[NBLUG/talk] RT - fastcgi or modperl

Eric Eisenhart eric at nblug.org
Sat Dec 13 13:01:01 PST 2003


On Sat, Dec 13, 2003 at 09:35:29AM -0800, Mark Street wrote:
> I am going with modperl as my web environment with rt, but I used fastcgi with 
> zope but have not used it in quite awhile.
> 
> I was wondering if anyone has noted any pro's or cons to modperl vs fastcgi?

Given a choice between using mod_perl or fastcgi for RT, mod_perl is
probably the better bet, but fastcgi is a perfectly acceptable fallback. 
mod_perl integrates better with Apache's throttling model (IOW: you always
have exactly as many RT processes available as apache processes), but if
you're trying to run multiple applications through mod_perl you can run into
conflicts (some modules simply weren't designed to be used from totally
distinct applications).  Also, mod_perl on apache2 hasn't hit "mod_perl 2.0"
yet, and a lot of larger applications (RT) have refused to try to develop
against a moving target.  I believe apache2+mod_perl2(1.99) is still an
unsupported and potentially very buggy setup configuration for RT.

If you use fastcgi, make sure to set the "-processes" option to at least
match the number of simultaneous users you'll have.  Take email injections
into account if you're talking about RT3.  It defaults to one process, which
will serialize access and quickly annoy you a lot if there's more than one
person using it.

The main advantage of fastcgi is that it's a lot easier to set up.  You're
not messing with the apache setup so much, you don't have to worry about
what other mod_perl applications are doing, etc.

You could alternately set up mod_perl on a separate apache instance on a
separate port just for RT and use mod_proxy to make it look like its on the
same apache instance, but that's kinda the same fundamental model as
fastcgi, only with more work.
-- 
Eric Eisenhart
NBLUG Co-Founder & Director-At-Large
The North Bay Linux Users Group
http://nblug.org/
eric at nblug.org, IRC: Freiheit at freenode, AIM: falschfreiheit, ICQ: 48217244



More information about the talk mailing list