Short update, I have tested the application on a different setup and when
running on Windows XP, single core CPU, jdk 1.6_u13, Grizzly patch 1.0.29
with moderate load, the CPU issue is not observed.
Thus it seems like the issue may be correlated to specific hardware/jdk etc.
and I think we need to pinpoint this better on our side before moving on. I
will try to get some time to make an isolated version of the application
(without the business end of it) that can trigger the behavior. The 1.0.29
patch certainly fixed the worst initial CPU leak though.
I am not sure what to do about the deadlock, or if to do anything at all...
Even if we get a fix backported, it will not be guaranteed anyway since we
have no way of reliably reproduce the deadlock. I think we might have to run
our application long term and see how frequent (if at all) we can reproduce
it before making any decisions.
--
View this message in context: http://www.nabble.com/Comet---growing-CPU-usage-tp23922811p24001089.html
Sent from the Grizzly - Users mailing list archive at Nabble.com.