Hello Team, we are running two nodes with artemis 2.30 brokers connected with a static network connector. CORE and OPENWIRE are enabled on the same client acceptor. >From remote there is a java client connecting with a failover transport, doing a periodic send while broker A and/or broker B can be shut down.
ActiveMQ classic "failover:(tcp://A:port,tcp://B:port)" does a client failover under the hood. On connection loss it retries until success and blocks sending until it is reconnected again. This is good for server maintainance and failsafe operation. On the other hand artemis netty libs "(tcp://A:port,tcp://B:port)" seem to close producer and session permanently, if the currently connected broker goes away. Thinking of complex clients this implies client restart and is not what we want. Since both variants are nearly code - compatible, I would say the netty libraries are missing the nice reconnect functionality. Can it be true or did I forget something? Regards Herbert