Hi Diego,
well passenger is quite aggressive with the threads spawned by hosted
rails applications.
There is this mention about threads in the Passenger guide :
But I never had time to study that.
> I am thinking about two posibilities:
>
> * Using nginx as a request dispatcher to serve requests torwards a mongrel
> cluster. This is a good deployment for rails applications and it is surely
> useful with concurrent requests. But I don’t know how to make mongrels share
> a single workflow engine instance (ruote instance).
Others, like Kenneth, have taken the ruote-rest way (and later
ruote-kit) to let the workflow engine sit in a web application behind
the front web application[s].
> * Using a JRuby/Tomcat deployment for my rails app. I don’t have experience
> in this architecture, and thus I don’t know if it is valid for serving
> concurrent requests.
>
> What is the best choice? Is there another posibility?
Do you really need serving "concurrent requests" ? Do you really have
that much users ? With bottleneck patterns ?
Have you taken a look at Thin ? It's a successor of Mongrel :
http://code.macournoyer.com/thin/
Best regards,
--
John Mettraux - http://jmettraux.wordpress.com
Others, like Kenneth, have taken the ruote-rest way (and later
On Wed, Jun 3, 2009 at 12:54 AM, Diego Moreno <dmo...@dit.upm.es> wrote:
> I am thinking about two posibilities:
>
> * Using nginx as a request dispatcher to serve requests torwards a mongrel
> cluster. This is a good deployment for rails applications and it is surely
> useful with concurrent requests. But I don’t know how to make mongrels share
> a single workflow engine instance (ruote instance).
ruote-kit) to let the workflow engine sit in a web application behind
the front web application[s].