[
https://issues.apache.org/jira/browse/AMQ-3669?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13279665#comment-13279665
]
Claudio Corsi commented on AMQ-3669:
------------------------------------
I was able to determine what the problem was with this issue.
It is that when we initially forward the message to the mirrored topic. The
message memory usage instance is set to the topic.
When the message is sent to the queue, the memory usage instance is not set to
the queue memory usage but it still uses the topic memory usage.
The message setRegionDestination method is used to set the message memory usage
but this is only performed if and only if the current message memory usage
instance is null. This is not the case when forwarding the message to the
queue and that is why the memory usage of the topic gets affected causing the
producer flow control to throttle the topic/queue when that is not the case.
> Pending producer with qMirror, messages are not spooled to disk
> ---------------------------------------------------------------
>
> Key: AMQ-3669
> URL: https://issues.apache.org/jira/browse/AMQ-3669
> Project: ActiveMQ
> Issue Type: Bug
> Components: Broker, Message Store
> Affects Versions: 5.5.1
> Environment: local test on windows, happens also on linux remotely
> Reporter: Darius Schier
>
> We fill a queue that is backed by a qmirror with data (persistent) that
> exceeds the configured memory limit.
> The producer (producerFlowControl="true" at queue and qmirror) will be
> throttled when the the limit is reached, no messages are spooled to disk.
> As an overall result, we are not able to write a lot of messages into amq.
> *Configuration:*
> {code:xml}
> <destinationInterceptors>
> <mirroredQueue copyMessage = "true" postfix=".qmirror"
> prefix=""/>
> </destinationInterceptors>
> <destinationPolicy>
> <policyMap>
> <policyEntries>
> <policyEntry topic=">" producerFlowControl="true"
> memoryLimit="2mb" />
> <policyEntry queue="created.static.for.persistent"
> producerFlowControl="true" memoryLimit="1mb" />
> </policyEntries>
> </policyMap>
> </destinationPolicy>
> <destinations>
> <queue physicalName="created.static.for.persistent" />
> </destinations>
> <persistenceAdapter>
> <kahaDB directory="${activemq.base}/data/kahadb"/>
> </persistenceAdapter>
>
> <systemUsage>
> <systemUsage>
> <memoryUsage><memoryUsage limit="10 mb" /></memoryUsage>
> <storeUsage><storeUsage limit="100 mb"/></storeUsage>
> <tempUsage><tempUsage limit="100 mb"/></tempUsage>
> </systemUsage>
> </systemUsage>
>
> <transportConnectors>
> <transportConnector name="openwire" uri="tcp://0.0.0.0:61616"/>
> </transportConnectors>
> {code}
> *Java test code:*
> {code}
> ActiveMQConnectionFactory factory = new
> ActiveMQConnectionFactory("tcp://localhost:61616");
> Connection connection = factory.createConnection("user", "pwd");
> connection.start();
> Session session = connection.createSession(false,
> Session.CLIENT_ACKNOWLEDGE);
> Destination destination =
> session.createQueue("created.static.for.persistent");
> MessageProducer producer = session.createProducer(destination);
> producer.setDeliveryMode(DeliveryMode.PERSISTENT);
> char[] m = new char[1024];
> Arrays.fill(m, 'x');
> // create some messages that have 1k each
> for (int i = 0; i < 12000; i++) {
> ActiveMQTextMessage message = new ActiveMQTextMessage();
> message.setText(new String(m));
> producer.send(message);
> }
> connection.stop();
> connection.close();
> }
> {code}
> *Expectation:*
> Messages should be written to disk when the memory limit exceeds, all
> messages should be available within the queue/topic.
> *Result:*
> INFO | Usage Manager memory limit (2097152) reached for
> topic://created.static.for.persistent.qmirror. Producers will be throttled to
> the rate at which messages are removed from this destination to prevent
> flooding it. See http://activemq.apache.org/producer-flow-control.html for
> more info
> Store percent used : 10
> Memory percent used : 20
> Temp percent used : 0
> Interesting: The smaller flow control for the queue (1mb) does not seem to
> catch but the qmirror does (2mb).
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
