I'm going through an problem, and that i can't determine what is happening. There's a distinctive website running about this VPS with Apache 2 and Passenger. Some occasions the application takes almost one minute do respond, and throughout these moments the development.record keeps running and serving pages within 50ms to 700ms. Newrelic does not show these lengthy response time demands also.

My monitoring system inspections the home page every minute, and logs the response time. Some periods of three to five minutes, if this shows 10,000ms to 100,000ms, I'm able to access the web site with extremely fast response.

I have a similar application running on different VPS with similar size, one-time all of them with 5x more visits, and keeps 500ms average response.

This can be a 1 Gb VPS with:

  • Debian 6
  • MySQL
  • Apache 2
  • RVM
  • Passenger 3..3
  • Ruby Enterprise Edition 2011.03
  • Sphinx indexing every a few minutes
  • Memcached

Here's my passenger.conf:

<IfModule mod_passenger.c>
  PassengerRoot /usr/local/rvm/gems/ree-1.8.7-2011.03@v2v/gems/passenger-3.0.3
  PassengerRuby /usr/local/rvm/wrappers/ree-1.8.7-2011.03@v2v/ruby
  PassengerPoolIdleTime 3600
  PassengerMaxInstancesPerApp 3
  PassengerMaxRequests 1000
  PassengerUseGlobalQueue on

Any ideas?