Corin Moss wrote:Keep in mind that the caching system is currently in a state of flux given that we are trying to move away from Jisp. A lot of the alternatives manage the paging aspect that now involves the persistent store and store janitors themselves.
I've just been through a several week excercise tuning
a fairly complex Cocoon implementation (for use by '000's
of concurrent users) I've picked up quite a lot of useful
snippets of info as I've delved into lots of different
areas of the code. I'd be happy to share these, if you'd
like to outline the steps you've taken, I'll reply with
other suggestions from my experience. I'm sure others will
do the same.
I am going through the cocoon.xconf and making sure that the
parameters and their descriptions do represent what is indicated
in the code. I am also fine-tuning the core Cocoon documentation,
starting with the concepts/mrustore.html, persistence.html, caching.html, and storejanitor.html, and performancetips.html
And i am cleaning the fluff from the Wiki page CocoonPerformance.
If you have any patches for those core xdocs or for javadoc in the code, then please send them via Bugzilla and i will integrate.
If you cannot manage that, then perhaps just dump your collection of notes as a Wiki page ...{{{ list of dot points }}} ... and we can glean stuff.
Also i am considering adding a couple of docs to describe both small and large scale performance tuning scenarios.
A related observation: I consider the current Jisp version unuseable for any production site. The problem outlined here http://issues.apache.org/bugzilla/show_bug.cgi?id=26753 is very serious and has forced me to use the EHCache store for some of our clients instead (The only useable alternative besides perhaps JCS (haven't tried to use it yet) ATM).
-- Unico
