Re: [Tsvwg] Re: WGLC on draft-ietf-tsvwg-prsctp-00

Sourabh Ladha (EED) wrote:

>maybe I am missing something.
>
>My reading of the draft says that in F3> "If" any delay is applied to sending of FW-TSN chunk, it SHOULD NOT exceed 200 ms and MUST not exceed 500ms. I dont understand how this "at LEAST 1 RTO wait" in sending the FWD TSN comes. 
>
>Anyway, one of the reasons of pointing this out was to to cause a thought process for blanton&allman's draft or any other work that "may" get influenced by PR-SCTP. If not, well and goo
>

You can't send the FWD-TSN until the left edge of the number
of TSN's is concurrent with your "abandoned chunk"

So

Consider

-----TSN101---->
-----TSN102----> XlostX
-----TSN103---->
..

at some point, 102 is abandoned.

the FWD-TSN(102)----->

can NOT be sent until:

1) TSN101 is acknowledged (aka the cumack is to the immediate left
     of the chunk to be forwarded... the left edge).

2) You would normally make a retransmission decision.

Just because you abandon it does not mean you tell the peer.. you have
to follow all or the normal RFC2960 procedures and ONLY when you
have marked it to abandon and you would have retransmitted and the
left edge is aligned will you send the FWD-TSN... you may decide
to delay a while... aka the 200ms but you send only at the normal
retrans opp..

R

>
>
>  
>

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