Re: [netlmm] Retransmitting a PBU when HI=1 and MN LL ID is non-zero (draft-ietf-netlmm-proxymip6-18)

Hi Tero,

On Thu, 24 Jul 2008, Tero Kauppinen wrote:

> Hello
>
> Let's consider the following case where MAG sends a PBU where HI=1
> (Attachment over a new interface) and MN LL ID is non-zero (no HNP).
> When LMA receives the initial PBU it creates a new BCE and sends back a
> PBA. If that PBA is lost, MAG will retransmit the PBU. If MAG leaves HI
> to 1 (I didn't find any indication from the spec that it should do
> otherwise) and retransmits the request, LMA will come to a conclusion
> that this is 'Binding Lifetime Extension - After handoff'. This because:
>

This is the correct behaviour. The LMA has the state for that session
and it should just update that session and not create a new session.
>From the MAG perspective, it never received the ACK and it just
retransmits the PBU with the same parameters, except seq num/time
stamp values.

> Section 5.4.1.1. skipped (no HNP)
> Section 5.4.1.2. hit (MN LL ID non-zero) checks 1) and 2) match -> a
> request for updating an existing Binding Cache entry
> Section 5.3.1. check 13 results in the third option (Binding Lifetime
> Extension - After handoff) because HI=1
>
> However, if HI would be set to a value of 5 (Handoff state not changed)
> for retransmissions, LMA would choose (Binding Lifetime Extension- No
> handoff) which I think should happen in this case. However, I'm still a
> bit unsure if this would break some other cases. Even if the initial PBU
> is not received by LMA, setting HI to 5 would result in the correct
> behavior because there is no existing BCE with matching {MN ID, AT, MN
> LL} triplet and thus LMA will create a new BCE regardless of the value
> of HI.
>

The MAG cannot set HI=5 (Re-Reg), we clearly require all the session
parameters in re-reg or de-reg messages. In this case, the MAG would not
even know the HNP as it never received the PBA.

Thanks
Sri

> Br,
>
> Tero Kauppinen                                 Tel: +358 9 299 3057
> Research Scientist, NomadicLab, IP Networks,   Fax: +358 9 299 3535
> Ericsson Research, Corporate Unit.             Mobile: +358 40 590 7560
> Oy L M Ericsson Ab
> mailto:tero.kauppinen@ericsson.com
> _______________________________________________
> netlmm mailing list
> netlmm@ietf.org
> https://www.ietf.org/mailman/listinfo/netlmm
>
_______________________________________________
netlmm mailing list
netlmm@ietf.org
https://www.ietf.org/mailman/listinfo/netlmm