Aaron/Anton,
We spent quite a bit of time performance tuning J2 recently...
specifically removing DB hits and optimizing for the
'guest' user. J2 spends a lot of time setting up user sessions for a
variety of reasons. So it is important to use a stateful
test harness to make sure that each hit is not starting a new session.
To be specific, one needs to use a HTTP cookies
aware test harness. As with most test suites, you need to determine how
often each "user" session will end up
grabbing a page. This can make a huge difference in how much load a
single session is going to put on the system.
We are always interested in performance results. I am not sure if most
of the improvements are in M4 or M3, so it
might be interesting to try out M4 if you can do that easily.
Randy
Aaron Evans wrote:
Anton <javantik <at> yahoo.com> writes:
Hi,
I am trying to use jetspeed2 M3. I noticed very slow
performance: with 25 concurrent users it is rendering
pages ~ 10 sec/page. I am using JMeter and hitting
home page of the sample site
(/jetspeed/portal/default-page.psml). Jetspeed runs
with default configuration (HSQL), additional
settings: set JAVA_OPTS=-Xmx512m -Xms128m in
startup.bat. (This is windows machine, P4 3.0, 1gb
RAM).
Also I noticed that it is using 100%CPU all the time.
Any ideas how to speed up jetspeed?
Anton,
one thing you should probably do is put your min heap size to the same as your
max (-Xms512m) because you could be running into a lot of garbage collection
issues.
Also, what version of Java are you running, 1.4 or 1.5?
As for the CPU, it should not be doing much when jetspeed is idle and not
being hit. So if it is, then you definitely have a problem. Memory might
be the main culprit though. Before setting your heap size to a fixed one,
you should note if the java process moves up to the 512 max quickly and
stays around there. If so, that could mean that you should crank it higher,
maybe around 768.
Anyway, I certainly hope that this is not typical! I don't know as I
haven't gotten that far yet. Has anyone else had similar results?
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
