[rt-users] How to run multiple instances of RT

Rich West Rich.West at wesmo.com
Wed Sep 27 12:07:37 EDT 2006


It is possible to do it with the same web server and the use of virtual
hosts using FastCGI.  It isn't possible to use the same web server and
mod_perl.  If you want to do it via mod_perl, you need to run completely
separate httpd processes with completely separate configuration files.

It is do-able, and I've done it for customers in the past.  We've run it
with fastcgi for a LONG time, but, with recent installations, we began
to have quirky problems with mod_fastcgi and Apache.

IF the RT installations are not hugely busy, you could always use the
"standalone_httpd" that is included with RT.  You can, say, run it on
port 8010, then, in your web server's configuration file, use ProxyPass
and ProxyReversePass within a VirtualHost to make it accessible via
separate your existing httpd installation and via the normal port 80
without the fear of any port conflicts.

I know that sounds pretty funky. As I read it, it didn't come out all
that clearly, but I cannot think of another way to put it..

The "standalone_httpd" app that is included with RT is _not_ threaded
and suffers from a large performance hit when hit heavily.  FastCGI is
much, much faster. :)

-Rich
> First note that you really don't need different instances
> of RT to handle different projects.  You can create
> different queues with different email aliases and different
> groups of people allowed to see them and the projects won't
> conflict with each other.  If you do need different instances
> for administrative reasons you can do it easily with
> fastcgi or with mod_perl you can run different httpd
> instances bound to different ports or IP addresses.
>   




More information about the rt-users mailing list