When shit hits the fan in your Redis db, Redis Health lets you know before your customers do

Install:

gem install 'redis-health'

Use:

require 'redis'
require 'redis_health'
require 'pony'

RedisHealth.configure do
  watch('worker count fell below 2 for more then 1 minute') do
    v = redis.scard('resque:party:workers')
    {triggered: v < 2, value: v}
  end

  watch('redis connected client count fell below 3 for more then 30 seconds', 30) do
    client_count = redis.info['connected_clients'].to_i
    {triggered: client_count < 3, value: client_count}
  end

  watch('redis connected client count less then resque registered worker count for more then 30 seconds', 30) do
    worker_count = redis.scard('resque:party:workers')
    client_count = redis.info['connected_clients'].to_i
    {triggered: client_count < worker_count, value: "client count: #{client_count}, Worker count #{worker_count}"}
  end

  watch('HOLY **** TEH FAIL LOGZ IZ HUGE', 1) do
    v = redis.llen('resque:party:failed')
    {triggered: v > 1000, value: v}
  end

  notify do |notices|
    #TEXT EMAIL FAX POSTCARD TWEET GOES HERE
    Pony.mail(:to => '[email protected]', :from => '[email protected]', :html_body => '<h1>Some stuff seems to have failed, take a look pretty please</h1> ' + notices.join(', '))
  end
end

redis = Redis.new

#runs watches and sleeps for 1 second in a loop
RedisHealth.new(redis).run

It calls your notifier block if any watches have been triggered for their specified number of seconds, passing it a list of triggered messages such as:

['HOLY **** TEH FAIL LOGZ IZ HUGE, value is now 100000', 'worker count fell below 2 for more then 1 minute, value is now 0']

And calls it again if a watch ceases to trigger, with a list of triggered messages such as:

['HOLY **** TEH FAIL LOGZ IZ HUGE no longer in effect, value is 10', 'something else failed, value is 10000000000']