[jboss-jira] [JBoss JIRA] (WFLY-12304) Upgrade Apache Artemis from 2.9.0 to 2.10.0

[Clebert Suconic (Jira)]

    [ https://issues.jboss.org/browse/WFLY-12304?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13781627#comment-13781627 ] 

Clebert Suconic edited comment on WFLY-12304 at 9/9/19 8:59 AM:
----------------------------------------------------------------

The REPLICATION_START_FINISH_SYNC  packet never arriving is a symptom of the issue I explained.

Look at this commit on artemis: https://github.com/apache/activemq-artemis/commit/85b93f0883bc06a2dfe2de9d560805a59d626d38


Since this commit, we are using NonFileClosingDefaultFileRegion to send the body of the file. it's sending it with zero copy.

On the same commit, look at the change on NettyConnectionm, method getFileObject.


NonFileClosingDefaultFileRegion is being ignored by the netty-xnio-transport handler, what will make the whole thing to freeze, not receive any data on the other side of the replication, and the replication-finish to never arrive.


I tried implementing a proper handler for the FileRegion, but it did not work.


The handler has code for FileRegion but for some reason that code is never happening... and I don't know the component well enough to fix it.



> Upgrade Apache Artemis from 2.9.0 to 2.10.0
> -------------------------------------------
>
>                 Key: WFLY-12304
>                 URL: https://issues.jboss.org/browse/WFLY-12304
>             Project: WildFly
>          Issue Type: Component Upgrade
>          Components: JMS
>            Reporter: Martin Stefanko
>            Assignee: Emmanuel Hugonnet
>            Priority: Blocker
>              Labels: blocker-WF18, downstream_dependency
>
> Upgrade Apache Artemis from 2.9.0.Final to 2.10.0.Final.
> Due to HA issue with Apache Artemis 2.10.0 this may not be the final version upgrade



--
This message was sent by Atlassian Jira
(v7.13.5#713005)