This will require us to support the idea of a client directed cluster type and make sure our tracking of the cluster's state is consistent with the specification's test cases.
The cluster type change will mean looking at the "replica set" name in the URI and making it match the name reported by the server. We should also add a "direct" type that disables monitoring entirely. This will be used for clients that want to connect to the server, only that server, and minimize additional over head.
The "Single" is our current "Standalone" but with a monitor thread. (I am not sure monitoring a standalone instance is worth it.)
This will probably spawn more tickets as the specification is published and we understand it better.
This will require us to support the idea of a client directed cluster type and make sure our tracking of the cluster's state is consistent with the specification's test cases.
The cluster type change will mean looking at the "replica set" name in the URI and making it match the name reported by the server. We should also add a "direct" type that disables monitoring entirely. This will be used for clients that want to connect to the server, only that server, and minimize additional over head.
The "Single" is our current "Standalone" but with a monitor thread. (I am not sure monitoring a standalone instance is worth it.)
This will probably spawn more tickets as the specification is published and we understand it better.
See: http://emptysqua.re/server-discovery-and-monitoring.html Official Blog Posting: http://blog.mongodb.org/post/99058181018/announcing-the-server-discovery-and-monitoring-spec