On Thu, Oct 22, 2009 at 10:04 PM, Ron Wells <[email protected]> wrote:

> ?? maybe not understanding... not failing just eating up machine .. IFL
> just about pinged out on usage . with one app.WAS with the Java

But if your WAS application has work to do and the CPU is there, would
you expect something else than that it uses what it can get?  Starting
a Java application can use a *lot* of CPU because the optimizer and
JIT (re)compile classes as CPU time permits. This can sometimes make
it continue using a lot of CPU cycles even after the application
server claimed it is ready and waiting for work.

Without instrumentation it does get hard because you don't have a
record of what happened. But I doubt this was your favorite way to
approach the problem without tools... ;-)

First you need to check that it really is the application using the
CPU, and not something else. So does it really get the resources from
z/VM or is something holding it back there? You also make sure you
don't swap excessively. When it really is the application using it,
you could use "ps" to see the total accumulated CPU usage per process
/ thread, take a javacore dump and see what were the big chunks. When
you find that the garbage collection is the big chunk, then run the
logs through a GC analyzer to see if someone seriously underestimated
the application memory requirements.

The other thing to look out for (especially since you say they are
just getting started) is that anything that you deploy on WAS will
take resources, even when you don't use it. Sometimes developers start
with deploying all kind of sample applications to learn.

Rob
--
Rob van der Heij
Velocity Software
http://www.velocitysoftware.com/

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to