Re: I-D Action: draft-ietf-hip-native-nat-traversal-12.txt

Hi Miika,
I was reviewing this section...

> * 4.12.3.  Handling Conflicting SPI Values
>     * Should the Responder send a notify on SPI collision?
>     * Removed text about registering with multiple addresses because I 
>think this does not work with HIP (or at least, requires multihoming)

When there is a SPI collision, it does seem that we would want a new type of NOTIFY to be sent.

Otherwise it seems the Initiator will be stuck in the state I2-SENT, retransmitting the I2 until going back to the failure state, when it can retry the BEX from the beginning again.

Maybe it needs to be an ICMP message (and not NOTIFY) since there is not yet an association between the two peers (RFC 7401 section 4.3).

-Jeff