[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: draft-ietf-shim6-proto-06 : Sending R2 in I2-SENT state



Hi Sebastien,


El 09/11/2006, a las 19:23, Sébastien Barré escribió:

Hi,

I have still a question regarding the proto draft. I'm sorry to post it so late, I was delayed in re-reading the draft in depth, and could not be present at IETF meeting. So, here is :

(p76) In section 10.4, it is stated that upon reception of an update request and if the state is I2-SENT or I2bis-SENT, we need to reply with an R2 then process the update. I agree with processing the update request, but I don't understand why to send an R2 : If the responder is in established state, it will anyway drop the R2. And I suppose that if the responder sent an update request, this implies that it is already in established state.

I suppose that you have a good reason of doing this, because I see in section 12.1 that in state I2-SENT, if payload is received, an R2 must also be sent (which I see equally as useless, since silently discarded by the peer). Also, in section 10.5, the same behaviour is defined. Can you explain the reason for this ?

btw, in appendix B, the simplified state machine is inconsistent with the text, saying that in I2-SENT state, an I2 (not an R2) must be replied to payload extension header packets, because the peer probably has sent an R2 which was lost. This appears to me more appropriate than what is said in sections 10.4, 10.5 and 12.1.

this is the correct behaviour, but it seems that i corrected one place and not the other....

the idea is that if a peer is in I2-SENT/I2bis-SENT state it means that it hasn't received a R2 message back, so it can move to ESTABLISHED state, right?

So, if the other peer is in ESTABLISHED state and is sending UPDATE/payload, this basically means that the R2 message was lost, ok? So we need the peer to retransmit the I2/I2bis message again so that the peer retransmit the R2 message

i will fix this,

thanks marcelo




otoh, the simplified state machine diagram in appendix B.1 agrees with the text (so disagrees with appendix B).

Sébastien.

--
Sébastien Barré
Researcher,
CSE department, UCLouvain, Belgium