We’ve just released the 1.2.1 Version of the MCA Java client.
It brings the following enhancements and fixes based on customer feedback:
The MultiClusterClient now lets it pass a FailureDetectorFactory with the proper generic argument, removing the need for downcasting on creation. So for exampleFailureDetectorFactory<NodeHealthFailureDetector> can be passed in directly instead of downcasting to FailureDetectorFactory.
Both the NodeHealthFailureDetector and TrafficMonitoringFailureDetector can be subclassed to override the methods which trigger the Coordinator. This can be helpful to "intercept" failure signals if a failure detector should be plugged in to work in a "dry run" mode. If the calls are intercepted and not forwarded to the coordinator but rather logged it will not have an effect on the system but can be used to test certain settings and see how they behave in testing or production environments.
Fixed a bug where subsequent service activations (after one has been activated on an entry) would not be picked up by the distribution logic. This mostly happens during interactive examples, since programmatically all services are usually enabled in one shot. This is also the work around for previous versions if needed.
Add support for ByteArrayDocument. It has been added to the Java SDK but the corresponding transcoder has not been included by default up until now.
The documentation and further information can be found here: https://subscription.couchbase.com/multi-cluster-java