| To: | kuznet@xxxxxxxxxxxxx |
|---|---|
| Subject: | Re: snd_cwnd drawn and quartered |
| From: | Werner Almesberger <wa@xxxxxxxxxxxxxxx> |
| Date: | Wed, 15 Jan 2003 16:37:11 -0300 |
| Cc: | netdev@xxxxxxxxxxx, chengjin@xxxxxxxxxxxxxx |
| In-reply-to: | <200301151843.VAA16069@xxxxxxxxxxxxx>; from kuznet@xxxxxxxxxxxxx on Wed, Jan 15, 2003 at 09:43:29PM +0300 |
| References: | <20030115152504.B1521@xxxxxxxxxxxxxxx> <200301151843.VAA16069@xxxxxxxxxxxxx> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
kuznet@xxxxxxxxxxxxx wrote: > We receive some amount of dupacks for segments >high_seq, before ACK > for retransmitted 99th segment arrives and terminates recovery. You meaned > this, right? :-) Yes, that's it. > Yup, this is bug. OK, this case is sorted out. I think that small fix > proposed by you is enough for beginning. I think, in addition to this fix, you also need to do RTO upon loss of retransmission. Otherwise, my fix would make you halve cwnd only once in this case. - Werner -- _________________________________________________________________________ / Werner Almesberger, Buenos Aires, Argentina wa@xxxxxxxxxxxxxxx / /_http://www.almesberger.net/____________________________________________/ |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: snd_cwnd drawn and quartered, kuznet |
|---|---|
| Next by Date: | Re: SCTP path mtu support needs some ip layer support., Mika Liljeberg |
| Previous by Thread: | Re: snd_cwnd drawn and quartered, kuznet |
| Next by Thread: | example showing how cwnd gets to one, Cheng Jin |
| Indexes: | [Date] [Thread] [Top] [All Lists] |