[
https://issues.apache.org/jira/browse/ARTEMIS-2185?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16701443#comment-16701443
]
juergen schmied commented on ARTEMIS-2185:
------------------------------------------
I'm running the spring boot 1.5.x out-of-the-box autoconfiguration for a
embedded broker. I will check what ClientConnectionPool is used. I might be not
properly pooled and close the connection after use. There are no errors
happening, so no stacktrace should be pregenerated since its a very expensive
operation.
> High allocation pressure due to pre generated ActiveMQExceptionMessage
> ----------------------------------------------------------------------
>
> Key: ARTEMIS-2185
> URL: https://issues.apache.org/jira/browse/ARTEMIS-2185
> Project: ActiveMQ Artemis
> Issue Type: Improvement
> Components: Broker
> Affects Versions: 2.6.3
> Reporter: juergen schmied
> Priority: Minor
> Labels: ChannelImpl, Memory, collection, garbage
> Attachments: tempsnip.png
>
>
> In org.apache.activemq.artemis.core.protocol.core.impl.ChannelImpl there is
> {code:java}
> public void returnBlocking(Throwable cause) {
> lock.lock();
> try {
> response = new
> ActiveMQExceptionMessage(ActiveMQClientMessageBundle.BUNDLE.unblockingACall(cause));
> sendCondition.signal();
> } finally {
> lock.unlock();
> }
> }
> {code}
> The response is created but not used normally.
> For every get/close connection there is the a ActiveMQExceptionMessage incl.
> stacktrace created and abandoned.
> This shows a high allocation pressure in jmc (8 MB garbage/200 calls).
> please don't precreate the unused exception (maybe use a producer producing
> exception when really needed)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
