[jira] [Commented] (AMQ-3738) Cursors not dealing with topic memoryLimit overflow

["Torsten Mielke (Commented) (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AMQ-3738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13218032#comment-13218032 ] 

Torsten Mielke commented on AMQ-3738:
-------------------------------------

IIRC then producer flow control will kick in. Paging out to disk only happens when 100% of the memoryLimit are used. At that same time producer flow control will kick in as well because of the 100% limit being reached. Paging to disk will only occur if producer flow control is disabled. 
But I presume the limit at which paging to disk happens may be configurable.
                
> Cursors not dealing with topic memoryLimit overflow
> ---------------------------------------------------
>
>                 Key: AMQ-3738
>                 URL: https://issues.apache.org/jira/browse/AMQ-3738
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker
>    Affects Versions: 5.5.1
>         Environment: OSX, Apple Java Runtime 1.6.0_26
>            Reporter: Jakub Korab
>         Attachments: activemq-cursors.xml, pom.xml
>
>
> I have defined a topic with a memoryLimit of 1mb using a fileDurableSubscriberCursor. I would expect that sending messages to a durable subscriber through that topic would result in more than 1mb worth of messages being able to be sent. Instead producer flow control kicks in at the 1mb limit. 
> To test this I am using the Maven activemq-perf plugin with a single consumer and producer:
> $ mvn activemq-perf:producer -Durl=tcp://localhost:61616 -Dproducer.destName=topic://orders
> $ mvn activemq-perf:consumer -Durl=tcp://localhost:61616 -Dconsumer.durable=true -Dconsumer.destName=topic://orders -Dconsumer.recvDelay=100
> This does not match the behaviour described in http://activemq.apache.org/my-producer-blocks.html or http://activemq.apache.org/message-cursors.html
> Going by the docs, I would expect messages to get paged out to disk until the systemUsage/tempUsage limit fills up.
> This same behaviour happens using the default cursor configuration.
> Attached are the activemq-perf pom and activemq.xml.

--
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