[
https://issues.apache.org/jira/browse/JAMES-3841?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17624447#comment-17624447
]
ouvtam commented on JAMES-3841:
-------------------------------
I'm running into a deadlock because I implemented the example code provided by
https://activemq.apache.org/statisticsplugin.html in
ActiveMQHealthCheck#check().
Basically, it creates a temporary queue and "waits" for the response in the
replyTo queue (similar to ActiveMQCacheableMailQueue#getSize).
When calling MessageConsumer#receive(timeout) it still can end in a deadlock
when prefetchSize is zero (see
org.apache.activemq.ActiveMQMessageConsumer#receive). That would explain some
observed deadlocks in disabled tests in the test package
org.apache.james.queue.activemq, but thats a story for another time.
{code:java}
MessageDispatch md;
if (info.getPrefetchSize() == 0) {
md = dequeue(-1); // We let the broker let us know when we timeout.
} else {
md = dequeue(timeout);
} {code}
So, I set the prefetchSize to 1 in order to avoid the deadlock. Now, I receive
a InterruptedException caused by the timeout in
MessageConsumer#receive(timeout).
I enabled statistics in EmbeddedActiveMQ by calling
brokerService.setEnableStatistics(true);
Somebody has an idea or knows any caveats when running an embedded ActiveMQ?
> ActiveMQ: Collect statistics and publish metrics
> ------------------------------------------------
>
> Key: JAMES-3841
> URL: https://issues.apache.org/jira/browse/JAMES-3841
> Project: James Server
> Issue Type: Improvement
> Components: Queue
> Reporter: ouvtam
> Priority: Minor
>
> Currently, only some James metrics are available for ActiveMQ which are
> exposed in WebAdmin. Namely, enqueuedMail and dequeuedMail for each
> registered queue (i.e. spool and outgoing). Some ActiveMQ broker stats
> (memoryUsage, memoryLimit, storeUsage, storeLimit, maxEnqueueTime, etc.) are
> currently not available.
> According to [https://activemq.apache.org/statisticsplugin.html] ActiveMQ
> Metrics can be obtained through subscription or by polling regularly. At the
> moment the statistics plugin is only used to get the size of a queue.
> I propose to poll ActiveMQ statistics via ActiveMQHealthCheck and publish
> those as metrics. Currently, the health check creates a session and checks if
> no exception occurred.
> What do you think?
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: server-dev-unsubscr...@james.apache.org
For additional commands, e-mail: server-dev-h...@james.apache.org
