Myles Braithwaite

• I was using four Mongrel process per site and had four web sites so in total I was running about 16 Ruby process.
• Mongrel also wasn’t hosting any static content (which i made sure of with one of these config.action_controller.asset_host in the config/environments/production.rb file).
• I used Apache2 mod_proxy balancer for the four mongrel instances.
• We didn’t add anymore RAM.
• Also to be noted two of those sites were running Radiant and one was running Typo.
  • The other one I developed and had aggressing cache settings.

Also the reason I switch was not because of the memory usage but because Mongrel is so slow and FastCGI was not any better. I actually like developing in Rails and even Ruby! Furthermore I do not really believe that this is a Rails problem but a Mongrel issue.

If you want more information about my setup you can message me on Freenode IRC with the username mylesbraithwaite.

Responses to Comments

Silveira Neto

But, are you serving less pages? Less visits?

I have received 594 unique visits (or at-least awstats tells me) on my previous post and the memory usage is still the same.

Kaaren

So you’re using less RAM now. Is that what you wanted to point out?
How many sites?
(had to fix that)

Four

sbraford

what applications were/are you running in rails and now django? (i see below this site is now powered by flother)
for this number of visits, this really isn’t the sweet spot for Rails. (sadly)
i’m experiencing this issue myself with several Rails apps all running the same software, but each receiving very little traffic individually. the mongrels chew up a lot of RAM/cpu.
but i still wouldn’t go back and rewrite my 50+ table app in python! :)

Nothing else; actually there are even more applications running on the server now than before. flother is the design I am using.