[e2e] TCP retransmissions at end of data stream

Eric Travis travis at gst.com
Sun Jun 15 11:22:18 PDT 2003


'Agthorr' wrote:

>On Mon, Jun 16, 2003 at 07:14:29PM +0200, Valentin Ossman wrote:
>  
>
>>Whenever an ACK is not received on the data and the transit time-outs,
>>slow start should be started. This means:
>>1. send the first un-acked packet again and wait for ack.
>>2. send the second and third packets again and wait for ack and so on.
>>    
>>
>
>Just so I am clear: after a timeout, I should begin retransmitting all
>unacked data, even when new data is available from the application?
>  
>
In absence of any other information (which is often
the case if you are responding to a timeout), you should
*only* be retransmitting the segment at snduna - especially
if there is new data available.

Your original question covered the special case where
you knew that there would never be any new data arriving
from the application. The fact that you've already
transmitted the FIN counts as "other information" above. ;o)

Eric




More information about the end2end-interest mailing list