database. So expecting an alert email for a database downtime with a simple monitor might just not work.

What should you really check?

Check if all the components of your stack are working fine.

• Most applications have a database. Check if database connections are working.
• Some applications have a caching layer. Check if you cache is reachable.

This would translate to the following code:

class HeartBeat
  class SystemsError < StandardError ; end

  def run
    raise SystemsError unless check_db && check_cache
  end

  private

  def check_db
    User.where('id  > 1').exists?
  end

  def check_cache
    Rails.cache.write('heartbeat', DateTime.current)
      && Rails.cache.read('heartbeat')
  end
end

You can use this model in a controller and mount it in your app at /appbeat or any url of your wish!

class HeartbeatController < ActionController::Metal
  def index
    self.response_body = HeartBeat.new.run
  end
end

Happy monitoring!