[
https://issues.apache.org/jira/browse/ARTEMIS-2355?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
clebert suconic updated ARTEMIS-2355:
-------------------------------------
Fix Version/s: 2.9.0
> Message does not redistribute in cluster when using AMQP and Multicast
> ----------------------------------------------------------------------
>
> Key: ARTEMIS-2355
> URL: https://issues.apache.org/jira/browse/ARTEMIS-2355
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Affects Versions: 2.7.0
> Reporter: Luis Miguel De Bello
> Priority: Major
> Fix For: 2.9.0
>
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> We are migrating from Artemis 2.6.3 to Artemis 2.7.0 and we hit and issue
> regarding clustering and message redistribution.
> Scenario:
> Broker 0:
> Create a Bridge to Broker 1
> Broker 1:
> Create a Cluster with Broker 2
> Broker 2:
> Create a Cluster with Broker 1
>
> Consumer in Broker 2 in Multicast Address.
> Producer Send Anycast Message to Broker 0, the message is Diver in Broker 0
> to Address for the Bridge. Message Arrive to Address in Broker 1, it is
> Divert to Multicast Address + Setting Routing Type Multicast.
>
> It works in 2.6.3 and it does not work in 2.7.0. I am almost sure the issue
> is while doing the redistribution and checking the isMatchRoutingType because
> in 2.6.3 is ok but in 2.7.0 they do not match. If someone can point me out
> some debugging class to see the message sent from Broker 1 to Broker for
> redistribution I can try to debug it.
> I will create a branch with the test
>
> Test Class: AmqpBridgeClusterRedistributionTest
> Repo: [https://github.com/ldebello/activemq-artemis/tree/ARTEMIS-2355]
>
> After some debugging + going back in commits the issue start happening after
> this commit
>
> [https://github.com/apache/activemq-artemis/commit/6a64f4e7a270f83c735200d6ef29561ef4927535#diff-4a5cefc3ba360bce9985ad42fc0c2231R698]
>
> I am assuming the redistribution is not "encoding" or updating the message to
> multicast because the buffer is already valid and in previous version
> reencode all the time so the message was ok.
>
> Removing the line fix this issue, anyway I think that is for performance so I
> will try to track were some validation should be added to reencode the
> message again, if someone has some idea where start looking that would be
> great.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
