htb and UDP packages bigger than 1500

htb and UDP packages bigger than 1500

[Arkadiusz Miskiewicz]

Hello,

I noticed that HTB doesn't properly limit traffic if someone sends UDP 
packages bigger than 1500. 

Does HTB have some problems/known limits in this area?

There is other traffic in that class and when I drop udp packets bigger than 
1500 then remaining traffic is limited properly to correct value.

udp part of tcpdump log:
17:35:24.041364 IP 172.16.4.23.3185 > 80.51.230.58.9267: UDP, length 35
17:35:24.048834 IP 172.16.4.23.46073 > 83.53.197.205.50882: UDP, length 35
17:35:24.071234 IP 172.16.4.23.2904 > 81.38.28.69.26349: UDP, length 35
17:35:24.076924 IP 201.232.209.115.50750 > 172.16.4.23.57590: UDP, length 8251
17:35:24.140895 IP 90.242.120.252.11982 > 172.16.4.23.19111: UDP, length 35
17:35:24.140976 IP 79.186.64.121.47130 > 172.16.4.23.19111: UDP, length 35
17:35:24.141039 IP 24.150.182.77.55984 > 172.16.4.23.41279: UDP, length 29
17:35:24.141120 IP 83.37.212.153.25153 > 172.16.4.23.32824: UDP, length 8251
17:35:24.154874 IP 172.16.4.23.19111 > 90.242.120.252.11982: UDP, length 8251
17:35:24.210940 IP 83.8.18.150.28955 > 172.16.4.23.24825: UDP, length 8251
17:35:24.240382 IP 172.16.4.23.19111 > 79.186.64.121.47130: UDP, length 8251
17:35:24.272529 IP 83.19.224.219.32977 > 172.16.4.23.19111: UDP, length 35
17:35:24.307164 IP 85.219.10.150.18601 > 172.16.4.23.51986: UDP, length 8251
17:35:24.312335 IP 83.26.249.97.10137 > 172.16.4.23.9383: UDP, length 8251
17:35:24.404250 IP 83.19.224.11.1833 > 172.16.4.23.21258: UDP, length 8251
17:35:24.467562 IP 196.206.89.182.58764 > 172.16.4.23.19111: UDP, length 35
17:35:24.560058 IP 172.16.4.23.50417 > 82.5.204.164.1024: UDP, length 25
17:35:24.563842 IP 172.16.4.23.24825 > 83.8.18.150.28955: UDP, length 35
17:35:24.567316 IP 172.16.4.23.59727 > 195.60.65.36.61323: UDP, length 35
17:35:24.569976 IP 83.11.67.228.31949 > 172.16.4.23.56823: UDP, length 8251
17:35:24.617104 IP 172.16.4.23.28945 > 76.11.24.115.13887: UDP, length 29
17:35:24.619235 IP 172.16.4.23.21258 > 83.19.224.11.1833: UDP, length 35
17:35:24.626488 IP 172.16.4.23.9383 > 83.26.249.97.10137: UDP, length 35
17:35:24.640367 IP 172.16.4.23.47366 > 91.146.230.1.38928: UDP, length 25
17:35:24.644314 IP 90.242.120.252.1955 > 172.16.4.23.19111: UDP, length 35
17:35:24.652024 IP 81.184.124.145.22454 > 172.16.4.23.57089: UDP, length 8251

-- 
Arkadiusz Miśkiewicz        PLD/Linux Team
arekm / maven.pl            http://ftp.pld-linux.org/

Re: htb and UDP packages bigger than 1500

[Arkadiusz Miskiewicz]

On Tuesday 11 of December 2007, Jarek Poplawski wrote:
> On 10-12-2007 18:10, Arkadiusz Miskiewicz wrote:
> > Hello,
> >
> > I noticed that HTB doesn't properly limit traffic if someone sends UDP
> > packages bigger than 1500.
> >
> > Does HTB have some problems/known limits in this area?
> >
> > There is other traffic in that class and when I drop udp packets bigger
> > than 1500 then remaining traffic is limited properly to correct value.
>
> Should work with proper "mtu" parameter (tc qdisc add htb help) or,
> for kernels >= 2.6.23, even without this. If this doesn't help,
> probably more information is required.

2.6.22 there. mtu wasn't specified at tc... Could you point to a git commit 
which "fixes" the issue in 2.6.23 so I could backport it to my 2.6.22?

> Regards,
> Jarek P.

Thanks,
-- 
Arkadiusz Miśkiewicz        PLD/Linux Team
arekm / maven.pl            http://ftp.pld-linux.org/

Re: htb and UDP packages bigger than 1500

[Jarek Poplawski]

On 10-12-2007 18:10, Arkadiusz Miskiewicz wrote:
> Hello,
> 
> I noticed that HTB doesn't properly limit traffic if someone sends UDP 
> packages bigger than 1500. 
> 
> Does HTB have some problems/known limits in this area?
> 
> There is other traffic in that class and when I drop udp packets bigger than 
> 1500 then remaining traffic is limited properly to correct value.
> 

Should work with proper "mtu" parameter (tc qdisc add htb help) or,
for kernels >= 2.6.23, even without this. If this doesn't help,
probably more information is required.

Regards,
Jarek P.

Re: htb and UDP packages bigger than 1500

[Jarek Poplawski]

On Tue, Dec 11, 2007 at 09:17:30AM +0100, Arkadiusz Miskiewicz wrote:
> On Tuesday 11 of December 2007, Jarek Poplawski wrote:
> > On 10-12-2007 18:10, Arkadiusz Miskiewicz wrote:
> > > Hello,
> > >
> > > I noticed that HTB doesn't properly limit traffic if someone sends UDP
> > > packages bigger than 1500.
> > >
> > > Does HTB have some problems/known limits in this area?
> > >
> > > There is other traffic in that class and when I drop udp packets bigger
> > > than 1500 then remaining traffic is limited properly to correct value.
> >
> > Should work with proper "mtu" parameter (tc qdisc add htb help) or,
> > for kernels >= 2.6.23, even without this. If this doesn't help,
> > probably more information is required.
> 
> 2.6.22 there. mtu wasn't specified at tc... Could you point to a git commit 
> which "fixes" the issue in 2.6.23 so I could backport it to my 2.6.22?

http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.23.y.git;a=commitdiff;h=c9726d6890f7f3a892c879e067c3ed839f61e745

From: Ranjit Manomohan <ranjitm@google.com>
Date: Wed, 11 Jul 2007 05:43:16 +0000 (-0700)
Subject: [NET_SCHED]: Make HTB scheduler work with TSO.
X-Git-Tag: v2.6.23-rc1~1109^2~22

But this is done with two different ways. And btw., if there is such
a big difference in sizes it could be not very accurate after all.

Jarek P.

Re: htb and UDP packages bigger than 1500

[Arkadiusz Miskiewicz]

On Tuesday 11 of December 2007, Jarek Poplawski wrote:
> On Tue, Dec 11, 2007 at 09:17:30AM +0100, Arkadiusz Miskiewicz wrote:
> > On Tuesday 11 of December 2007, Jarek Poplawski wrote:
> > > On 10-12-2007 18:10, Arkadiusz Miskiewicz wrote:
> > > > Hello,
> > > >
> > > > I noticed that HTB doesn't properly limit traffic if someone sends
> > > > UDP packages bigger than 1500.
> > > >
> > > > Does HTB have some problems/known limits in this area?
> > > >
> > > > There is other traffic in that class and when I drop udp packets
> > > > bigger than 1500 then remaining traffic is limited properly to
> > > > correct value.
> > >
> > > Should work with proper "mtu" parameter (tc qdisc add htb help) or,
> > > for kernels >= 2.6.23, even without this. If this doesn't help,
> > > probably more information is required.
> >
> > 2.6.22 there. mtu wasn't specified at tc... Could you point to a git
> > commit which "fixes" the issue in 2.6.23 so I could backport it to my
> > 2.6.22?
>
> http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.23.y.git;a=commi
>tdiff;h=c9726d6890f7f3a892c879e067c3ed839f61e745
>
> From: Ranjit Manomohan <ranjitm@google.com>
> Date: Wed, 11 Jul 2007 05:43:16 +0000 (-0700)
> Subject: [NET_SCHED]: Make HTB scheduler work with TSO.
> X-Git-Tag: v2.6.23-rc1~1109^2~22

Backported.

> But this is done with two different ways. And btw., if there is such
> a big difference in sizes it could be not very accurate after all.

Hm, here htb qdisc is attached to imq devices which have mtu 16000 by default 
(in latest patch; previously that was 1500). I'll just lower it to 1500 for 
now. Thanks.

> Jarek P.

-- 
Arkadiusz Miśkiewicz        PLD/Linux Team
arekm / maven.pl            http://ftp.pld-linux.org/