Link Details

Is that accurate? By default Ruby on Rails is *single-threaded* and can only handle a single request at a time?

For a web application framework, that just seems kind of... ridiculous.

This article seems useful as a rough outline for setting up actual multi-server clustering, but I'm still looking at it askance just based on those opening statements. I'm no RoR expert -- anyone want to clarify?

The default WEBrick server is single-threaded for sure. I'm not so sure if mongrel_rails is single-threaded, however it can work on only on process. Recently, I had to create a webconsole with ping, mtr and similar utilities. While working with only one mongrel_rails instance if someone issued a "system ping" - the application server wasn't able to serve any other requests untill that sysmem ping finished. I solved that problem by setting up a mongrel_rails cluster with apache as described in that tutorial.

It's the Ruby interpreter that is single threaded. More properly, it's threading is not reliable so it needs to be treated as being single threaded. I use a mongrel cluster and the pound reverse proxy which feeds requests to the "pack of mongrels".

hmm I got this problem and well glad you gave this cool solution.