You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@activemq.apache.org by "Gary Tully (JIRA)" <ji...@apache.org> on 2017/01/12 12:26:53 UTC

[jira] [Commented] (AMQ-6424) Duplicate dead letter queues

    [ https://issues.apache.org/jira/browse/AMQ-6424?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15820885#comment-15820885 ] 

Gary Tully commented on AMQ-6424:
---------------------------------

as a possible workaround you could provide a "discarding" dead letter strategy for queues of the form DLQ, which would ensure that duplicates from DLQs don't result in further nested DLQ creation.
Getting to the root cause and fixing would need some sort of test to speed up the investigation

> Duplicate dead letter queues
> ----------------------------
>
>                 Key: AMQ-6424
>                 URL: https://issues.apache.org/jira/browse/AMQ-6424
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker, KahaDB
>    Affects Versions: 5.13.0, 5.13.3, 5.14.0
>         Environment: JDK 1.8.0_93 , Windows Server 2012 R2
>            Reporter: Berge Stillingen
>              Labels: database, journal, mssql, paging
>
> In version 5.14 (and 5.13.3) it appears the broker is able to 'occasionally' create extra DLQ-queues holding duplicate messages. 
> Browsing the admin console, you will sometimes see something like this: 
> - DLQ.myEventQueue 
> - DLQ.DLQ.myEventQueue 
> That both hold the samme message ID, eg. ID:ClientABC-61753-1471513949441-1:4:1:3:1	
> The one from DLQ.myEventQueue is a standard entry caused by a client transaction rollback. 
> The (unwanted) one from DLQ.DLQ.myEventQueue has a 
> - dlqDeliveryFailureCause: java.lang.Throwable: duplicate paged in from store for queue://DLQ.myEventQueue
> At some point you can even get a third queue named 
> - DLQ.DLQ.DLQ.myEventQueue holding another message with the samme MessageID and the same dlqDeliveryFailureCause 
> Reproduce: Haven't found a way to provoke the error. It happens occasionally in a semi-active test environment running with Mule 3.6.2. 
> Best guess is that there is some issue with the use of Journaling and synchronization between Kahadb and the MSSQL-database. 
> Configuration can be found at 
> http://activemq.2283324.n4.nabble.com/file/n4715868/activemq-anonymous.xml



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)