if the controller can not connect to a configured netconf server, RESTCONF is dead until connectivity to the server is established. This is 100% repeatable: try issuing a RESTCONF request when connectivity to a node is down (even to a different node) - the request does not come back, until it times out, or until connectivity to the netconf server is re-established. Note also that RESTCONF seems to be dead for other nodes too, not just the node to which we have no connectivity

What we need to do when there is no connectivity to a configured netconf erver is to drop the server from operational space, and only have it show up in the config space. We may want to show the status of the node as unreachable if we can't connect to it. (It is configured, but we can not reach it). We should also maintain whatever configuration we have for the node.

We may want to have a configurable parameter for a retry timeout and retry count, which would specify how frequently and how many times we retry the attempt to connect to the node. For Hydrogen we may keep it simple - retry when a request to show the node or show inventory comes from RESTCONF (nobody is interested in the node until they actually ask for its status, which will cover most apps).