Lin Sun wrote:
One quick way would be allow users to start a tomcat server from a
geronimo tomcat javaee5 assembly or little G tomcat assembly(say
geronimo.sh start tomcatOnly=true). It is possible to just launch
the tomcat server, and read the configuration files (conf/server.xml,
etc) and start a tomcat server from a geronimo tomcat assembly, by
using the Catalina.java provided by tomcat. But this server/app
would have no relationship with geronimo, other than using the jars
provided by the geronimo tomcat assembly. The deployed app would be
tracked only by tomcat, and not by geronimo. We should be able to
achieve this without adding any new jars.
If we need more than that, I can for seen the following issues that
need investigation -
1. We'll have to provide better server integration with tomcat and be
able to read the server configuration from tomcat's server
configuration files, along with using config.xml configurations.
This would be an absolute minimal requirement. Would this be really
difficult?
2. We'll have to migrate user's app automatically for the user, when
the user's app is a bit complicated that contains any need to require
a geronimo-web.xml
This is where things get more interesting.... lots of permutations and
edge cases to consider.
Lin
On Mon, May 18, 2009 at 3:49 PM, Bill Stoddard <wgstodd...@gmail.com> wrote:
I know G can't consume tomcat configs today, but is this a feature that
could be developed for G 2.2?
Say I have an application successfully deployed and running under Tomcat..
wouldn't it be nice if I were able to drop the tomcat server config into a
Geronimo-Tomcat assembly, start the server, deploy the app and be up and
running in seconds. I'm talking about a seamless, zero effort/zero touch
migration from Tomcat to a Geronimo-Tomcat assembly. Is it possible? If
not, what simplifying assumptions could be made to 'mostly' achieve a
zero-touch migration?
What are the primary challenges with consuming a tomcat config unchanged
with a G-Tomcat assembly? Same Q's apply for Jetty... what's 'doable'
with-in reason?
Thanks,
Bill