Broker 10.5 | webMethods Broker Documentation | webMethods Broker Messaging Programmer's Guide | JMS Policy Based Client-Side Clustering | Failover Mechanism in Clusters
 
Failover Mechanism in Clusters
In a cluster, if any Broker goes offline or comes online, the webMethods API for JMS throws a connection exception. Configure your JMS client application to catch the connection exception and reconnect to the cluster.
When the webMethods API for JMS receives the request to reconnect from the JMS client application, the webMethods API for JMS switches the publishing task to an available Broker in the cluster, based on the policy.
For example, consider a Broker cluster consisting of Broker A, Broker B, and Broker C. At any point of time, if Broker A goes offline or comes online, failover is achieved as follows:
1. Broker A in the cluster goes offline.
2. webMethods API for JMS throws a connection exception.
3. The JMS client application catches the connection exception and reconnects.
4. webMethods API for JMS switches the publishing task to Broker B and Broker C.
5. The JMS client application continues publishing using Broker B and Broker C.
6. If Broker A comes online, the webMethods API for JMS again throws a connection exception.
7. The JMS client application catches the connection exception and reconnects.
8. webMethods API for JMS includes Broker A for publishing messages.
9. The JMS client application continues publishing using Broker A, Broker B, and Broker C.