Rails - May 2005 - Scary FastCGI settings?

After experiencing the frustrating "Oops! Someone''s making a
request!
I''d better start a couple Ruby processes!" a couple times with
Apache
and FastCGI, I eventually put the following settings in my config:

FastCgiConfig -killInterval 3600 -minProcesses 0 -restart -autoUpdate

However, for some reason this seems to die without warning and I
cannot tell why. The platform is Linux, and I just installed the whole
Ruby/FastCGI/Gems/Rails deal a few weeks ago from source, so it should
be current.

What are good options to pass to FastCgiConfig for a server that will
keep at least one Ruby process ready to go at all times in case of a
traffic slump? What options work for you? What kind of options would I
expect to see in a config for a site like Basecamp or Backpack?

Sincerely,

Tom Reinhart
tom-V0YqjHVuocLQT0dZR+AlfA@public.gmane.org
http://AllTom.com/

> What are good options to pass to FastCgiConfig for a server that will
> keep at least one Ruby process ready to go at all times in case of a
> traffic slump? What options work for you? What kind of options would I
> expect to see in a config for a site like Basecamp or Backpack?
<IfModule mod_fastcgi.c>
   FastCgiIpcDir /tmp/fcgi_ipc
   FastCgiServer /app/public/dispatch.fcgi -initial-env  
RAILS_ENV=production -processes 15 -idle-timeout 60
</IfModule>

That would be how the Basecamp and Backpack definitions look. The key  
is to use a static server definition, like the one I have with  
FastCgiServer.
--
David Heinemeier Hansson
http://www.loudthinking.com -- Broadcasting Brain
http://www.basecamphq.com   -- Online project management
http://www.backpackit.com   -- Personal information manager
http://www.rubyonrails.com  -- Web-application framework

On 5/24/05, Tom Reinhart <alltom-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
wrote:
> On 5/16/05, David Heinemeier Hansson
<david-OiTZALl8rpK0mm7Ywyx6yg@public.gmane.org> wrote:
> > > What are good options to pass to FastCgiConfig for a server that
will
> > > keep at least one Ruby process ready to go at all times in case
of a
> > > traffic slump? What options work for you? What kind of options
would I
> > > expect to see in a config for a site like Basecamp or Backpack?
> >
> > <IfModule mod_fastcgi.c>
> >    FastCgiIpcDir /tmp/fcgi_ipc
> >    FastCgiServer /app/public/dispatch.fcgi -initial-env
> > RAILS_ENV=production -processes 15 -idle-timeout 60
> > </IfModule>
> >
> > That would be how the Basecamp and Backpack definitions look. The key
> > is to use a static server definition, like the one I have with
> > FastCgiServer.
> 
> Making that change to my configuration seems to have helped a lot over
> the past two weeks or so. Thank you very much!
I admit, failing has taken a lot longer for the server to do, but it
has happened twice in the time that I have used this configuration. By
failing I mean getting the "script returned zero bytes" error in the
Apache logs, which won''t fix itself until I restart Apache manually as
root.

I''m in RoR ''development'' mode at the moment. Has
anyone noticed a
difference in Apache-FastCGI stability between development and
production modes? Does anyone know what exactly causes the "Rails
application failed to start" error? (Dead Ruby processes? Too many
Ruby processes? Exhausted server? Apache having a bad day?)

We''re preparing to preview the site to our sponsors, and launch soon
after, which is why I''m a bit nervous about the problem. If installing
LigHTTPd and serving Apache requests to it makes the application more
stable, and the overhead of forwarding from one server app to another
for every request is negligible (and I can find a nice guide for
setting it up), I may go that route, although it''s late in the game to
be changing server platforms now. There are other sites on the server
outside our control, which is why we cannot change from Apache 1.3x.

Thanks for any help.

Sincerely,

Tom Reinhart
tom-V0YqjHVuocLQT0dZR+AlfA@public.gmane.org
http://AllTom.com/