From: Wesley Eddy (weddy@elitists.net)
Date: 10/30/03
Date: Thu, 30 Oct 2003 11:12:33 -0500 From: Wesley Eddy <weddy@elitists.net> Subject: Re: tcptrace RTT time evolution; SACK in tcptrace Message-ID: <20031030161233.GA26958@irg.cs.ohiou.edu>
On Thu, Oct 30, 2003 at 08:49:47AM -0500, Joshua Blanton wrote:
> When SACK is enabled, fast retransmits have a different definition
> ...
>
I don't think this is the main issue here, but I could be wrong.
> Juan Francisco Torreblanca wrote:
> > Regarding the question of retransmission due to RTO or packet loss
> > detection, I've been checking tcptrace stats with the relative ethereal
> > trace following Mani suggestion and they don't match: retrx packets 5,
> > triple dupacks 2, but no more than 1 RTO has happened.
> >
I think the problem is that you're confusing events that trigger loss
recovery with lost packets, when there isn't a one-to-one relationship
there. Multiple packets can (and often are) lost in the same outstanding
window, and the loss recovery is only entered once in order to fix them.
If all you want to do is count RTOs, there is an IsRTO() function in rexmit.c
that's used for a couple of things. A slight bit of hacking could probably
use that to count RTO events, but since that function only makes a best
guess (tcptrace doesn't have a crystal ball for looking inside a sender's
state), be a little careful.
-Wes
----------------------------------------------------------------------------
To unsubscribe, send a message with body containing "unsubscribe tcptrace" to
majordomo@tcptrace.org.
This archive was generated by hypermail 2.1.7 : 10/30/03 EST