Different close_notify behaviour in DTLS 1.2 and DTLS 1.3 about draft-westerlund-tsvwg-dtls-over-sctp-bis

[gloinul]

I can think of two likely reasonable solutions to this issue.

A) Require DTLS 1.3 as that avoids the issue with the DTLS 1.2 Close Notify alert behavior. It also avoids version related negotiation issues at this stage of the protocol but they can resurface with a DTLS 1.4. Also it requires DTLS 1.3 stack immediately.

B) We define a DTLS/SCTP message and registers a PPID for it and identify it that way. Thus the initiating party tells the peer that it want to close and have stopped incoming and have drained its queue. When peer process this message on DTLS/SCTP layer it will close for more incoming data and drain its buffer. When that is accomplished it sends the close_alert that is then echoed back by the original requesting endpoint.

So lets consider the impact of the options and make a recommendation to the WG.