Re: [Dime] Diameter retransmission -- to use same hop-by-hop and end-to-end IDs?

[Jouni Korhonen <[e-mail]>]

Hi,

On top of my head.. and in a tired condition :)

3/4/2016, 7:55 AM, Dave Dolson kirjoitti:
> Hello Diameter experts,
>
> We’ve been looking to the specs in order to answer the question as to
> whether a
>
> retransmitted Diameter request MUST/SHOULD/SHOULD NOT/MAY use the same
>
> hop-by-hop and end-to-end identifiers as the original request.
>
> Can anyone point to the position of the standards, or ad hoc standards
> in this regard?

RFC6733 Section 3 says "The sender MUST ensure that the Hop-by-Hop 
Identifier in a request is unique on a given connection at any given 
time". So if a different transport connection is used as a result of 
retransmission, the h-b-h id could potentially be different than in the 
original request. If different h-b-h ids were used over the same 
connection for the same buffered request messages that should also 
work.. I do not recall any text specifically prohibiting that. Not that 
I would find this kind of behaviour good, either..

RFC6733 Section 3 says "The End-to-End Identifier MUST NOT be modified 
by Diameter agents of any kind." Also "Duplicate requests SHOULD cause 
the same answer to be transmitted (modulo the Hop-by-Hop Identifier 
field and any routing AVPs that may be present), and they MUST NOT 
affect any state that was set when the original request was processed." 
And Section 5.5.4. says "The End-to-End Identifier field in the Diameter 
header along with the Origin-Host AVP MUST be used to identify duplicate 
messages."

The combination of Origin-Host and e-t-e id must be unique for duplicate 
detection. Now if either one changes the receiver potentially has issues 
determining correctly to which previously seen message the request was a 
retransmission for.

> A secondary question is whether an agent MUST/SHOULD/SHOULD NOT/MAY
>
> use the same hop-by-hop identifier when forwarding a retransmitted
> request that it used the
>
> first time the message was seen.
>
> My sense is that an agent is not required to do so, but may it do so?

See above.

- JOuni

>
> Thanks in advance,
>
> David Dolson
>
> Senior Software Architect, Sandvine Inc.
>
>
>
> _______________________________________________
> DiME mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/dime
>