1. Ripple goes through the process of taking any configuration settings in the server group (cluster) and distributing them amongs your declared clones. this proecess also performs the a sequential stop and restart of the cloned app server in order for the configurations setting to take place through out your cluster without causing any down time.
2. it is my understanding that the only time you need to regen the plugin confing is when either the console in the admin console states that "the action preformed ... invalidates the server pluggin, you will need to regen the plugin" or when there is a change within your EAR (adding a new servlet, adding another clone, changind the transport setting).
3. if you are in a cloned environment and you regen the plugin, you will have to manually copy it out to another server only if the http server is located on a remote machine from where your nodes are setup. easier ways might include writting a script to do this or getting technologies such as AFS working within your cluster.
hope some of this helps,