[jira] [Created] (ARTEMIS-2589) Automatic Client failover is not working

["richard krasowski (Jira)" <ji...@apache.org>]

richard krasowski created ARTEMIS-2589:
------------------------------------------

             Summary: Automatic Client failover is not working
                 Key: ARTEMIS-2589
                 URL: https://issues.apache.org/jira/browse/ARTEMIS-2589
             Project: ActiveMQ Artemis
          Issue Type: Bug
          Components: ActiveMQ-Artemis-Native
    Affects Versions: 2.10.0, 2.6.4
            Reporter: richard krasowski
            Assignee: Clebert Suconic


+*Problem*+:

Automatic Client failover (reconnection due to server shutdown) is not working properly.

 

*+Description+*:

I created a live/backup group over server-discovery mechanisms. They both broadcast their connectors over udp and a multicast address. When the both servers (live and backup) are started, they are successfully recognized as live and backup.

 

I have 2 clients both implemented in Java Spring. One client acts as the producer and one client acts as consumer. They are configured over a ActiveMQJMSConnectionFactory and over DiscoveryGroup mechanisms (udp://{color:#008000}231.7.7.7:{color}{color:#0000ff}9876){color}. The producer has a jmsTemplate and the consumer is realized as a Message driven bean.

When I send 20 messages over the producer the live server acknowledges them and forwards them to the consumer. When I shutdown the live-server the consumer successfully failover to the backup server but the producer says: "*Backup is not active, trying original connection configuration now.*"

And after a minute it throws an Error "*org.apache.activemq.artemis.core.client : AMQ214016: Failed to create netty connection*" with an *java.net.UnknownHostException*:

It seems like the client receives the live/backup configuration and tries to reconnect to live server but after that it does not successfully failover to backup server.

 

I created a github repo with a docker-compose environment. There is a producer, a live/backup artemis broker pair and a consumer.

 https://github.com/krasowskir/messaging_poc_infrastructure

 

My goal is to send 200 messages and during the sending process I want to shutdown the server (activemq0) and I want to see a successful client failover of the producer to backup server (activemq1)

 



--
This message was sent by Atlassian Jira
(v8.3.4#803005)