Re: very high latency network problem : 2
Fred M Goolsby <fmgoolsb-5wfLAf0ZXl5/[email protected]>
| Newsgroups | gmane.comp.hardware.microcontrollers.tini |
|---|---|
| Message-ID | <[email protected]> |
Thank you for your reply,
>You are seeing a non-changing timeout retry
>operation?
I can sniff the connection and see an outbound packet. After about 450 ms,
having not seen a TCP ack from the receiver, the
TINI (assuming the packet was lost ?) transmits the packet a second time.
After another 450 ms, a third transmission. And after a third 450 ms the
TINI shuts down the window (accepting no more data). We then see the ack
from the original packet but the port is now closed.
I think we are definitely seeing a non-changing backoff timer.
Fred
Bryan Armstrong
<[email protected] To: Fred M Goolsby <[email protected]>
alsemi.com> cc: [email protected]
Subject: Re: [TINI]very high latency network problem : 2
02/23/04 11:25 AM
Fred M Goolsby wrote:
> Sorry to repost this but I got no responce the first time I posted this
> back in december.
>
> It appears that TINI TCP/IP does not follow the Geometric Backoff
algorithm
> for the retransmit timer. This has not been a problem until
> we started dealing with VERY SLOW networks (800 - 900 ms latency).
>
> Has anyone dealt with anything similar to this? Is there any workaround
> short of switching from TCP to UDP ?
Robert might have more insight on this, but as I understand TINI should
support the backoff. You are seeing a non-changing timeout retry
operation? This should be doubling each time.
--
--
Bryan Armstrong
bryan.armstrong-6tN4nzCoH/[email protected]
_______________________________________________
TINI mailing list
TINI-6tN4nzCoH/[email protected]
To UNSUBSCRIBE, edit your profile, or see list archives:
http://lists.dalsemi.com/mailman/listinfo/tini