Erron Austin wrote:
In load testing our portal on Linux, Java 1.4.x (not quite sure of the
Ram), I was throwing concurrent 400 users. An issue that I'm seeing is
that I keep getting this message in the logs for each page:
..........
DEBUG (2007-04-06) 12:07.40:905 [sitemap] (/home/ese)
http-0.0.0.0-8080-105/DefaultComponentFactory: ComponentFactory
decommissioning instance of
org.apache.cocoon.portal.acting.PreparePortalAction.
Typically, if I was seeing this, I would pool the generator or
transformer. But this is an Action, which doesn't appear to be poolable.
Question #1, is this a performance bottleneck?
Probably, yes.
Ouestion #2, if so, what is the solution?
Just did some cursory web searching, and from what I found there are two
possibilities:
1) Just as you said, make your Action poolable. AFAICT all you have to
do is make your Action class implement the Poolable interface.
2) Make your Action class implement the Threadsafe interface, which has
the effect of making the same instance be used for all requests, which
is even faster than pooling. You have to make sure your code is
actually thread safe, of course. ;-)
See:
http://mail-archives.apache.org/mod_mbox/cocoon-users/200402.mbox/[EMAIL
PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
