from my side, it's http server configuration limitation.After peformance tune, it works well now.
doing much testing, I am able to recreate the "frozen server"
situation on demand.
Strangely enough, it seems like the problem is caused by my use of
org.apache.commons.lang.builder.ToStringBuilder in a (non-infinite)
loop. When I remove the usage, the problem disappears. I'll have to do
more investigation before I can reach a conclusion though
On Fri, Feb 27, 2009 at 12:42 AM, Stella Lok <email@example.com> wrote:
> Hi Kevan,
> The machine has one Xeon processor which is a dual core. I am running
> WAS CE on IBM JSE 5.
> I do have the thread dump showing the blocked MQ threads, which I have
> just forwarded to you, if you would like to take a look (if anyone
> else is interested in taking a look, please let me know)
> Btw, the activemq used is not the one embedded in WASCE, but an external one.
> Thank you very much
> On Thu, Feb 26, 2009 at 11:13 PM, Kevan Miller <firstname.lastname@example.org> wrote:
>> On Feb 26, 2009, at 5:58 AM, Stella Lok wrote:
>>> I am using WAS CE 220.127.116.11 with ActiveMQ 5.2.0. While doing load
>>> testing, I encountered situations when the web server (which is the
>>> Tomcat embedded in WAS CE) permanently (until restart) stopped
>>> responding to any HTTP requests (the catalina log stops logging new
>>> requests), and none of the apps nor the admin console can be accessed
>>> anymore. When that happens, the WAS CE java process' CPU utilization
>>> would invariably be permanently stuck at around the 50% level, even if
>>> I completely stop accessing the server.
>>> I did a thread dump whenever this occurred, and most of the time I do
>>> not see any blocked threads, only runnable, parked and
>>> waiting-on-condition threads. Once, I was able to see that there were
>>> 4 blocked ActiveMQ threads, though I'm not sure if this is related.
>>> I would be most grateful if someone could help guide me to possible
>>> causes of the problem.
>> Hmm. Do you still have the thread dumps? Esp. the one showing the blocked
>> AMQ threads? Is this a 2 CPU machine (e.g. a single dual core processor?)
>> Sounds like you have a single thread spinning. A thread dump should help
>> identify the culprit... Are you running on Sun or IBM JSE?