On Sat, Sep 01, 2001 at 07:17:29PM +0200, Jamie Lokier wrote:
> The appended "tcpdump -i ppp0 -n" trace shows an excessive number of
> retransmits from the remote POP server. The retransmits are triggered
> by excessive duplicate ACKs from the local client. By excessive, I mean
> lots of retransmits of the same data.
The duplicate ACKs should not cause any retransmits (unless the sender
is badly broken), because they contain a high enough ACK number.
The problem is really that a TCP sender cannot recover from a too short RTT
estimate; if the RTT is longer and it doesn't get any acks it'll assume
packet loss and never has a chance to find out about the longer RTT, because
that only works with new ACKs.
The standard initial RTT is 3 seconds; but your RTT is 5.2s.
What you can do is to change the initial RTT on the sender side. On Linux
it can be done with the "irtt" option of route or the equivalent one of
iproute2. Most other OS have a similar way to configure the IRTT.
-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Sep 07 2001 - 21:00:12 EST