> > i have a strange effect when I run code inside of cocoon, it seems to
> > slowdown more then 10 times!! :(
> > I call the main-method from my test class the command line.
> > - The method does lookup an Entity EJB on JBoss server.
> > - Get a Property from the Bean
> > - Serialize the Bean using Betwixt
> > Then I tested the same by callingthe same method from an cocoon action.
It
> > slowes down dramatically:
> > |Command Line | Cocoon Action
> > ----------+-------------+--------------
> > Lookup | 144 ms | 2825 ms
> > Getter | 17 ms | 3560 ms
> > Serialize | 708 ms | 73342 ms
> > Has anybody an idea what causes this slowdown, or anybody seen something
> > similar before?
> > Any help would be apriciated.
>
> How much memory is allocated for your jvm in your test environment? Can
> you tell if memory appears to be paging out to disk during your cocoon
> test?
I have adjusted the jvm startup parameters, so the command-line test starts
up with "-server -Xms1200m -Xmx1200m", the same as tomcat/cocoon.
> Is the code literally identical - if so, how are you calling it from
> your action?
Yes, it's the same class. In cocoon from my action I call the method
"TestClient.main(new String[] {});", the same as from command line. The
output is written to stdout, under tomcat it's in catalina.out-file.
That's what makes the problem so strange, it's exactly the same code!
> What else is "installed" in Cocoon - in other words, what blocks, etc.
> (trying to get an idea if classloading is an issue).
That was my first Idea, I thought, perheaps reflection slows down because of
all the classes loaded with the jboss-client/tomcat/cocoon-jars. So I added
the complete classpath to my command line test. The commandline test slows
down also, but only about 10% or so, not 1000%!
Perheaps the problem has nothing to do with cocoon, perheaps it's about the
tomcat classloader, that is what I will test next, I will run the
main-method of the test class from within an servlet.
> Does this happen over and over, or only soon after startup of cocoon?
There is no perfomance increase, when I request the pipeline (containing the
test action) a second, third... time, so it seems that it has nothing to do
with the cocoon startup.
Any Ideas?
Regards
Christoph
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
