Hi Alexey,
After treating HP-UX like Linux in the test for isLinux in Controller.java
(in the 1.9.19 snapshot code base), our QA engineer ran a test that made a
total of 85 requests from 10 different threads (monitored by the profiler).
This resulted in 12146 calls to getSpinRate(), 69 calls to
resetSpinCounter(), and 20 calls to workaroundSelectorSpin() over the course
of 55 seconds.
Based on the result, in your opinion, would this classify as the Linux
selector spin issue? Or would this be considered normal? Unfortunately for
us, we don't have another system to run the test on. Our application only
runs on the HP-UX system.
If there is any more detail you would like, please let me know. This test
was run on a system that does not exhibit the CPU spikes that prompted the
original posting. We are setting up another test on a system that is more
like the client experiencing the CPU spikes to see if the results are any
different.
Thanks for your help. I really appreciate it.
--Dennis
Oleksiy Stashok wrote:
>
> Hi Dennis,
>
>> the System.getProperty("os.name") is "HP-UX". Our contact at HP
>> states that
>> the selector spin test that fails (see link below) is a sun bug, and
>> will be
>> fixed in 1.7 which makes me believe all the more the issues are
>> related. I
>> wonder if HP-UX and Linux share the same sun code base?
>>
>> Just for fun, I have modified the grizzly Controller class to treat
>> "HP-UX"
>> the same as "linux" and we will be running some tests. I'll let you
>> know
>> the results.
> Great, it should be my next suggestion :)))
> Please let me know the result.
>
> WBR,
> Alexey.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_grizzly.dev.java.net
> For additional commands, e-mail: users-help_at_grizzly.dev.java.net
>
>
>
--
View this message in context: http://old.nabble.com/Grizzly-Web-Server-on-HP-UX-PA-RISC-Performance-Issue-tp26733514p26845878.html
Sent from the Grizzly - Users mailing list archive at Nabble.com.