Allowing raft client activity even if there's no defined ctl-svc entry?

[00pauln00]

I'm wondering if we should create a ctl-svc entry for a client whose info had not been declared at server startup time? This would reinstate some of the functionality which had existed prior to tcp support while still (hopefully) allowing for tcp operation to function properly.

If not we need to address the comment in raft_server_client_recv_ignore_request() which incorrectly states:

 /* Lookup the client in the ctl-svc-node tree - existence is not                                     
   * mandatory.                                                                                        
   */

[kitwestneat]

That looks doable. I think we'd have to start sending the port #s in the handshake since the TCP from port is a random number.

Do you think it'd be alright to replace from with csn in the calls to ri_server_recv_cb and ri_client_recv_cb? That would move a lot of message validation to raft_net, which I think makes sense.

(Sorry for not responding earlier, I'm still getting used to github's workflow)