[
https://issues.apache.org/jira/browse/ARTEMIS-2180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16791506#comment-16791506
]
Leo Provido commented on ARTEMIS-2180:
--------------------------------------
Hi [~jbertram].
I've tried reproducing this and this is what I did:
* I spun 3 c5.4xlarge in AWS, - using Artemis 2.6.3, I create a master broker
on one server, a slave broker on another, and a test client. I attached my
broker configs ( [^server2-primary-broker.xml] [^server3-backup-broker.xml] )
* I then ran our test on our test client, which sends messages with mean
message size of 1,000,000 bytes and concurrency of 40 for a total duration of
600 seconds.
* Half-way through the test, I invoked _artemis stop_ on the slave broker, and
then noticed WARN messages logged by the master broker. Attached the
artemis.log produced by the master broker. [^artemis.log], which was concluded
by "[org.apache.activemq.artemis.core.server] AMQ222010: Critical IO Error,
shutting down the server".
I ran this test 3 times using a fresh stack of servers with interruptions
(invoke _artemis stop_ on the slave broker) and got the same outcome.
I also ran this test 3 times using a fresh stack of servers without
interruptions and didn't encounter the issue.
I've yet to try this on 2.6.4 and the latest build.
> Host and broker runs out of memory when stopping a backup in a cluster
> ----------------------------------------------------------------------
>
> Key: ARTEMIS-2180
> URL: https://issues.apache.org/jira/browse/ARTEMIS-2180
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Components: AMQP
> Affects Versions: 2.6.3
> Reporter: Simon Chalmers
> Priority: Critical
> Attachments: artemis.log, hs_err_pid15534.log,
> server2-primary-broker.xml, server3-backup-broker.xml
>
>
> When running a live-backup cluster pair and stopping the backup, the live
> broker starts leaking memory, runs out of memory and core dumps.
> This occurs during a performance test when the broker is under load; the
> slave has been running successfully but is then terminated whilst the broker
> is still under load.
> Core dump attached.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
