I am looking into using ws-notification pullpoints in several parts of my
application and have noticed something that is causing me concern.
(I am using the Soap-UI ws-notification script for most of my testing.)
The basic flow is:
Create a pullpoint,
Set up a ws-notification subscription to the pp (saving the subscription
id).
Send a few messages to the topic (via ws-notification notify)
Get the messages from the pullpoint.
Unsubscribe the PP from the topic (using the subscription id), then destroy
PP.
I put the following in a loop and run it continually and here is what I see
in the jconsole.
Every subscription creates several new queueus:
1) a queue with the just subscription id
2) a jca queue:
org.apache.servicemix.jca.{http://servicemix.org/wsnotification}<SUBSCRIPTION_ID>
3) a jms queue:
org.apache.servicemix.jms.{http://servicemix.org/wsnotification}<SUBSCRIPTION_ID>
In subscription/non-durable/Queue there is a another mbean with the
subscription id.
In the topic section there is another 3 mbeans that show up for everytime I
run through that loop.
AFAIK, I have cleaned up the subscription and the pullpoint, so there should
not be any left over resources.
This seems to be wasting resources on the activemq server (even if it's just
mbean's) for stuff that is no longer needed. These 7 mbeans (and possibly
other resources) seem to be UOR's that will never go away. If I start
multiplying this by many dynamic pullpoints, this could cause a problem.
Am I not cleaning up something that should be?
Has anyone done any testing to determine how big of an issue this is?
-Kevin
--
View this message in context:
http://www.nabble.com/pullpoint-subscriptions-using-activemq-resources-tp15502439s12049p15502439.html
Sent from the ServiceMix - User mailing list archive at Nabble.com.
