1 of 1 people found this helpful
The whole deploy-hasigleton directory concept is not there anymore.
It should be suceeded by the domain controller. Not sure if the functionality is there yet though.
I tot domain controller is just for management and has got nothing to do with clustering,
One of the disadvantages to the deploy-hasingleton concept in previous JBoss AS releases was that membership change of the cluster resulted in a period of time during which the deployment was not available. Only after the previously hosting member leaves the cluster would the deployment get deployed on the newly selected master node. The approach I've recommended people to take in AS7 is to deploy your application on every node in the cluster, and create an msc Service<Locator>, where Locator provides the means of connecting to your deployment. This service can then be wrapped by org.jboss.as.clustering.singleton.SingletonService (analogous to HASingleton in previous AS releases).
Consider the case of an EJB deployment that should you only want accessed on one node at a time. You would create a Service<Remote>, where Remote is the remote interface of your bean. Install this service into the MSC as: new SingletonService(myService, myServiceName). Any time your application needs to access this EJB, it would query this service for the bean's remote stub. SingletonService will ensure that the remote stub refers to the master node of the cluster, regardless of the node from which you requested the EJB. This has the same effect of deploying the EJB via deploy-hasingleton, but doesn't have the problem of deployment downtime.
For an example of how to use SingletonService, refer to the testsuite:
Hope the guides for HA and MSC come up soon, cant' wait until then... .