Russell Stuart wrote:
The HTB qdisc has a compile time option, HTB_HYSTERESIS,
that trades accuracy of traffic classification for CPU
time. These patches change hysteresis to be a runtime
option under the control of tc.
The effects of HYSTERESIS on HTB's accuracy are significant
(see
yes but I'm not sure if RB three lib is in 2.4.14
On Wed, 7 Jul 2004 [EMAIL PROTECTED] wrote:
Hi All,
Can I backport 2.4.20 kernel version's HTB related changes to 2.4.14 ? Will this
work w/o any issues ?
Please consider this urgent and replay asap.
Thanks in advance,
Reema.
1. In order of priority, we satisfy all leaf classes' rates (while the class
is ?green?)
2. When the leaf classes' rate is reached (all the leaf classes are ?yellow?),
borrow the unused speed from parent classes if they have something to give
(if they are not ?red?). In this case, each leaf
remove cburst 1
thanks!
class htb 1:5500 parent 1:2 leaf 5500: prio 0 quantum 65536 rate 5Mbit
ceil 5Mbit burst 20Kb/8 mpu 0b cburst 8151b/8 mpu 0b level 0
Sent 45107618 bytes 42138 pkts (dropped 0, overlimits 0)
rate 608617bps 548pps backlog 40p
lended: 42098 borrowed: 0 giants: 0
1) 6kbit is really too small it should be at least 10 ..
2) it should workeven with 6k:
- look at stats (tc -s class show dev eth0) before and
after the test - you are interested in drops. Also try
it during the test to look whether queues are build up.
devik
On Sat, 5 Jul 2003, Sergiusz
Don't expect good quality if use local QoS for incomming
data. To have real non-best effort you need to have QoS
on both sides of link - when transmitting packets.
Then yes - HTB should be able to do it using priorities.
But I doubd you can do it if IMQ is in the way.
devik
On Sat, 5 Jul 2003,
and do not participate in mark recognition and appropriate
queuing. it falls back to a best effort, lossy network, regardless of
what you do at the edges.
gerry
Martin Devera wrote:
Don't expect good quality if use local QoS for incomming
data. To have real non-best effort you need to have
you probably need IMQ if eth0 is input for your case
On Thu, 3 Jul 2003, [EMAIL PROTECTED] wrote:
Dear folks,
Here goes my bandwidth manager:
INTERNET
|
|eth0 202.14.41.1
BW.Manager
| |
| +eth1192.168.1.0/24
|
+--eth2192.168.2.0/24
Total incoming bandwidth to eth0 is
you should provide more info, version, script ..
On Thu, 3 Jul 2003 [EMAIL PROTECTED] wrote:
Did someone solve this problem?
I receive this error (on dmesg) when I launch my script for eth1:
KERNEL: assertion (cl cl-un.leaf.q-q.qlen) failed at sch_htb.c(959)
Why ?
Ok, try please 2.4.21. Also can you trigger the bug
often or only sometimes ? Does it depend on traffic ?
I'm looking how to reproduce it here ..
On Thu, 3 Jul 2003 [EMAIL PROTECTED] wrote:
OK No problem!
I'm useing Slack 9.0/2.4.20( AMD K6 2-333Mhz with RAM 128MB) with HTB from
kernel...and
you should upgrade to 2.4.20. There was some problems with
TC API changes between 2.4.18-2.4.20.
The problem you see is with timer value overflow.
devik
On Sat, 22 Mar 2003, Damian Halloran wrote:
Am having a problem with HTB. The following message is generated twice
every 4 minutes in
that doing it all on the client would be better, but for a distributed
viewing, a webpage makes a nice interface.
Still, rrdtool is not made for real time graphing. A 10 sec interval is not
real-time. For me, real-time is now. And for that you need something that
can graph the files
BUT:
tc filter add dev eth1 protocol ip parent 10:1 prio 1 handle 100 fw
classid 10:100
OUTPUT : RTNETLINK answers: Invaid argument
cls_fw is not loaded ? (quick guess only).
devik
___
LARTC mailing list / [EMAIL PROTECTED]
try SFQ.
devik
On Mon, 9 Sep 2002, Michael O'Shea wrote:
Hi,
How do I allocate a guaranteed rate to each individual connection.
For example I can allocate, say 1Mb, to all HTTP traffic by defining the
appropriate class filter. In this case, if I have one connection it will
get all 1M; if
Hi yuxiao,
On Mon, 9 Sep 2002, yuxiao wrote:
Hello Alex
Sorry for my stubborn question. I read the S.Floyd paer Link-sharing and
resource mamagement On page 5, they said The estimator estimates the
bandwidth used byeach class over the appropriate time interval, to determine
whether or
heh you was a bit faster Alexey ;-)
On Mon, 9 Sep 2002 [EMAIL PROTECTED] wrote:
Hello!
Sorry for my stubborn question. I read the S.Floyd paer Link-sharing and
resource mamagement On page 5, they said The estimator estimates the
bandwidth used byeach class over the appropriate time
include pfifo or bfifo as leaf and set limit on it.
devik
On Tue, 27 Aug 2002, Luigi Romano wrote:
Hi all,
can I buffer traffic flow with HTB (like with TBF)?
HTB hasn't the limit parameter that represent the buffer size in tbf.
there is another manner to buffering traffic with HTB?
Thank
[EMAIL PROTECTED] wrote:
- Does tcng support HTB ? syntax ?
Not yet. Someone's working on implementing support for it, so I
hope to have it in the not too distant future.
Jacob have it done. He sent me some results for checking
and althought I have no time to test the latest one it
seems
it is the same as PRIO queue in linux tc.
devik
On Tue, 9 Jul 2002, bert hubert wrote:
On Tue, Jul 09, 2002 at 12:02:40PM +0800, Patrick Chan wrote:
There is priority queueing in Cisco router.
Is there any equivalent implementation for TC on Linux?
If yes, how can I configure and can
sfq MUST be leaf because it is classless.
devik
On Mon, 8 Jul 2002, Joshua Snyder wrote:
I have read most of the documents regarding both Htb and Sfq and I
still am not sure if it is best to put the Sfq on the leaf Htb or to put
it closer to the root class. All of the docs for Htb say to
When I try to attach filters to it i get the following answer:
# tc filter add dev eth1 protocol ip parent 1:0 prio 1 u32 match ip
dport 80 0x flowid 1:10
RTNETLINK answers: Invalid argument
What am I doing wrong?
did you load/compile U32 classifier ?
1. why do I have lended for children classes almost the same as
sent, who do they lend to? the neighboring classes?
Yes, to other classes.
lended is # of packets donated by this class (from its rate)
I did some experimenting, and had a root class and just 1 child class,
and this child
I change it regulary. I didn't experienced any problems - only
lockups on slow systems which can't handle it.
There should be no problem regarding the change.
devik
On Mon, 24 Jun 2002, Alexey Talikov wrote:
I can't answer you completely ( this question for kernel maintainers ) :(
But as I
I'm currently working on 2.5 inclusion. Marcello will not probably take it
without approval by QoS maintainers (Werner and jamal).
devik
On 18 Jun 2002, john huttley wrote:
Devik, Can you send Htb to Marcello Please?
Its only a module, but its about time it went mainstream.
Maybe its too
Wow. It is very interesting. Did you tried to read counters
in iptables -vL and compare counts ? Like to read value from
/proc/net/dev compare to count of packets at INPUT chain and
then compare with no of packets in DROP chains.
It could give us better picture where are the packets going to.
What is the problem ? use cat ethloop_script|./ethloop
for example ..
devik
On Thu, 6 Jun 2002, King Yung Tong wrote:
Dear all,
I copy a Ethloop script for tc script, but no response???
Does anyone know how to use ethloop?
Ethloop;
#1st parameter -- time in milliseconds from program
No, i do not think like this. If we think just for TCP connections
we may end up with ideas like changig window size. TCP tunes to send this
way but it also tries to send more each 5 seconds, if it have data.
please not this. :-) Packetizer sw goes this way and it is
- not allowed by
that quite possible ... The only way to equalize bandwidth fairly in
these scenarios still seems to be to implement the hierarchial approach
of hashing against destination IP (the user receiving the packets) and
exactly. The discussion should be about how to implement if
efficiently. What
or by proportional
to rate in each class.
Pat
On Fri, 7 Jun 2002, Martin Devera wrote:
Ok. It seems you generate only 40kbps by ethloop script. Then you
can't expect your lines to go higher !
Generate 100kbps to 1:11 and 1:10 ...
devik
On Fri, 7 Jun 2002, King Yung Tong wrote
, Martin Devera wrote:
ethloop output (as text)
On Fri, 7 Jun 2002, King Yung Tong wrote:
Pat
On Fri, 7 Jun 2002, Martin Devera wrote:
Ok. It seems you generate only 40kbps by ethloop script. Then you
can't expect your lines to go higher !
Generate 100kbps to 1:11
Did you read HTB user manual ?
On Thu, 6 Jun 2002, [gb2312] hanhbnetfilter wrote:
tc qdisc add dev eth0 root handle 1: htb default 12
tc class add dev eth0 parent 1: classid 1:10 rate
30kbps ceil 100kbps burst 2k
tc class add dev eth0 parent 1: classid 1:11 rate
10kbps ceil 100kbps burst
you set both directions to use the same htb. Am I right ? Which
kind of line, 11kBbs .. what is it ??
On 6 Jun 2002, josh wrote:
hi there!
For weeks I am trying to set my network in such a way:
server (linux) is connected through ppp0 to internet. There is MLDonkey running on
that and it
No. It would require tcp-qos cooperation.
On Thu, 6 Jun 2002, Mancinelli Giovanni wrote:
Hello all,
I have some doubts about HTB algorithm:
Is it possible to deny client connection when bandwidth is saturated ?
I mean bandwidth manager manages 1Mbit and link is full, can I stop new
And also if you feel to be able to write new qdisc then
I'd expect that you are able to find places to rewrite quickly.
As Stef suggested, in HTB patch you will see all modified
files.
devik
On Wed, 5 Jun 2002, Stef Coene wrote:
On Wednesday 05 June 2002 12:35, alouini khalif wrote:
I want
ESFQ !? Did I missed something ? :-)
devik
On Wed, 5 Jun 2002, Alexey Talikov wrote:
use latest version iproute2-2.4.7-now-ss020116-try
it work without problem (for me)
esfq is more simple in configuration that wrr
but in wrr more features for traffic control
and its use another principle
OR you can use binary in tar - the iproute compilation
is not needed then.
devik
On Wed, 5 Jun 2002, Stef Coene wrote:
On Wednesday 05 June 2002 20:33, Chris K Ellsworth wrote:
when i go to install htb i know i need to patch the kernel (DONE)
and do i need grab the iprotue2+tc and patch
Use independent root. I.e. create 1:40 with 1:0 as root not 1:2.
Roots of hierarchy are isolated each against other.
And like someone other wrote, use ceil to bound both 1:2 and 1:40
to appropriate values.
devik
My problem is One faculty (Class 1:40) don`t want to borrow and lend
bandwith to
Probably the more cheap way is to simply increase no of buckets.
With adaptive bin number you will probably end up with linear
hashing which is still fast but increasing no of bucket and keep
single level search will probably be yet faster.
Just my opinion ...
devik
On Tue, 4 Jun 2002, Michael
This is often discussed and is on TODO for someone ;)
SFQ is connection oriented. right?
Would be a good idea to make the queues per ip rather than per tcp flow?
So there would be per host fairnes.
Regards.
--
... ___ ...
| /|
only note, as of HTB 3.6 it detects non-work conserving discipline
as leaf and should work ok (only will eat more cpu).
devik
On Thu, 30 May 2002, Stef Coene wrote:
On Thursday 30 May 2002 22:33, Leandro Stasi wrote:
If I create a cbq qdisc as a leaf of a cbq class, the bandwith rate is
Ehhh sorry , I wanted to ask: do you know .. ?
devik
On Wed, 29 May 2002, [ISO-8859-1] Julin Muoz wrote:
Yes, but will we understand ?? :-)
On Wed, 29 May 2002, Martin Devera wrote:
So, the --set-mark 0x10020 trick is not working in this case ??
Yes
So, the --set-mark 0x10020 trick is not working in this case ??
Yes, it is the conclusion. With tc filter the packets go to the desired
queue, Devik's trick doesn't work in this case :-)
I'm interested here. Do you want why didn't it work ?
devik
Everytime when ceil is computed then cburst is used and vice versa ..
I will try to answer to myself:
-
| |
| |
-
Say you have a burst like this. When transmiting all the available
bandwitdh is
Seems there is no imq.o file. I'm not sure where should it be, Patrick ?
Did you enabled IMQ in make menuconfig ?
devik
On Mon, 27 May 2002, Arindam Haldar wrote:
hi all,
i want to try htb3 with imq
downloaded patched imq htb3 with kernel-2.4.18 on rh7.2 but got
STUCK on step 1 :(
this
in
netfilter !
what did i miss ?
Ciprian Niculescu wrote:
Original Message
Subject: Re: [LARTC] htb3 with imq
Date: Mon, 27 May 2002 17:44:26 +0200
From: Ciprian Niculescu [EMAIL PROTECTED]
To: Martin Devera [EMAIL PROTECTED]
References: [EMAIL PROTECTED]
[EMAIL
Probably you will have to do closer look (tcpdump
and tcptrace).
From my experiance I'd not believe CBQ much but without
closer expertise it is hard to say what's goin' on.
Also you tried to add IP adresses as both SRC and DST
which is nonsense to do on single interface. It seems
like if you
and 5, and got
much better results.
- Jason
-Original Message-
From: Martin Devera [mailto:[EMAIL PROTECTED]]
Sent: Friday, May 24, 2002 4:27 AM
To: Jason Hunter
Cc: [EMAIL PROTECTED]
Subject: Re: [LARTC] tc causes errors on HTML downloads
Probably you will have
...yeah...578 and 868 and I'm looking for 600 and 900 :-)
How accurate would you normally expect HTB to be ??
-Original Message-
From: Martin Devera [mailto:[EMAIL PROTECTED]]
Sent: 24 May 2002 18:47
To: Justin Owens
Cc: lartc
Subject: Re: [LARTC] Some help please !
Ehh and another
in typical scenarios you can use default values. Higher
one will result in higher burstiness but also to slightly
faster computation (for 1Gbit links :).
It is here probably becaus older htb didn't compute defaults.
devik
On Thu, 23 May 2002, Alexandre M. Mello wrote:
why the exemple in the
Hi,
as I promised in LARTC list I created patch for new connbytes
match. It matches packets which bellongs to connection which
transfered given range of bytes.
For those interested it is at
http://luxik.cdi.cz/~devik/connbytes/
It was originaly meant to be used with HTB or other qdisc to
I've seen patch to squid somewhere. The trick is to encode original
requester information into new request packets. It can go into
priority field - it is 32bit and is userspace-settable.
I'm currently thinking about doing something like it but have no
time ...
On Wed, 22 May 2002, Gavin White
Hi,
I sent this during LARTC outage so that I'm resending it.
It is off-topic but I feel here like at home so that
I'd like to share my experiences with my latest interesting
project.
I created nice HW toy with interesting results so that if
you want break from networking feel free to look at
On Wednesday 22 May 2002 10:43, you wrote:
I've seen patch to squid somewhere. The trick is to encode original
requester information into new request packets. It can go into
priority field - it is 32bit and is userspace-settable.
That sounds exactly like the solution I've been looking
the linux box uses redhat 7.2 with 2.4.18 kernel patched with htb 2 + bridge
firewalling, i have 60 htb classes (from 64 to 512 kbit)each with sqf leaf and I am
using U32 filter to select source, destination or port.I also have some firewalling
iptables rules. I tested the setup for 15 days
Hello,
what HTB version are you using ?
On Mon, 20 May 2002, Muhammed Soyer wrote:
Hi,
I am tryin to shape my routers traffic ..I have read lartc document .HTB is
suggested there .So I used htb ..My kernel is 2.4.18 .I patched my kernel and
activated the supports ..
I have created my
Yes .. try 3.4 - it should fix it. Let me know.
Hint: current patch can be reverted by patch -p1 -R .
devik
On Mon, 20 May 2002, Muhammed Soyer wrote:
The patch files name was htb3.3-020512.tgz (means 3.3 ? )
- Original Message -
From: Martin Devera [EMAIL PROTECTED
It is documented in HTB manual. The complaint was not displayed
before and sharing was silently wrong.
Quantum of leaf should be between 1500 and 6 and it is computed
as rate/r2q (or can be supplied independly).
So that if your maximal leaf rate is 1Mbit (12Bps) then r2q
should be 3. If
Hi,
I just finished connbytes match. You can use it to move
long downloads into lower prio class. It is for hackers
only just now.
Needed files are at luxik.cdi.cz/~devik/qos/connbytes.tgz.
IT IS NOT PATCH YET !
I'm going to sleep. More later ..
devik
HTB 3.4 is at http://luxik.cdi.cz/~devik/qos/htb/.
The only change is incorporation of patch to the latest
bug I find in linux kernel headers ..
devik
___
LARTC mailing list / [EMAIL PROTECTED]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO:
You have to enable QoS and HTB in make menuconfig. If you
then set HTB up as module you have to insmod it of course.
The error bellow seems like if you have no htb compiled
or loaded into kernel.
devik
On Fri, 17 May 2002, Justin Owens wrote:
Hi,
Could someone send me a dump of what options
Hi All,
in htb 3.3 I turned on more debugs for bad timestamps. As result
we found another bug in all 2.4 (and probably 2.2) kernels.
Because the bug is from today I attached patch. It is no harm
to apply it in every case.
It will get rid of these:
May 14 09:31:00 kernel: HTB: bad diff in charge,
:) You can drop recreate it safely. User will
not see it. The change function for SFQ will be
probably considered redundant by other kernel
developers because of it.
But you loose counters! This is bad IMHO.
That's good point ! So that go ahead and implement it for
SFQ :)
:) You can drop recreate it safely. User will
not see it. The change function for SFQ will be
probably considered redundant by other kernel
developers because of it.
On Tue, 14 May 2002, Catalin BOIE wrote:
OK.
But for SFQ?
what do you want to change in sfq !!??
perturb and quantum
I find a line in tbf_change() in sch_tbf.c: ptab = xchg(q-P_tab, ptab);
Can I consider that the pointers are exchanged between q-P_tab and ptab? as
following:
prev=q-R_tab;
q-R_tab = rtab;
rtab = prev;
Is my understanding righ?
Yes
OK.
But for SFQ?
what do you want to change in sfq !!??
devik
___
LARTC mailing list / [EMAIL PROTECTED]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
tc -s qdisc|class will give you all stats qdisc is maintaining.
For HTB it also does include actual rate (however I'm not sure
whether this part will survive merge into 2.4 kernel...).
If you want more infor you have to use user-space daemon which
will compute them.
What are you missing ?
devik
I suppose SFQ could still do something like above even if HTB were
dropping expired packets at dequeue. And this would save all qdiscs
the work of dropping expired packets at dequeue, which is still needed
even if you also check at enqueue. Also, only dropping at dequeue is
a good
I have an SFQ with max queue size 128 and very low rate, say about
1 packet/sec that I use to limit the rate of SYN's.
Now as part of a test I send a syn flood, say 200 packets in one
second. In that first second SFQ drops 200-128 but the time limit
won't drop any more on enqueue. Now
Q1: -on phys, iface it will work but will be hard to filter.
-i've been told by one user that it works with virtual vlan ifaces
but because you want to share between these interfaces you will want
to use physical one and use iptables to define classes (or u32 filter
and match vlan-id
Q1: -on phys. iface it will work but will be hard to filter.
. 1 VLAN ~ 1 IP segment , in this case I can filter according to dest.
IP addres on internal iface (I hope)...
the problem will be in that that physical interface's qdisc will see
packet with vlanid added (probably).
packets at dequeue time only in clasfull qdisc. It also makes better
sense because qdosc can also instruct SFQ to drop packet - this drop
I don't understand. What other qdisc instructs SFQ to drop?
CBQ only at the moment. It can call q-ops-drop on inner
qdisc as response to overlimit
Hi,
only a few notes on the theme. You are right with the displacement
and bad enqueue byte counters. Maybe it would be better to cound
packets at dequeue time only in clasfull qdisc. It also makes better
sense because qdosc can also instruct SFQ to drop packet - this drop
is then not decremented
The first line only mark packet. After it leaves iptables it
will enter IMQ because of the mark.
devik
On Sun, 5 May 2002, Ciprian Niculescu wrote:
whitch is the logic in this peace of code
iptables -t mangle -A PREROUTING -j IMQ
iptables -t mangle -A PREROUTING -s 1.2.3.4 -j MARK
It how it is supposed to work IMHO :)
On Sun, 5 May 2002, Ciprian Niculescu wrote:
i want to do:
iptables -t mangle -A PREROUTING -j MARK
multiples marks with different mark
so the ideea is too mark the trafic in the PRE, and POST ROUTING table
(this is the place where i have
And reason ? dep is needed only once on fresh source tree. Durin
htb developement I use make _mod_net which recompiles only
network modules and then insmod net/sched/sch_htb.o to test it.
For quick compile test you can make net/sched/sch_htb.o (but
it will not load as module then).
devik
On Tue,
And it doesn't make sense either. Why did you want to do it Stasi ?
devik
On Fri, 26 Apr 2002, Stef Coene wrote:
On Friday 26 April 2002 01:10, Stasi wrote:
Can I use the same SFQ queue dicipline for all the classe in a CBQ or
HBT system?
No.
If you add a SFQ, you can specify a handle.
The new code is private at this time - it can't be compiled - in
reality I;m working on it 4 weeks and I never tried to compile it ;)
I'll create web page for it at luxik.cdi.cz/~devik/qos/ in hour
or so ..
devik
On Fri, 26 Apr 2002, Pavel Mores wrote:
know. The main problem lies in my time
If you read the manual, the algorithm will not work correctly
with {,c}burst MTU ...
devik
On Fri, 26 Apr 2002, Anton Yurchenko wrote:
Hello,
for example I have a parent class 1:1 defined as:
tc qdisc add dev eth0 root handle 1: htb default 12
# shape at 15 kbit almost no borrowing and
I just tried to change {,c}burst to 1600, or leaving them by default but
no visible result.
here is the latest tc -s -d class show dev eth0
class htb 1:101 parent 1:1 prio 0 rate 40Kbit ceil 40Kbit burst 1599b/8
mpu 0b cburst 1599b/8 mpu 0b quantum 512 level 0 Sent 392526 bytes 5898
pkts
ceil 1kbit !!??
Wouldn`t it prevent 1:1 from borrowing from 1: ? I think I`m
it will but also it will prevent 1:1 to lend anything.
missunderstanding the ceil. As I see it it is if I fe. have rate 100kbit
and ceil 100kbit then 1:1 would go at at least 100kbit and can borrow
100kbit from
;) Try Patrick's IMQ version it should be able to do it.
On Fri, 26 Apr 2002, Anton Yurchenko wrote:
ok I got it :)) that is what I suspected :)) finaly. that is common
logic, I`m sorry, after reading some of the docs you go mad. my bad,
thank you and Stef Coene very much for you help.
We are using Ipchains. I guess that makes no difference. At the moment we do the
following:
For Example (all rules for the same device):
1. filter ACKs by using u32
2. filter a specific IP by using ipchains -m (because we masquarade) with its
own tc fw
3. filter ToS by using u32
4.
Very interesting devik, as soon as I have some time, I will begin
graphing and studying the behaviour of HTB.
I will find useful if you have some notes in gathering data, and
graphing with gnuplot. ( I think you used gnuplot in your page ).
I'm planning to write some page on this theme ..
At first look all seems good. What kind of traffic are tyou sending ?
Can you do quick tcpdump -ni eth1 and sent me a sample ? It seems like
is your packets has some other destinations or what ..
Also include tc filter show dev eth1 please.
devik
On Mon, 22 Apr 2002 [EMAIL PROTECTED] wrote:
and set classid directly in iptables like:
iptables -t mangle -A to-dsl -p tcp --dport 80 -j MARK --set-mark 0x10010
iptables -t mangle -A to-dsl -p tcp --sport 24 -j MARK --set-mark 0x10020
Oh an and I forgot to ask: Is there any other improvement exept for having a
shorter script?
It
Getting in ballance should be very quick. The decision is made
ASAP the hi prio packet is ready. Don't forget to set lower prio
number for your Citrix class.
devik
Thanks for your help,
it's working with 110k for entire bandwidth and fluent downloading via ftp.
In real environment, one of
Hi,
I'm happy that HTB gained so much popularity ;) Only
one hint for you - you can completely avoid all these
tc filter add fw ...
You can use only one
tc filter add dev ppp0 parent 1: protocol ip prio 1 handle 1 fw
and set classid directly in iptables like:
iptables -t mangle -A to-dsl -p
The other thing that is current sub-optimal is the division of long uploads
vs short uploads. I've attempted to ensure that normal webpages are
downloaded as fast as possible with the burst parameter but if someone is
downloading a large file from my website all other web users suffer. I've
it is not possible to attach qdisc to inner node. If you
do it it is deleted when you create 10:5 ..
On Wed, 17 Apr 2002, Nils Lichtenfeld wrote:
Hi there!
If I use a structure (with HTB) like this one:
10:0
| \
10:1 10:2
|
10:3
| \
10:5 10:6
Sould I only use sfq on
Otherwise iptables will do the whole to-dsl list for every packet. In
your case ot wouldn't matter except for some extra CPU usage. But if you
would like to mark port 80 as bulk-traffic and ACK's as interactive
traffic, then those port 80 ACK's could be marked as bulk which you
wouldn't
Hi it seems as if there is something other throttling
the data - is not it the case that you tested the downloads
thru your 128k internet connection ? If yes then throttling is
done at other side of line. The try to set BAND to 110k for
example ...
devik
On Sun, 14 Apr 2002, Lumir Unzeitig
Im playing with HTB and police filters. It seems that the drop action
doesnt work with htb (I think it sends the traffic to the default
configured class).
ohh yes you are probably rigght. I'll try to make it better in
the nest release.
devik
___
Thanks for the answer.
Now correct me if I got it right, what you said it is a tc bug? Because it
looks for the filters like you said ? I mean the kernel doesnt have all
those filters, right?
well it is really bug in kernel in show part. But yes it is
only cosmetic bug - kernel have only one
nice to see another active developer in field !
Let Power be with you Patrick ;)
:)
I thought about changing qos subsystem to support real ingress queueing
without imq or something. Kind of two root qdiscs per device. What do
you
(and others) think about this ?
There already is
So, is it a bug, or a misconfiguration ?
Sorry about that, it IS a bug :(
It has been fixed and i will put out a new version tonight.
The problem was that packets passed through non-work-conserving qdiscs
were delivered to dev_queue_xmit_nit twice and somewhere there a NULL
pointer
don;tknow other place. Try to download
with wget or lynx - netscappe | IE tries to be too
smart sometimes.
What says file tc.gz about downloaded file ?
What is the size of the file ?
devik
On Sun, 7 Apr 2002, Omer suat wrote:
Hi,
I am new to tc, please forgive me for simple question.
I
- Original Message -
From: Martin Devera [EMAIL PROTECTED]
To: Stef Coene [EMAIL PROTECTED]
Cc: Gavin [EMAIL PROTECTED]; [EMAIL PROTECTED]
Sent: Friday, April 05, 2002 1:40 PM
Subject: Re: [LARTC] htb ceiling
tc qdisc add dev eth1 root handle 1 htb default 10 r2q 100
tc class add dev
the rate is really measured - it is used as source for DRR weight
(here it is used relative to other classes) and to see whether we
need to borrow from parent - here absolute value matters.
So that the example above should work and give 512k.
So each class gets it's rate if it can by
Yes. Patch SFQ to use only IP in hash and no port :)
On Sat, 23 Mar 2002, Luiz Felipe Ceglia wrote:
Hi Folks,
Is there a way to deal with download accelerator programs, such as
Download Accelerator Plus?
--
Luiz Felipe Ceglia- Staff TereNet
[EMAIL PROTECTED]-
all are used everytime.
you have to use filters to direct unneeded packets
into fake queue (X:0 in case on HTB, some hiprio
band in case of prio ...).
devik
On Fri, 22 Mar 2002, Nils Lichtenfeld wrote:
Hello ML!
How to mark a device not to be used with IMQ?
And how can I get a list of
1 - 100 of 102 matches
Mail list logo