activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "C Velo (JIRA)" <j...@apache.org>
Subject [jira] Closed: (AMQ-2906) Seeing messages getting "stuck" on brokers w/o application consumers
Date Fri, 10 Sep 2010 20:12:41 GMT

     [ https://issues.apache.org/activemq/browse/AMQ-2906?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

C Velo closed AMQ-2906.
-----------------------

    Resolution: Fixed

> Seeing messages getting "stuck" on brokers w/o application consumers
> --------------------------------------------------------------------
>
>                 Key: AMQ-2906
>                 URL: https://issues.apache.org/activemq/browse/AMQ-2906
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker
>    Affects Versions: 5.4.0
>         Environment: CentOS 5.2 (2.6.18-128.1.10.el5.xs5.5.0.51xen), ActiveMQ server
5.4.0, ActiveMQ client 5.2.0 (w/ connection pooling)
>            Reporter: C Velo
>         Attachments: activemq_broker1.xml
>
>
> Seeing messages moved around between brokers and occasionally getting "stuck" on brokers
w/o application 
> consumers on it (even though the broker might have an ActiveMQ consumer attached to it).
 I can still see messages
> getting Enqueued and Dequeued on the broker containing the "stuck" messages, but the
"stuck" messages never
> move (have some messages that have been on the queue for 1hour+).
> Current setup:
> • 3 Producers
> • 3 brokers
> • 3 Application consumers (which are using ActiveMQ connection pooling)
> • 2 application consumers --> broker1
> • 1 application consumer --> broker2
> • broker3 <-- 0 application consumers
> ----
> This is what the networkConnector string looks like in the activemq.xml file on the server
side:
> (note: line breaks are for readability - in practice it is a single line w/o the "\"
at the end of each line)
> {code}
>         <networkConnector name="ActiveMQ Cluster" uri="static:(tcp://broker1:61616,tcp://broker2:61616,tcp://broker3:61616)"
\
>           conduitSubscriptions="true" dynamicOnly="true" decreaseNetworkConsumerPriority="false"
\
>           suppressDuplicateQueueSubscriptions="true" duplex="false" networkTTL="5"/>
> {code}
> ----
> This is what the broker.url string looks like in on the client side:
> (note: line breaks are for readability - in practice it is a single line w/o the "\"
at the end of each line)
> {code}
> amq.publisher.broker.url=failover\:tcp\://broker1\:61616?wireFormat.maxInactivityDuration=0&daemon=true,\
>                                                               tcp\://broker2\:61616?wireFormat.maxInactivityDuration=0&daemon=true,\
>                                                               tcp\://broker3\:61616?wireFormat.maxInactivityDuration=0&daemon=true
> {code}
> ----
> +To reproduce:+
> * start up multiple Producers, sending messages to the 3 brokers
> * start up only 2 consumers connected to 2 of the 3 brokers.
> ----
> It would be nice if there was a maxQueueTimeout parameter that could
> be set by the user, so that if a message has been on a queue for over 
> a certain amount of time, it would be automatically routed to another broker.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message