Hello, We are using Apache Camel with Active MQ extensively for our integration needs. For one of our customers, a high availability solution is a must. So we are in the process of solutioning this.
Deployment model: 1) Active MQ(5.9.0) is deployed in a VM. 2) Apache Camel is deployed on a Apache Tomcat instance. We are using Spring DSL for all the routes. We will have multiple VM having one instance of tomcat. For our testing purpose we have only 2 of this instance. We thought of having an Active-Active model so that it addresses both the scalability and availability aspects. 1) An Active MQ instance in a master-slave mode(on a distributed file system mode). 2) 2 Camel instances acting upon this MQ instance. There are routes which both produce and consume. We are thinking of using the linux LVS and ldirectord to handle the scenario of keeping track of one of the instance going down. When we tested with multiple Camel instance on a non-failover ActiveMQ instance things were fine. We could see the producer putting into the queue and anyone of the camel instance picking up for acting on the queue. But when we tested the multiple Camel instance on a failover Active MQ we are seeing that the message are not picked up consistently. We tried with Exclusive, Temporary and Shared active MQ configuration. Please let me know your suggestions and also if anyone has done a similar solution, please share your thoughts. Thanks.
