Hi,
I am using servicemix 3.3.1 version. And wanted to load test one of the CXF
web service.
I was using soapUI 3.5 and tested up to 15 threads (15 Threads for 1 minute
with a test delay of 1 second). It works fine. As soon as I increase the no
of Threads to 16 after sometime, I was getting time outs. And then found out
the whole server was frozen with no error message in the log. This is the
last line of the log I saw
=======================================
16:17:31,904 DEBUG [32980...@qtp0-33] (SedaFlow) - Called Flow send
16:17:31,904 DEBUG [32980...@qtp0-33] (DeliveryChannelImpl) - Waiting for
exchange ID:127.0.0.1-128980e3012-8:52 (6c97c1) to be answered in
DeliveryChannel{servicemix-cxf-bc} from sendSync
=======================================
This behavior could be repeated exactly at 16 threads with few CXF web
services I have . Then I started searching in the forum and found this
http://old.nabble.com/cxf-bc-load-testing-tp23195483p23195483.html
http://old.nabble.com/cxf-bc-load-testing-tp23195483p23195483.html
As suggested, I increased the following properties (in
conf/servicemix.properties).
servicemix.corePoolSize = 32
servicemix.maximumPoolSize = 64
servicemix.queueSize = 512
Now it is failing exactly at the 32 Threads.
In that form entry it was already said that
https://issues.apache.org/activemq/browse/SM-1717 was a defect related to
that and it was already fixed in servicemix 3.3.1 (the version I am using).
I think it is OK to get the time out but, if the server is going to be
frozen, that is not acceptable to me. At the same time, increasing the
servicemix.corePoolSize and other parameters is just a short term solution
for me in our production environment.
Has anyone come across something similar ? Any suggestions ?
--
View this message in context:
http://old.nabble.com/CXF-web-service-load-testing-freeze-the-server-tp28564246p28564246.html
Sent from the ServiceMix - User mailing list archive at Nabble.com.
