RE: [Ipsec] Rekeying IKE_SAs with the CREATE_CHILD_SA exhange

This is actually one of those issues we haven't reached 
consensus yet.

It seems that we have consensus on the following:

  - It is mandatory to include one or more type 4 (D-H) transform 
    substructures in the SA payload when rekeying the IKE_SA
    (Section 3.3.3).

  - Doing a new Diffie-Hellman exchange is not mandatory when 
    rekeying the IKE_SA (e.g. Section 2.18 says that g^ir term
    is optional). Thus, although including a D-H transform
    substructure is mandatory, it can contain value "NONE".

  - Usually implementations will want to do a new D-H exchange
    when the IKE_SA is rekeyed; i.e., while NONE is allowed here, 
    it's probably not the most common case.

  - Doing D-H is not mandatory when rekeying or creating a CHILD_SA.

  - When you're rekeying or creating a CHILD_SA, and you're not
    doing D-H, the KEi/KEr payloads are not included.

My interpretation of the spec is that

  - When you're rekeying the IKE_SA, and you're not doing D-H,
    the KEi/KEr payloads are not included.

But at least Tero and Paul disagreed with this conclusion back
in April (i.e., you have to include a dummy KEi/KEr payloads
even when you're not doing D-H --- but only the IKE_SA case,
not in the CHILD_SA case)...

There hasn't been much discussion about this since, though..

Best regards,
Pasi 

--------------------
> From: Alejandro Perez Mendez
> Sent: 27 October, 2005 11:16
> To: IPsec WG
> Subject: [Ipsec] Rekeying IKE_SAs with the CREATE_CHILD_SA exhange
>
> In order to get a better clarification document, we report a
> misprint found in the Appendix A.
>       
> The clarifications document says in its section 5.1:
>
>    << KEi and KEr are required for rekeying an IKE_SA.>>
>       
> But in the "Appendix A. Exchanges and payloads" in the section
> "A.5. CREATE_CHILD_SA exchange for rekeying the IKE_SA" shows:
>
>    << A.5.  CREATE_CHILD_SA exchange for rekeying the IKE_SA
>
>    request             --> SA, Ni, [KEi]
>       
>    response            <-- SA, Nr, [KEr] >>
>       
> where KEi and KEr seems to be optionals.
>       
> Regards!
>       
> Alejandro Pérez Méndez
> Pedro J. Fernández Ruiz

_______________________________________________
Ipsec mailing list
Ipsec@ietf.org
https://www1.ietf.org/mailman/listinfo/ipsec