Thanks andi for the response but I have some comment to tell.
1- The TBF does not drop packets asI have made the "limit" value to big
(1MB).. Also I have checked that using the -s option in TC where I found no
packets are dropped. Finally I can see the packets released after a while
(using tcpdump) so that means that no drooping is happennig.
2- The application can not be fixed as it is commercial MS product.
3- Setteing the burst size to one full big datagram will make the traffic
nonconforment to the EF statndard and it will be dropped in the next
policing piont.
Thanks
Wael
----- Original Message -----
From: "Andi Kleen" <ak@suse.de>
To: "Wael Ashmawi" <washmawi@ee.upenn.edu>
Cc: <linux-kernel@vger.kernel.org>
Sent: Friday, September 22, 2000 5:16 PM
Subject: Re: traffic shapping problem with fragmented packets
> On Fri, Sep 22, 2000 at 04:56:55PM -0400, Wael Ashmawi wrote:
> > I have tried several ways of doing that the script shown below does the
> > shaping ok but for packet less than the Ethernet MTU. if I start to have
> > fragments everything is missed up. I tried it with pings givinig
> > different packet size in the command line.
>
> The problem is not really solvable unless you fix the application to send
> smaller packets. The only way to shape traffic in IP is to drop packets,
and
> with these big fragments a single packet drop the whole fragment is lost.
TBF
> doesn't know anything about fragments so it'll just arbitarily drop a rate
> which results in at least one fragment dropped from every application
datagram,
> which results in no data transfered. You could try to set the burst size
> >= application datagram size, but even that will only work properly with a
> single stream.
>
>
>
> -Andi
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 21:00:28 EST