Mongrel users - Jun 2006 - Bare naked server

Thanks, Zed, for mongrel :!)

I''ve been on TextDrive for some time running my sites behind apache
w/lighttpd. Once it got running it was cool enough. When it ran. And
there was still the webmin dance and all.

So I decided to set things up on my little dedicated redhat server -
since I''d had so much success locally w/mongrel. Fast. Simple. It has
been great and I managed to move all my sites into it and off TD.
Smooth as glass thus far.  So, I decided I did not need no stinking
apache ;-)

Zed said :
>Mongrel was
>specifically designed to be run as a non-privileged user on a higher
>port in order to avoid these kinds of things.  In reality if you''re
>running Mongrel you should put it behind another web server and let that
>web server defend port 80 for you.
Doh. My understanding was that running a server in front of apache was
primarily because apache(other) was good/better at serving static
files. But since my performance thus far has been good(much faster now
than apache/mod_ruby) I''ve not sweat it.(thinking in the future
I''d
serve images from another server if nec.)

Are there other reasons, besides load, that I might be stupid for
leaving mongrel running bare on port 80? Are the mongrel police going
to come get me?

I did have one interesting error that was being thrown because I was
unable to grab the host name for a request. For some reason
http:///w00tw00t.at.ISC.SANS.DFind: was landing on my port. I''m
sending :status => 404 now(rather than dieing, which is caught by
exception notifier) Are these the type of things I should be worried
about?

http://www.atlink.it/~conti/2006/03/04/w00tw00tatiscsansdfind-update/

Multiple incoming requests?  Mongrel (or, more specifically, Rails itself)
can only handle a single request at a time, right?

 - James Moore

Hmm, I did not know - that would be a good reason :) So, apache,
lighty and all divvy up the requests so mongrel does not have to?

On TD, I figured we were using apache/lighty so multiple ports could
be used - and therefor use  a shared ip, with a lot of instances of
lighty(which I was thinking played a role in the server dying so
much). I thought that was the main reason.