[jboss-jira] [JBoss JIRA] (WFWIP-65) [Artemis 2.x upgrade] Artemis RA connections disconnected after failover

[Yong Hao Gao (JIRA)]

    [ https://issues.jboss.org/browse/WFWIP-65?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13619513#comment-13619513 ] 

Yong Hao Gao commented on WFWIP-65:
-----------------------------------

[~clebert.suconic] well, I couldn't reproduce it so I rely on the attached logs.
So far what I see is that there may be some problem with the pooled connection factory. (jmsXA).
When the MDB retries the reconnection, the log shows the max retry attempts is 1 and there is no backup connector during the retry.
For example:

06:40:55,834 DEBUG [org.apache.activemq.artemis.core.client.impl.ClientSessionFactoryImpl] (Reset MessageHandler after Failure Thread) Connector towards NettyConnector [host=127.0.0.1, port=8080, httpEnabled=false, httpUpgradeEnabled=true, useServlet=false, servletPath=/messaging/ActiveMQServlet, sslEnabled=false, useNio=true, activemqServerName=default, httpUpgradeEndpoint=acceptor] failed
06:40:55,834 DEBUG [org.apache.activemq.artemis.core.client.impl.ClientSessionFactoryImpl] (Reset MessageHandler after Failure Thread) Backup is not active, trying original connection configuration now.

Here is the pooled cf :

<pooled-connection-factory name="activemq-ra" entries="java:/JmsXA java:jboss/DefaultJMSConnectionFactory" connectors="connector-remote-live" ha="true" min-large-message-size="102400" block-on-acknowledge="true" retry-interval="1000" retry-interval-multiplier="1.0" reconnect-attempts="-1" transaction="xa" min-pool-size="-1" max-pool-size="-1"/>

You can see the ha is true. 

So far I'm not quite clear how this factory is initialized (may be it's in wildfly code base).

Howard


>  [Artemis 2.x upgrade] Artemis RA connections disconnected after failover
> -------------------------------------------------------------------------
>
>                 Key: WFWIP-65
>                 URL: https://issues.jboss.org/browse/WFWIP-65
>             Project: WildFly WIP
>          Issue Type: Bug
>          Components: Artemis
>            Reporter: Miroslav Novak
>            Assignee: Clebert Suconic
>            Priority: Blocker
>              Labels: feature-branch-blocker
>         Attachments: DedicatedFailoverTestCaseWithMdb.testShutdownWithFailback.zip
>
>
> There is regression againt Artemis 1.5.5.0012 in scenario with failover of Artemis RA from live to backup (shared store, dedicated topology). After backup activates after failover, connections of RA does not failover to backup (or are disconnected from backup just after failover)
> Test scenario:
> * There are three servers. Live server (Node 1) and backupserver (Node 2) are in dedicated HA topology. 
> * InQueue and OutQueue are deployed on live and backup. Send messages to InQueue on live server.
> * When all messages are sent, deploy message driven bean on Node 3. 
> * MDB sends messages form InQueue to OutQueue. 
> * Shutdown live server. Wait for backup server to alive, then start live server again and stop backup.
> * Receive messages from OutQueue from live server.
> Results:
> After live server is stopped and backup activates, no messages are processed by MDB. Looking at trace logs it seems that MDB does reconnect to backup but it's then disconnected. Then I don't see connector to backup server between tried connectors during reconnection. 
> Attaching reproduce and logs.



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)