Re: [LARTC] HTB - problem with one thread
On Wednesday 28 August 2002 20:45, Stef Coene wrote: The problem is that clients can have their bandwidths at maximum only if they use a download accelerating software like DAP with multiple symultanous threads. Using only one thread, a single download can reach about 60% of the allocated bandwidth. Quantum is used if 2 or more classes with the same parent are fighting for bandwidth. First, they are allowed to get the rate you gave them. For the remaining bandwidth, quantum is used. Each class may send quantum bytes. You have to make sure quantum MTU (typical 1500 bytes for ethernet). What if you use no sfq qdiscs? Stef I have tried with pfifo queues and it's the same. I made many tests changing quantum, queueing for leaf but in no case I could reach the allocated bandwidth with a single download thread. The same limit set on cisco router with traffic-shape, worked fine. Could there be a problem when there are many qdiscs ? Are there many calculations to be done, so the packets get delayed and the TCP transfer rate is lowered. Ciprian ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] htb guarantee bandwidth
hi, you give to all the clients all the bandwith or limited at 64k, but with different priorities. class parent 1: classid 1:5 htb rate 128kbit class parent 1:5 classid 1:1 htb rate 64kbit ceil 128kbit prio 1 class parent 1:5 classid 1:2 htb rate 64kbit ceil 128kbit prio 2 you put the client that you want to have guaranted 64kbps in 1:1, and the other two in 1:2. it's the clasic example from the devik page. C Rohan Almeida wrote: Hi People, I am really losing sleep over concept of bandwidth guarantee using htb. Let me say that i have tried cbq and was quite satisfied. A simple example. I have a link of 128 Kbps, and 3 clients. Now i give each of them a bandwidth restriction of 64 Kbps, and one out of them i want guaranteed bandwidth. Now with cbq, i create an isolated class for the guaranteed guy, and i observerd that regardless of the other 2 machines being online, He always gets his allocated 64 Kbps. For htb, since there is no keyword isolated, i created 3 root classes. Also in htb, its documented that root classes do not share from each other. In other words, can i assume that they are isolated? I tried the same scenario as above using htb, but unfortunately the guaranteed machine was not receiving the allocated bandwidth. Has neone doen nething like this? Thanx -- arc_of_descent ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/ --- Xnet scaneaza automat toate mesajele impotriva virusilor folosind RAV AntiVirus. Xnet automatically scans all messages for viruses using RAV AntiVirus. Nota: RAV AntiVirus poate sa nu detecteze toti virusii noi sau toate variantele lor. Va rugam sa luati in considerare ca exista un risc de fiecare data cand deschideti fisiere atasate si ca MobiFon nu este responsabila pentru nici un prejudiciu cauzat de virusi. Disclaimer: RAV AntiVirus may not be able to detect all new viruses and variants. Please be aware that there is a risk involved whenever opening e-mail attachments to your computer and that MobiFon is not responsible for any damages caused by viruses. ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
RES: [LARTC] Help with tc TBF filter
Hi, I just want to know if HTB is alredy in stock kernel or if I have do recompile the kernel for it to work. Thanks. Roberto Campos ___ Meu Provedor Tecnologias e Informatica ltda. Rua Camerino, 128 Gr. 302 - Centro Rio de Janeiro - RJ - CEP 20080-010 Tel.: 21 - 25181011 Fax: 21 - 25181911 ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] how to force iproute2 to FORWARD to another interfacepackets with a destinatioin IP on this machine?
Jan Macek wrote: I have this situation: My machine debina, has three interfaces: eth1 212.126.24.129 ppp0 10.2.0.1 (Point-to-Point: 212.31.242.98) nsc5 10.2.0.250 (Point-to-Point: 172.23.140.32) I want packets which come in through the nsc5 interface, to be FORWARDED to the ppp0 interface to 212.31.242.98, even when their destination address is 212.126.24.129 (even so this is the IP of eth1 on this machine). How to achieve that? you can use set of 'ip rule' and other routing table ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: RES: [LARTC] Help with tc TBF filter
On Thursday 29 August 2002 13:03, Roberto Campos wrote: Hi, I just want to know if HTB is alredy in stock kernel or if I have do recompile the kernel for it to work. It will be in kernel 2.4.20. Stef -- [EMAIL PROTECTED] Using Linux as bandwidth manager http://www.docum.org/ #lartc @ irc.oftc.net ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
[LARTC] masq vs snat
Hi all Is snat /x - 1IP equal to masquerade? If I want specify address for masquerade, could I use snat for this? -- Best regard, Alexander Trotsai aka MAGE-RIPE aka MAGE-UANIC My PGP at ftp://blackhole.adamant.net/pgp/trotsai.key[.asc] ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] how to log pacets which hit routing rules?
Jan Macek wrote: Hi, Is there some way to make iproute2 log to syslog or to a file selected packets? I have a problem, that my packets dissapear somewhere, and I want to be sure if they get to the routing stage or not, and where do they get routed. use the iptables or ipchains with --log switch ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
[LARTC] HTB: messages in my log
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hi! I'm getting following messages in my log, don't know what I'm doing wrong. I have that messages on 2.419 and 20pre1 first call of my TC script, after the boot Aug 29 14:30:06 whitestar kernel: HTB init, kernel part version 3.6 Aug 29 14:30:06 whitestar kernel: HTB: quantum of class 10001 is big. Consider r2q change.4HTB: quantum of class 10010 is big. Consider r2q change.6HTB init, kernel part version 3.6 second call Aug 29 15:35:25 whitestar kernel: HTB: quantum of class 10001 is big. Consider r2q change.4HTB: quantum of class 10010 is big. Consider r2q change.7htb*g j=1476817 Aug 29 15:35:25 whitestar kernel: HTB init, kernel part version 3.6 Aug 29 15:35:25 whitestar kernel: HTB: quantum of class 10001 is big. Consider r2q change.4HTB: quantum of class 10010 is big. Consider r2q change.6HTB init, kernel part version 3.6 here is my script #!/bin/bash # Written by Robert Penz ([EMAIL PROTECTED]) # Script is under GPL # Thx for the help goes to the Linux Advanced Routing Traffic Control HOWTO # at http://lartc.org/HOWTO//cvs/2.4routing/lartc.html # and http://www.docum.org/ and http://luxik.cdi.cz/~patrick/imq/index.html # = # = # # all in Mbit # $1 = max bandwidth down # $2 = max bandwidth up # first check the parameter [ -z $1 ] echo parameter 1 missing exit 1 [ -z $2 ] echo parameter 2 missing exit 1 if [ -n `tc -s qdisc ls dev imq0 | grep htb` ] ; then tc qdisc del dev imq0 root handle 1:0 tc qdisc del dev eth0 root handle 1:0 echo old stuff killed fi # start with the download stuff # we use htb tc qdisc add dev eth0 root handle 1: htb default 10 # 100mbit nic tc class add dev eth0 parent 1: classid 1:1 htb rate 100mbit # 80mbit is just a fake value, the ceil value is more important tc class add dev eth0 parent 1:1 classid 1:10 htb rate 80mbit ceil 100mbit # traffic shaping to $1 mbit tc class add dev eth0 parent 1:1 classid 1:11 htb rate $1mbit # we use sfq for all tc qdisc add dev eth0 parent 1:10 handle 30:0 sfq tc qdisc add dev eth0 parent 1:11 handle 40:0 sfq ## now call the filters # put the capped marked stuff into that chain, uncapped is default so we don't # need to do anythink # mark with -j MARK --set-mark 1 tc filter add dev eth0 protocol ip pref 1 parent 1: handle 1 fw classid 1:11 ## continue with the upload stuff # we use HTB # all traffic to this device is to the ip 141 .. put it by default into the TC tc qdisc add dev imq0 handle 1: root htb default 10 # 100mbit virtual nic tc class add dev imq0 parent 1: classid 1:1 htb rate 100mbit # 80mbit is just a fake value, the ceil value is more important tc class add dev imq0 parent 1:1 classid 1:10 htb rate 80mbit ceil 100mbit # traffic shaping to $2 tc class add dev imq0 parent 1:1 classid 1:11 htb rate $2Mbit # we use sfq for both tc qdisc add dev imq0 parent 1:10 handle 30:0 sfq tc qdisc add dev imq0 parent 1:11 handle 40:0 sfq ## now call the filters # put the capped marked stuff into that chain, uncapped is default so we don't # need to do anythink # mark with -j MARK --set-mark 2 tc filter add dev imq0 protocol ip pref 2 parent 1: handle 2 fw classid 1:11 ## bring up the imq virtual nic ip link set imq0 up - -- Regards, Robert - Robert Penz [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE9biO48tTsQqJDUBMRAgxKAJwKbnPrqDbfl4Il6OGXyQc0CGkGHgCgl16J FxcZjs4+Rovn92EWiR8c2tA= =r2/R -END PGP SIGNATURE- ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] HTB: messages in my log
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 All you have to do is chaning r2q so quantum is smaller :) ok, I've done that, set r2q to 60, but still the same message, my problem is also that I don't know a class 10010 now my qdisc line looks that way tc qdisc add dev eth0 root handle 1: htb default 10 r2q 60 - -- Regards, Robert - Robert Penz [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE9bijy8tTsQqJDUBMRAq7OAKCCmYs9GrDntAXBFIq8ncdAJMjSFQCdEX1V tUq54Ojsmd4K+fdIe61nEBY= =EOE7 -END PGP SIGNATURE- ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] HTB: messages in my log
On Thursday 29 August 2002 16:00, Robert Penz wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 All you have to do is chaning r2q so quantum is smaller :) ok, I've done that, set r2q to 60, but still the same message, my problem is also that I don't know a class 10010 That's an internal number. now my qdisc line looks that way tc qdisc add dev eth0 root handle 1: htb default 10 r2q 60 The biggest quantum allowed is 6 (I think). So if you calculate all quantums, make sure none of them is bigger then 6. or smaller then 1500 Stef -- [EMAIL PROTECTED] Using Linux as bandwidth manager http://www.docum.org/ #lartc @ irc.oftc.net ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
[LARTC] Confusion about kernelconfig QoS-options
Hello there! Using Kernel 2.4.19. The modules I intend to use are SFQ queue, HTB packet scheduler, Firewall based classifier and the U32 classifier. Witch of the following options _NEED_ to be enabled? I ask, because the goal is to have the kernel as small as possible. QOS and/or fair queueing [*] QoS support [*] Rate estimator [*] Traffic policing Thank you! Greetings, Nils ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Confusion about kernelconfig QoS-options
On Thursday 29 August 2002 17:40, Nils Lichtenfeld wrote: Hello there! Using Kernel 2.4.19. The modules I intend to use are SFQ queue, HTB packet scheduler, Firewall based classifier and the U32 classifier. Witch of the following options _NEED_ to be enabled? I ask, because the goal is to have the kernel as small as possible. Then use modules and load only the modules you need. Stef -- [EMAIL PROTECTED] Using Linux as bandwidth manager http://www.docum.org/ #lartc @ irc.oftc.net ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Confusion about kernelconfig QoS-options
Hi Stef! Stef Coene schrieb: Witch of the following options _NEED_ to be enabled? I ask, because the goal is to have the kernel as small as possible. Then use modules and load only the modules you need. That was not my question. The listed items are not selectable as modules. They can only be compiled into the kernel or turned off. I list them again, they reside in the QOS and/or fair queueing section: [*] QoS support [*] Rate estimator [*] Traffic policing * is the default selection. Thank you! Greetings, Nils ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Confusion about kernelconfig QoS-options
On Thursday 29 August 2002 18:25, Nils Lichtenfeld wrote: Hi Stef! Stef Coene schrieb: Witch of the following options _NEED_ to be enabled? I ask, because the goal is to have the kernel as small as possible. Then use modules and load only the modules you need. That was not my question. The listed items are not selectable as modules. They can only be compiled into the kernel or turned off. I list them again, they reside in the QOS and/or fair queueing section: [*] QoS support Only needed if you want to implement QOS support so other devices on the network can send requestst for bandwidth. [*] Rate estimator This can be used for the Traffic policing to calculate the rate of the packets. [*] Traffic policing This is sort of tbf-in-filter. You can configure a filter to only match packets at a certain rate. You don't need these options to use SFQ, HTB, fw and/or u32. Stef -- [EMAIL PROTECTED] Using Linux as bandwidth manager http://www.docum.org/ #lartc @ irc.oftc.net ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Confusion about kernelconfig QoS-options
Hello Stef! Stef Coene schrieb: On Thursday 29 August 2002 18:25, Nils Lichtenfeld wrote: That was not my question. The listed items are not selectable as modules. They can only be compiled into the kernel or turned off. I list them again, they reside in the QOS and/or fair queueing section: [*] QoS support Only needed if you want to implement QOS support so other devices on the network can send requestst for bandwidth. [*] Rate estimator This can be used for the Traffic policing to calculate the rate of the packets. [*] Traffic policing This is sort of tbf-in-filter. You can configure a filter to only match packets at a certain rate. You don't need these options to use SFQ, HTB, fw and/or u32. Thank you very, very mutch! This is exactly the information I was hoping for. Greetings, Gundy ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
[LARTC] Old linuxbox as BwM
Hello, I'm playing with tc, htb, and other QoS features on an old pc I have. The PC runs, by now, a shrinked version of Debian Woody, with patched versions of kernel 2.4.18 and iproute in order to support HTB. What i'm wondering is if with a AMD 100Mhz (stealed from a museum ;-), 16Mbps RAM i can do bandwidth management to 20 network hosts, which connect wirelessly (not all, but mostly of these) sharing a wireless link of 800 kbps (kbit in tc grammar) and i'm simulating an external interfase of 256 kbps. The ASCII art is as follows: e0||e1 _to client | Linuxbox ||HUB ||Wireless | 800kbps devices | 100MHz16MB ||||AP | ) ) ) ) (PC,Hheld) || | |_| | `-- Wired Network Where e0 and e1 means eth0 and eth1. eth0 works at 256kbps, and eth1 might receive traffic at 800kbps as much (from Wireless Access Point) or at 10mbps from the wired network. --- is wire cable ) ) ) ) is air link (aka wireless) All this stuff is just for asking if i could use the old museum AMD 100MHz for BwM of 800 kbps average traffic. This old PC also works as testbed for mi diskless linux test (ramdisks, initrd setups, etc) so, it is possible that it will work with only 8MB of RAM if I use 8MB for the ramdisk. So, consider both cases (8 and 16 of RAM). Thanks. -- Esteban A. Maringolo Buenos Aires, Argentina signature.asc Description: This is a digitally signed message part
Re: [LARTC] how to log pacets which hit routing rules?
On Thu, Aug 29, 2002 at 03:17:55PM +0200, Wojtek wrote: | Jan Macek wrote: | Is there some way to make iproute2 log to syslog or to a file selected | packets? | | use the iptables or ipchains with --log switch iptables does not have a --log switch. You must use the '-j LOG' target instead. -James ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Old linuxbox as BwM
On Fri, Aug 30, 2002 at 02:59:54PM -0300, Esteban Maringolo wrote: Hello, I'm playing with tc, htb, and other QoS features on an old pc I have. The PC runs, by now, a shrinked version of Debian Woody, with patched versions of kernel 2.4.18 and iproute in order to support HTB. What i'm wondering is if with a AMD 100Mhz (stealed from a museum ;-), 16Mbps RAM i can do bandwidth management to 20 network hosts, which connect wirelessly (not all, but mostly of these) sharing a wireless link of 800 kbps (kbit in tc grammar) and i'm simulating an external interfase of 256 kbps. No problem. -- http://www.PowerDNS.com Versatile DNS Software Services http://www.tk the dot in .tk http://lartc.org Linux Advanced Routing Traffic Control HOWTO ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Re: [LARTC] Weird(?) HTB3 setup
I want to be able to specify actions for different classes of traffic in any of these four ways, and I'd like to use only HTB if possible: 1. No guranteed rate, No ceil 2. Guaranteed rate, No ceil 3. No guranteed rate, Ceil specified 4. Guaranteed rate, Ceil specified What do you mean with no ceil? Do you mean that the classes can send at full device speed? Then the ceil = device speed. For htb, no ceil means ceil = rate. No guaranteed rate can be simulated by creating 2 classes : root class rate = cail = 100 % class 1 = rate 1%, ceil 100% class 2 = rate 99%, ceil 100% Class 1 will have (allmost) no guaranteed bandwidth. In worst case senario, it get's only 1 % of the bandwidth. But if class 2 uses only 20%, class 1 can get's the remaining 80%. Of course you can change the ceil to match case 3. For types 2, 3 and 4 there can be several classes of each, with different rates and ceilings. 4 is ofcourse easy. 2 is also easy - just set ceil to the ceil of the parent class. But I'm not sure whether 1 can be accomplished with this: there is build-in passthru class named X:0 where X is your handle. Simply set default 0 when creating htb and all unclassified packets will go directly thru. devik Does go directly thru mean that unclassified packets are sent *before* packets belonging to a class with a guarateed rate? Or does it mean that unclassified packets get sent when there is bandwidth to spare (which is what I want) ? And what about lending? In what proportion does this passthru class lend bandwidth compared to other classes? It means that all the packets will get sended as fast as the hardware can. The packets will end up in the queue just before the device so they can eat bandwidth from other classes (and that's not what you want). Stef -- [EMAIL PROTECTED] Using Linux as bandwidth manager http://www.docum.org/ #lartc @ irc.oftc.net ___ LARTC mailing list / [EMAIL PROTECTED] http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
