Home / server / Questions / 212775
In Process
Ramon Tayag
Asked: 2010-12-15 23:05:31 +0800 CST

Apache/Passenger: not following PassengerMaxPoolSize/PassengerMaxInstancesPerApp?

  • 772

If I'm correct when I think that a line of

30193 ramon     16   0  170M  119M  3628 S  0.0 11.6  0:08.47 Rack: /var/www/salcedoauctions/current

in htop means that there's one spawn of the salcedoauctions app, then how come there are so many more spawns than I've specified?

Here's my httpd.conf:

....
PassengerMaxPoolSize 4
PassengerMaxInstancesPerApp 2
....

Here's the semi-complete htop:

  PID USER     PRI  NI  VIRT   RES   SHR S CPU% MEM%   TIME+  Command
30193 ramon     16   0  170M  119M  3628 S  0.0 11.6  0:08.47 Rack: /var/www/salcedoauctions/current
30197 ramon     18   0  170M  119M  3628 S  0.0 11.6  0:02.08 Rack: /var/www/salcedoauctions/current
30198 ramon     18   0  170M  119M  3628 S  0.0 11.6  0:01.96 Rack: /var/www/salcedoauctions/current
30199 ramon     18   0  170M  119M  3628 S  0.0 11.6  0:01.96 Rack: /var/www/salcedoauctions/current
30067 ramon     15   0  167M  117M  3648 S  0.0 11.5  0:08.77 Rack: /var/www/salcedoauctions/current
30202 ramon     16   0  167M  117M  3648 S  0.0 11.5  0:04.55 Rack: /var/www/salcedoauctions/current
30203 ramon     16   0  167M  117M  3648 S  0.0 11.5  0:03.43 Rack: /var/www/salcedoauctions/current
30204 ramon     16   0  167M  117M  3648 S  0.0 11.5  0:03.32 Rack: /var/www/salcedoauctions/current
28454 ramon     16   0 73764 51732  2520 S  0.0  4.9  0:00.66 Rack: /var/www/unstilted/current
.....

Some notes:

  • Yes, I've restarted apache so it can reload the config (several times, in fact, since the server was running out of memory)
  • CentOS 5.5
  • Ruby EE 1.8.7
  • Rails 3.0.1

  • Passenger 3.0.0

  • UPDATE #1 Doing passenger-status shows the numbers I expect to see though though. I suppose I just understood it wrong. The reason I'm asking is because my 1GB memory is getting used up and sometimes the server grinds to a halt. It seems that those processes are hogging the RAM.

  • UPDATE #2: I asked around in #rubyonrails and I was asked if any of the apps were forking any processes. Both apps don't fork anything, but I do use Dragonfly to handle my images and attachments. However, I don't know how I can pinpoint if this is the cause of the problem.
  • UPDATE #3 (Oct 14, 2011): I upgraded to Ruby 1.9.2, Rails 3.1.x, dragonfly 0.9.3, ImageMagick 6.7; and I'm still getting the same problem. It's not Dragonfly anymore because the newer versions don't use rmagick. It does a command line call to convert images.
ruby-on-rails phusion-passenger ruby-rack

1 Answers

  1. For tracking down these problems, running passenger-status -v on the server will give a good overview of what is actually active. Likewise, passenger-memory-status will give you a nice overview of mem usage.

    Regarding versions, consider upgrading to passenger > 3.0.7 - I had similar problems with rails/dragonfly and memory leaks under early 3.0 releases. ImageMagick 6.6.0 and rmagick 2.13.1 have been working well for a while in production.

    • 0