Raised by Christian Dickmann on the mailing list:
Hancock, Robert wrote:
> hi christian,
>
>
>> -----Original Message-----
>> From: Christian Dickmann [mailto:mail@christian-dickmann.de]
>> Sent: 09 March 2007 21:16
>> To: nsis@ietf.org
>> Subject: [NSIS] Use of 3-Way Handshake in D-Mode case
>>
>> Hi Robert,
>>
>> section 4.3.3 "Message Transmission" states:
>> "If a messaging association is not required, the message is sent in
>> D-mode. The processing in this case depends on the message type
>> and
>> whether routing state exists or not. ...
>> - If the message is not a Query, and routing state exists, ...
>> - If the message is a Query, ...
>> - If no routing state exists, GIST can attempt to use Q-mode as in
>> the Query case. ..."
>>
>> Thus, for a message that is not a Query and where no routing state
>> exists, the third rule applies. The third rule gives only the option
>> of sending the message in Q-mode. Does that mean, that there is no
>> way to carry out a 3-way handshake just for transport of D-Mode
>> messages?
>>
>
> that is not the intention. this text hasn't changed for a while; the
> scope (in fact, the scope of this whole section 4.3) is supposed to be
> 'how do you send a message based on whatever routing state you already
> have', and 4.4 explains how to set up routing state if you don't have
> it. it may be that that final bullet should be extended to say that
> lack of routing state can be used as a trigger to make a handshake
> take place.
>
Very good. I was quite worried after reading that text :) So I think that this
short addition to the last bullet would be a good idea.
>> The SendMessage primitive has a parameter that suggests "whether GIST
>> should attempt to set up or maintain forward routing state." Also
>> text like "A Confirm MUST be sent in C-mode if a messaging
>> association is being used for this routing state" suggests that there
>> is case of having routing state without C-Mode.
>> Am I missing something?
>>
>
> no, it is quite possible for the handshake to take place without
> creating an MA (although you SHOULD only send messages outside an MA
> when you have provisioned bandwidth for signalling traffic - this is a
> congestion control issue, and will be in the next version, see
> http://nsis.srmr.co.uk/cgi-bin/roundup.cgi/nsis-ntlp-issues/issue208).
>
Thanks for pointing me to this issue, answered another question I had.
Christian
|
