Hi Bruce, I would say what you are running is not a master / slave pair, but just two parallel broker.. If the falover transport of the Consumer and Broker A point to different Brokers, you do not get Messages.
What you need is a locking mechanism. Best Regards Herbert Von: "Bruce Cooper" <[email protected]> An: [email protected] Datum: 23.02.2024 03:24 Betreff: [Ext] Issue with network of brokers and active/passive pair Hi, I'm running into a strange issue configuring activemq 5.17.6 to forward messages from a satellite broker to a pair of brokers running in an active/passive (failover) pair. I'm hoping somebody here can help me I have three brokers. The first is running stand-alone, and is configured with a one way network bridge to the other two, who are operating in an active/passive (also known as master/slave) mode. /----> Broker B. <---\ Producer -> Broker A ---+ +---- Consumer \----> Broker C <----/ This works fine, up until I try to test what happens when the redundant pair is rebooted/goes down. My test procedure is this: Start all brokers, and ensure that everything is connected Run the producer script multiple times to send persistent messages to a queue, seeing that messages are delivered to the consumer, and printed out. Shut down (via activemq stop cmd) the broker that the consumer is connected to (B) Observe that the consumer reconnects to the other broker (C), as does the satellite broker (A) Run the producer script again. Observe that one set of messages is delivered Run the producer script again - This time no messages are delivered. Messages are accepted by Broker A, and it seems that it has forwarded them to broker C (The queue shows the same number of enqueued and dequeued messages in Broker A's console, as does the Network Connector), but no messages (apart from the first set) appear in Broker C's queue. Get consumer to fail back to the original server (B) by restarting B, then stopping C Any subsequent attempts to send messages fail. Restart Broker A Messages now flow, but any that were transmitted at step 7 have been lost If I change the topology to remove Broker C and restart Broker B, messages are appropriately queued, forwarded and delivered to the consumer once it reconnects after the restart. It behaves as expected. The configuration for the brokers is the default example configuration, except that Broker B and C use the same persistenceAdapter kahaDB directory and broker A has a network Connector that looks like <networkConnectors> <networkConnector name="laptop" uri="static:failover:(tcp:// 100.127.41.128:61616,tcp://100.127.41.128:61617)" duplex="false" staticBridge="true" userName="bruce" password="sekrit" networkTTL="10"> <staticallyIncludedDestinations> <queue physicalName="TestQueue"/> </staticallyIncludedDestinations> </networkConnector> </networkConnectors> I have attached the configuration files for reference, along with the test program I am using to send and receive. What am I doing wrong? -- https://mechination.com.au/ Ph: 0448 341 729[Anhang "nodeA.xml" gelöscht von Herbert Helmstreit/RG/SYSTEMA/DE] [Anhang "nodeC.xml" gelöscht von Herbert Helmstreit/RG/SYSTEMA/DE] [Anhang "nodeB.xml" gelöscht von Herbert Helmstreit/RG/SYSTEMA/DE]
