Right now all cluster members must be specified in a comma separated list in the JDBC url. While we plan to continue to support that, we should also allow the single cluster endpoint to be provided. When a cluster endpoint is provided we should then discover the individual cluster members (ideally even dynamically with a running application).