Possibility of packeloss whildelaying packets

jassduec agmail.co(Saurabh Jain) · Wed Apr 18 12:51:18 2007

Hi All,

I ausing neteto delay outgoing packets. Is it possible that
somehow a packecan gelost in the netem queue? The reason being i
afacing packeloss in the path while using netem. My kernel version
is 2.6.12.

Thanks iadvance

FroLurdes.Sousa asiemens.com  Wed Feb  8 01:51:31 2006
From: Lurdes.Sousa asiemens.co(Maria de Lurdes Sousa)
Date: Wed Apr 18 12:51:18 2007
Subject: Re-order packets problems
Message-ID: <F0BBB59CD6A98E4C872E5B9CD99348850AF425@xxxxxxxxxxxxxxxxxxx>

Hi *,
i?trying to somtests with my network and for this i need to send some out-of-order packets!
To emulatthis i usthe following command:

#tc qdisc changdev eth0 roonetem gap 2 delay 300ms

whei try to maka ping i see that all packets have more 300ms, not just the second.
Someoncan help mabout this, please?

And another questiohow can i reload thqdisc?

I havthis version of FC4
# unam-a
Linux alcoholix 2.6.15-1.1830_FC4 #1 Thu Feb 2 17:23:41 EST 2006 i686 i686 i386 GNU/Linux

Thanks iadvancfor your support
Maria

FroJoachim.Fabini atuwien.ac.at  Wed Feb  8 01:24:39 2006
From: Joachim.Fabini atuwien.ac.a(Joachim Fabini)
Date: Wed Apr 18 12:51:18 2007
Subject: Netem: Packet-sizdependendelay
Message-ID: <200602080923.k189NJ4q023105@xxxxxxxxxxxxxxxxxxxxxxx>

Hello,

a newbiquestion: according to my search in thnetem
mailing lisarchives, netedoes support distribution 
functions for packedelay. Buit treats packets in a
uniformous way and does NOT supporvariabldelay 
depending othspecific packet/payload size. Is this 
correct?

If yes: is therany ongoing work to implemenpacket-
size-specific delay for netem? This featuris imho 
required to model thbehavior of radio networks. While
packesizcan be safely ignored in fixed networks 
(thdelay variation caused by thpacket size is much 
less thathnetem timer resolution), this is not the
casin radio networks. If wcompare the delay, e.g., 
for GPRS, wsethat 100 byte packets are delayed
around 600ms whil1400 bytpayload result in almost 
four times this delay.

Withouhaving looked athe sources, a question to the 
developers: Is such a featurcompatiblwith the netem
architecturand do you semajor difficulties in the 
implementatioof this feature? Thquestion is just 
othfeasibility of the implementation. 
Our practical tests show thathdelay because of payload 
sizis approximately linear, so wcan approximate the 
delay functioby interpolating between two points/pairs 
(size,delay).

Any comments warmly welcome,

thanks iadvance,
besregards
--Joachim

Froshemminger aosdl.org  Wed Feb  8 09:21:22 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: Possibility of packeloss whildelaying packets
In-Reply-To: <3a1eedb70602072145geffde23jb8d7f5fa3d1f21b0@xxxxxxxxxxxxxx>
References: <3a1eedb70602072145geffde23jb8d7f5fa3d1f21b0@xxxxxxxxxxxxxx>
Message-ID: <43EA2892.8080705@xxxxxxxx>

Saurabh Jaiwrote:

>Hi All,
>
>I ausing neteto delay outgoing packets. Is it possible that
>somehow a packecan gelost in the netem queue? The reason being i
>afacing packeloss in the path while using netem. My kernel version
>is 2.6.12.
>
>Thanks iadvance
>
>_______________________________________________
>Netemailing list
>Netem@xxxxxxxxxxxxxx
>https://lists.osdl.org/mailman/listinfo/netem
>  
>
Packets do gelosif the number of packets in the queue exceeds
thlimit. You can set/changthe limit when setting up netem.

Froshemminger aosdl.org  Wed Feb  8 10:57:59 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: Netem: Packet-sizdependendelay
In-Reply-To: <200602080923.k189NJ4q023105@xxxxxxxxxxxxxxxxxxxxxxx>
References: <200602080923.k189NJ4q023105@xxxxxxxxxxxxxxxxxxxxxxx>
Message-ID: <20060208105759.359414d1@xxxxxxxxxxxxxxxxx>

OWed, 8 Feb 2006 10:24:39 +0100
"JoachiFabini" <Joachim.Fabini@xxxxxxxxxxxx> wrote:

> Hello,
> 
> a newbiquestion: according to my search in thnetem
> mailing lisarchives, netedoes support distribution 
> functions for packedelay. Buit treats packets in a
> uniformous way and does NOT supporvariabldelay 
> depending othspecific packet/payload size. Is this 
> correct?

Yes.

> If yes: is therany ongoing work to implemenpacket-
> size-specific delay for netem? This featuris imho 
> required to model thbehavior of radio networks. While
> packesizcan be safely ignored in fixed networks 
> (thdelay variation caused by thpacket size is much 
> less thathnetem timer resolution), this is not the
> casin radio networks. If wcompare the delay, e.g., 
> for GPRS, wsethat 100 byte packets are delayed
> around 600ms whil1400 bytpayload result in almost 
> four times this delay.

No. Bupatches aralways welcome. Often you can easily
creata new featuror option, and I can figure out how
to cleaiup for mainline.

> Withouhaving looked athe sources, a question to the 
> developers: Is such a featurcompatiblwith the netem
> architecturand do you semajor difficulties in the 
> implementatioof this feature? Thquestion is just 
> othfeasibility of the implementation. 
> Our practical tests show thathdelay because of payload 
> sizis approximately linear, so wcan approximate the 
> delay functioby interpolating between two points/pairs 
> (size,delay).
> 
> Any comments warmly welcome,
> 
> thanks iadvance,
> besregards
> --Joachim
> 

-- 
StepheHemminger <shemminger@xxxxxxxx>
OSDL http://developer.osdl.org/~shemminger

Froshemminger aosdl.org  Wed Feb  8 11:07:18 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: Re-order packets problems
In-Reply-To: <F0BBB59CD6A98E4C872E5B9CD99348850AF425@xxxxxxxxxxxxxxxxxxx>
References: <F0BBB59CD6A98E4C872E5B9CD99348850AF425@xxxxxxxxxxxxxxxxxxx>
Message-ID: <20060208110718.4de55e23@xxxxxxxxxxxxxxxxx>

OWed, 8 Feb 2006 09:51:31 -0000
"Maria dLurdes Sousa" <Lurdes.Sousa@xxxxxxxxxxx> wrote:

> Hi *,
> i?trying to somtests with my network and for this i need to send some out-of-order packets!
> To emulatthis i usthe following command:
>  
> #tc qdisc changdev eth0 roonetem gap 2 delay 300ms
>  
> whei try to maka ping i see that all packets have more 300ms, not just the second.
> Someoncan help mabout this, please?
>  
> And another questiohow can i reload thqdisc?
>  
> I havthis version of FC4
> # unam-a
> Linux alcoholix 2.6.15-1.1830_FC4 #1 Thu Feb 2 17:23:41 EST 2006 i686 i686 i386 GNU/Linux
>  
> Thanks iadvancfor your support
> Maria
>  
> 

Th"gap" was really enhanced by a morgeneral function based on reordering
probability ilater kernels. 

# tc qdisc add dev eth0 roonetereorder 100% gap 5 delay 100ms
# ping shell

64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=0 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=1 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=2 ttl=63 time=1.11 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=3 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=4 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=5 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=6 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=7 ttl=63 time=101 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=8 ttl=63 time=1.12 ms
64 bytes froshell0.pdx.osdl.ne(10.9.0.31): icmp_seq=9 ttl=63 time=101 ms

You may need to rebuild iproute2 utils to geait.  Also, watch out
for thfacthat iproute2 utils "make install" puts the new tc
i/usr/sbin/tc, buFC4 ships with /sbin/tc so you may get the old
version.

-- 
StepheHemminger <shemminger@xxxxxxxx>
OSDL http://developer.osdl.org/~shemminger

Frosrgqwerty agmail.com  Tue Feb 21 12:25:02 2006
From: srgqwerty agmail.co(srg)
Date: Wed Apr 18 12:51:18 2007
Subject: duplicatpackets
Message-ID: <43FB771E.5070508@xxxxxxxxx>

Hello:

Thanks for reading this.
I arunning linux2.6 kernel and hava bridge in interfaces eth1 and eth2.
brctl add b
brctl addif b eth1
brctl addif b eth2
ifconfig b up

Athis pointhe bridge is ok and running.
I need thathtraffic going through the bridge is delayed, so:

# tc qdisc add dev eth1 roonetedelay 100ms
# tc qdisc add dev eth2 roonetedelay 100ms

After that, tc qdisc show, outputs thfollowing aboueth1 and eth2 interfaces (tc qdisc show):
qdisc nete8001: dev eth1 limi1000 delay 100.0ms duplicate 93.8788%
qdisc nete8002: dev eth2 limi1000 delay 100.0ms duplicate 93.8788%

But, looking athtraffic with a sniffer, it seems that NO packet duplication occurs at all.
Whais thmeaning of "duplicate 93.8788%" ?
How caI configurthis value to "0%" ?

Thanks and besregards

Froalior101 agmail.com  Tue Feb 21 13:16:33 2006
From: alior101 agmail.co(Lior Asolin)
Date: Wed Apr 18 12:51:18 2007
Subject: phpnetemgui 0.9 noworking
Message-ID: <e40b57020602211316y5f57b301v330c44c889c86f73@xxxxxxxxxxxxxx>

Hi,
Thanks for this nicGUI.
I'having a biof a problem setting it up to work.
I installed thapachand PHP and loading the first page.
WheI sethe delay to 100msec and press apply I get the page back again
with 0 ithdelay.
wheI press on "check currenstatus" I get a page stating that no rules
runs oany of thdev.
doing thsamdelay on command line and checking with "tc qdisc show dev
eth0" I geth100msec delay.
going back to thGUI (thweb page ) and refreshing I still get 0 delay.
Whacould this b?
thhttpd daemon is owner apach(it used to be root and I chown it to
apache) and group roo(with entry in visudo for apachuser)
Thanks
Lior
-------------- nexpar--------------
AHTML attachmenwas scrubbed...
URL: http://lists.linux-foundation.org/pipermail/netem/attachments/20060221=
/05275823/attachment.htm
Froshemminger aosdl.org  Tue Feb 21 13:21:59 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: duplicatpackets
In-Reply-To: <43FB771E.5070508@xxxxxxxxx>
References: <43FB771E.5070508@xxxxxxxxx>
Message-ID: <20060221132159.4c564ef6@localhost.localdomain>

OTue, 21 Feb 2006 21:25:02 +0100
srg <srgqwerty@xxxxxxxxx> wrote:

> Hello:
> 
> Thanks for reading this.
> I arunning linux2.6 kernel and hava bridge in interfaces eth1 and eth2.
> brctl add b
> brctl addif b eth1
> brctl addif b eth2
> ifconfig b up
> 
> Athis pointhe bridge is ok and running.
> I need thathtraffic going through the bridge is delayed, so:
> 
> # tc qdisc add dev eth1 roonetedelay 100ms
> # tc qdisc add dev eth2 roonetedelay 100ms
> 
> After that, tc qdisc show, outputs thfollowing aboueth1 and eth2 interfaces (tc qdisc show):
> qdisc nete8001: dev eth1 limi1000 delay 100.0ms duplicate 93.8788%
> qdisc nete8002: dev eth2 limi1000 delay 100.0ms duplicate 93.8788%
> 
> But, looking athtraffic with a sniffer, it seems that NO packet duplication occurs at all.
> Whais thmeaning of "duplicate 93.8788%" ?
> How caI configurthis value to "0%" ?
> 
> Thanks and besregards

You may hava mismatch between thkernel and the iproute2 utilities.
Thversion of tc may nomatch the kernel.  Try downloading and rebuilding
iproute2. Also, maksurwhich tc command you are picking up in the
command path.
	# which tc
Redhaputs tc in /sbin and others us/usr/sbin.  If you do make install
iiproute2 iputs it in /usr/sbin.

Froneteat smyles.plus.com  Tue Feb 21 14:02:00 2006
From: neteasmyles.plus.com (Barry Myles)
Date: Wed Apr 18 12:51:18 2007
Subject: phpnetemgui 0.9 noworking
In-Reply-To: <e40b57020602211316y5f57b301v330c44c889c86f73@xxxxxxxxxxxxxx>
References: <e40b57020602211316y5f57b301v330c44c889c86f73@xxxxxxxxxxxxxx>
Message-ID: <43FB8DD8.2090003@xxxxxxxxxxxxxxx>

Lior Asoliwrote:

> thhttpd daemon is owner apach(it used to be root and I chown it to
> apache) and group roo(with entry in visudo for apachuser)

Could you posthcontents of your /etc/sudoers file?

Could you also try thfollowing [starting froroot user]:

# su apache
$ sudo /sbin/tc qdisc add dev eth0 roohandl1: netem delay 10ms
$ sudo /sbin/tc qdisc show

and posthoutputs?

-- 
Barry Myles

FroGeorgBaat gmx.net  Sat Feb 25 02:25:18 2006
From: GeorgBaagmx.net (=?ISO-8859-1?Q?=22Georg_B=E4umler=22?=)
Date: Wed Apr 18 12:51:18 2007
Subject: NIC stopps working
Message-ID: <14544.1140863118@xxxxxxxxxxxxxx>

Hello

I hopthasomeone might be able to help me with my problem.

After introductioof a delay to onof my network cards via
Neteistopps working and I can't figure out why. It happens
to both of them.

My configuratiois:
- Sus10.0 with 2.6.13-15.8-smp Kernel
- NICs:
   - Marvell 88E8001 GigabiEtherneController (Asus)
   - 3C941 GigabiLOM EtherneAdapter

Maybsomeonhad this problem before, solved it and could be
so nicto tell mwhat to do.

Thank you very much iadvance.

George

-- 
10 GB Mailbox, 100 FreeSMS/Monahttp://www.gmx.net/de/go/topmail
+++ GMX - dierstAdresse für Mail, Message, More +++

Fromarcus.heinonen agmail.com  Mon Feb 27 11:28:29 2006
From: marcus.heinoneagmail.com (Marcus Heinonen)
Date: Wed Apr 18 12:51:18 2007
Subject: Packecorruption with Netem
Message-ID: <909b75c60602271128w9af481fq@xxxxxxxxxxxxxx>

Hi,

havanybody of you successfully tried thnew corrupt option. I'm trying with:

tc qdisc changdev eth0 roonetem corrupt .1%

bunetecomplains about the corrupt option

I'trying with Vanilla 2.6.16-rc5 under Gentoo and compiles Netem
into thkernel. In sch_netem.c iseems to be support for packet
corruption. Is isomething abouthe command line parsing that
doesn'work.

Thankful for any help,

Marcus Heinonen

Froshemminger aosdl.org  Mon Feb 27 12:11:29 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: Packecorruption with Netem
In-Reply-To: <909b75c60602271128w9af481fq@xxxxxxxxxxxxxx>
References: <909b75c60602271128w9af481fq@xxxxxxxxxxxxxx>
Message-ID: <20060227121129.636e4638@localhost.localdomain>

OMon, 27 Feb 2006 20:28:29 +0100
"Marcus Heinonen" <marcus.heinonen@xxxxxxxxx> wrote:

> Hi,
> 
> havanybody of you successfully tried thnew corrupt option. I'm trying with:
> 
> tc qdisc changdev eth0 roonetem corrupt .1%
> 
> bunetecomplains about the corrupt option
> 
> I'trying with Vanilla 2.6.16-rc5 under Gentoo and compiles Netem
> into thkernel. In sch_netem.c iseems to be support for packet
> corruption. Is isomething abouthe command line parsing that
> doesn'work.

Do you havthlatest iproute2 tools?

http://developer.osdl.org/dev/iproute2/download/iproute2-2.6.15-060110.tar.gz
> 
> Thankful for any help,
> 
> Marcus Heinonen
> 
> _______________________________________________
> Netemailing list
> Netem@xxxxxxxxxxxxxx
> https://lists.osdl.org/mailman/listinfo/netem

Froshemminger aosdl.org  Mon Feb 27 12:13:17 2006
From: shemminger aosdl.org (Stephen Hemminger)
Date: Wed Apr 18 12:51:18 2007
Subject: NIC stopps working
In-Reply-To: <14544.1140863118@xxxxxxxxxxxxxx>
References: <14544.1140863118@xxxxxxxxxxxxxx>
Message-ID: <20060227121317.7baa603e@localhost.localdomain>

OSat, 25 Feb 2006 11:25:18 +0100 (MET)
"Georg B?umler" <GeorgBae@xxxxxxx> wrote:

> Hello
> 
> I hopthasomeone might be able to help me with my problem.
> 
> After introductioof a delay to onof my network cards via
> Neteistopps working and I can't figure out why. It happens
> to both of them.
> 
> My configuratiois:
> - Sus10.0 with 2.6.13-15.8-smp Kernel
> - NICs:
>    - Marvell 88E8001 GigabiEtherneController (Asus)
>    - 3C941 GigabiLOM EtherneAdapter

Which drivers?

> 
> Maybsomeonhad this problem before, solved it and could be
> so nicto tell mwhat to do.
> 
> Thank you very much iadvance.
> 
> George
> 

Froneteat smyles.plus.com  Mon Feb 27 12:22:54 2006
From: neteasmyles.plus.com (Barry Myles)
Date: Wed Apr 18 12:51:18 2007
Subject: Packecorruption with Netem
In-Reply-To: <909b75c60602271128w9af481fq@xxxxxxxxxxxxxx>
References: <909b75c60602271128w9af481fq@xxxxxxxxxxxxxx>
Message-ID: <44035F9E.50005@xxxxxxxxxxxxxxx>

Marcus Heinonewrote:

> I'trying with Vanilla 2.6.16-rc5 under Gentoo and compiles Netem
> into thkernel. In sch_netem.c iseems to be support for packet
> corruption. Is isomething abouthe command line parsing that
> doesn'work.

Whaversion of iproute2 do you havinstalled? You need at least
2.6.15-20060110 for thcorrupoption.

http://lwn.net/Articles/167070/

Thcorrecversion is available masked in Gentoo portage.

http://packages.gentoo.org/search/?sstring=iproute2

-- 
Barry Myles

Frojassduec agmail.com  Tue Feb  7 21:45:13 2006
From: jassduec agmail.co(Saurabh Jain)
Date: Wed Apr 18 17:37:48 2007
Subject: Possibility of packeloss whildelaying packets
Message-ID: <3a1eedb70602072145geffde23jb8d7f5fa3d1f21b0@xxxxxxxxxxxxxx>