Skip to contentSkip to conversation

Problems with multipath routing.

LARTC Raúl Alexis Betancort Santana 2 participants 2 messages Jan 28, 2004 – Jan 29, 2004

lore
Share & tools
lore
#0 lore raw

Hi all, I have setup two multipath route tables on my system for doing failover routing, What I want it's that if GW at route1 of the MP is dead, traffic goes by route2, for doing that I have created the multipath routes as follows:

ip route add table mail.traffic proto static nexthop via ${GW1} dev eth1 weight 1 nexthop via ${GW2} dev eth1 weight 250

But it does not run as I espected, I want that most (all if posible) the traffic goes by GW1, and if it is down (DgD patches are applied) traffic must goes by GW2, the kernel it's not taking into account the weight parameter or maybe I'm doing it wrong.

Any hit will be apreciated ...

Best regards
_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

#1 lore raw

hello,
you can use the following script,
IP=/sbin/ip
${IP} route add default via ${STATICGW1} dev ${STATICIF1} src ${STATICIP1} proto static table 200
${IP} route append prohibit default table 200 metric 1 proto static ${IP} route add default via ${STATICGW2} dev ${STATICIF2} src ${STATICIP2} proto static table 200

${IP} rule add prio 200 from all lookup 200
this will first lookup the first line, i.e. gateway-1 and if it is failed, it will try for gateway-2,
it would be good if you apply julian's routing patch fromm the following location,
http://www.ssi.bg/~ja/#routes-2.4
for more info. you can refer this document too,
http://www.ssi.bg/~ja/nano.txt
regards,
Tushar Thakker

----------------------------------------------------------------
Success usually comes to those who are too busy to be looking for it
----------------------------------------------------------------
----- Original Message -----
From: <[email protected]>
To: <[email protected]>
Sent: Thursday, January 29, 2004 2:55 AM
Subject: LARTC digest, Vol 1 #1563 - 9 msgs

Quoted 15 lines

Send LARTC mailing list submissions to
[email protected]

To subscribe or unsubscribe via the World Wide Web, visit
http://mailman.ds9a.nl/mailman/listinfo/lartc
or, via email, send a message with subject or body 'help' to [email protected]

You can reach the person managing the list at
[email protected]

When replying, please edit your Subject line so it is more specific than "Re: Contents of LARTC digest..."

Quoted 7 lines

Today's Topics:

1. Re: tncg and bandwidth limiting (Martin A. Brown)
2. IProute and Traffic Control of PPC (David Bierce)
3. Re: Shaping Device Aliases (Martin A. Brown)
4. Re: IMQ Stability (Alexander Trotsai)
5. Problems with multipath routing.

(=?iso-8859-15?q?Raúl Alexis Betancort Santana?=)

Quoted 232 lines

6. small netwok traffic shaping (Adrian Coman)
7. Re: small netwok traffic shaping (Michael Renzmann)
8. Burst Rate? (Scott Baker)
9. Question(s) for the programming gurus (Michael Renzmann)

--__--__--

Message: 1
Date: Wed, 28 Jan 2004 00:18:24 -0600 (CST)
From: "Martin A. Brown" <[email protected]>
To: Scott Baker <[email protected]>
Cc: [email protected]
Subject: Re: [LARTC] tncg and bandwidth limiting

Scott,

: Basically ignore that last message.

Earlier message ignored.

: I'm trying to do some very simple rate-shaping on an interface. I want : to limit my 100baseT interface to 7 megs both ingress and egress of the : interface.

You'll notice that Rubens suggested you use a TBF. This would be perfectly adequate solution for your transmitted traffic. Note that an HTB class and a TBF qdisc are essentially performing the same function. Shaping!

Note there is a difference in the traffic control structures created by your tcng configuration. Your egress section will actually be two HTB classes inside an HTB qdisc attached to the INTERFACE in question. In your situation, you do not need both classes (created as siblings), since you are classifying everything into class $all.

: I'm curious if some of the other experts out there wouldn't have a : "better" way to do what I'm doing. I'd like to do HTB ingress as well, : but it complains that the the ingress qdisc doesn't allow inside : classes or something like that. I think this will work for me, I just : want to make sure this is the best way to do things.

This is a limitation of traffic control under Linux. You can only shape what you transmit [ see IMQ if you want to know how to break this rule ]. So, unless you are going to use IMQ, you'll not be able to shape your local input traffic (if you are a router, you should be able to slow down conversations by "artificially" delaying the packets on the internal interface).

However, you don't need to care that you are not shaping on your inbound traffic. You can police the traffic. For the difference between shaping and policing, try here [0].

[ snip ]

: htb () {
: class ( rate 100Mbps, ceil 100Mbps ) ; /* remove this */
: $all = class ( rate 7Mbps, ceil 7Mbps ) ;
: }

: ingress {
: $p = bucket(rate 7Mbps, burst 100kB, mpu 200B);
: class (1) if (conform $p && count $p) || drop;
: }

After you run your tcng config file through tcc ("tcc < $FILE | less"), you should see (lines broken for readability) the following for the ingress traffic control. I left INTERFACE in the config file--obviously you have #defined it someplace else.

  tc qdisc add dev INTERFACE ingress
  tc filter add dev INTERFACE parent ffff:0 protocol all prio 1 \
    u32 match u32 0x0 0x0 at 0 classid ffff:1 \
    police index 2 rate 875000bps burst 102400 mpu 200 action drop/pass
    ^^^^^^

Note that the policer will (somewhat harshly) accommodate your desires to limit the traffic accepted inbound on an interface.

Best of luck,

-Martin

[0] http://tldp.org/HOWTO/Traffic-Control-HOWTO/elements.html#e-shaping
http://tldp.org/HOWTO/Traffic-Control-HOWTO/elements.html#e-policing

--
Martin A. Brown --- SecurePipe, Inc. --- [email protected]

--__--__--

Message: 2
To: [email protected]
From: David Bierce <[email protected]>
Date: Wed, 28 Jan 2004 00:46:19 -0600
Subject: [LARTC] IProute and Traffic Control of PPC

Has any one ever done it? We are in the planning stages and the majority of the hardware we have available is PPC, what are some thoughts?

--__--__--

Message: 3
Date: Wed, 28 Jan 2004 00:54:39 -0600 (CST)
From: "Martin A. Brown" <[email protected]>
To: Gordan Bobic <[email protected]>
Cc: [email protected]
Subject: Re: [LARTC] Shaping Device Aliases

Gordan,

I've noticed that you are trying to use aliased IP addresses and traffic control together, and you are a bit frustrated that tc doesn't handle aliased interface names.

: > > I understand that device aliases (e.g. eth2:3) are not shapeable. : > > Does anybody know if this functionality is planned in the future? : >
: > None of the new(er) networking tools recognise device aliases, : > because on all recent linux releases, aliases don't exist. : > the ethX:X notation is a legacy notation used only by the ifconfig : > program. everything else just sees a ethX with more than one IP : > address.
: >
: > So you just run your shaping rules on the real interfaces, and : > restrict it's operation with IP address filtering.
:
: Yes, I am aware of that. However, that makes shaping multiple : independent "streams" going through one interface much more difficult.

I don't understand why this becomes much more difficult--it just becomes a little more difficult, depending on the number of IP addresses you have active on a given interface. If you can handle multiple addresses on an interface, then shaping traffic on these (known) addresses shouldn't be much more difficult than managing each address.

: The only other thing I can think of is setting up a dummy network : device and giving it the IP addresses on all the non-primary subnets : (e.g. multiple DSL lines), and setting up the arp and routing to make : the packet actually go via the primary interface.

This sounds like a very confused idea. I'm not sure it's worth the hassle--as I hope I can convince you below.

[ more stuff snipped ]

: Has anybody got any thoughts on this?

I have some thoughts, which I hope can help you understand why you will be able to use the traffic control tools to accomplish your filtering. For posterity, I'll reiterate some of what has come before.

IP aliases don't exist. This is a convention for ifconfig. "ip addr show" will display all IP addresses active on a given interface.

Traffic control is the last thing performed before turning the packet over to the device driver and hardware. Similarly, it is the first thing called on receipt of a packet. See diagrams KPTD [0] and ebtables packet flow [1].

In this case, you can use any number of techniques to identify the packets with tc tools based on their IP addresses--the convenience of the aliased interface naming is simply an obstruction of the real path the packet takes.

: If this would work, maybe it should be documented in the advanced : routing howto, as I can see how there might be a lot of people out : there who would find it useful.

Let me suggest a possibility, if we assume a nested configuration. Let's say you have IP0 and IP1 active on interface eth3 and you want to make sure that bandwidth is split 75/25 between these two and you want them to share bandwidth. Classic bandwidth-sharing situation....in the tcng config below, you'd need to #define IP0 and IP1, but then you'd have a simple configuration. If you needed to further subdivide traffic within each of the IP0 and IP1 classes, you'd have an easy way to do so.

    dev eth0 {
        egress {
            class ( <$ip0> )  if ip_src = IP0 ;
            class ( <$ip1> )  if ip_src = IP1 ;
            htb () {
                class ( rate 1544kbps, ceil 1544kbps ) {   /* T1 speed */
                    $ip0 = class ( rate 1024kbps, ceil 1544kbps ) ;
                    $ip1 = class ( rate  384kbps, ceil 1544kbps ) ;
                }
            }
        }
    }

Alternately, you may wish to simulate virtual circuits with each of the IP addresses on a machine. In this case, you could use separate root classes attached to the HTB qdisc, or another class. You can prevent the two classes from competing with each other by setting the rate and ceil to the same value. Here's a very simple permutation of the above.

    dev eth0 {
        egress {
            class ( <$ip0> )  if ip_src = IP0 ;
            class ( <$ip1> )  if ip_src = IP1 ;
            htb () {
                class ( rate 1544kbps, ceil 1544kbps ) {   /* T1 speed */
                    $ip0 = class ( rate 1024kbps, ceil 1024kbps ) ;
                    $ip1 = class ( rate  384kbps, ceil  384kbps ) ;
                }
            }
        }
    }

Best of luck, Gordan!

-Martin

[0] http://www.docum.org/stef.coene/qos/kptd/
[1] http://ebtables.sourceforge.net/br_fw_ia/PacketFlow.png

--
Martin A. Brown --- SecurePipe, Inc. --- [email protected]

--__--__--

Message: 4
Date: Wed, 28 Jan 2004 10:25:00 +0200
From: Alexander Trotsai <[email protected]>
To: "Michael S. Kazmier" <[email protected]>
Cc: [email protected]
Subject: Re: [LARTC] IMQ Stability

Quoted 4 lines

On Fri, Jan 23, 2004 at 10:29:13AM -0700, Michael S. Kazmier wrote:
MSK>Hello all,

MSK>I have been doing a lot of archive searching over the last week

reading

Quoted 2 lines

MSK>posts on IMQ and it's apparent stability / instability. I have seen a MSK>number of posts about it not being maintained as well. Can anyone

talk to

Quoted 1 line

MSK>me about IMQ's stability in a heavy throughput environment (20 Mbps)

and

Quoted 23 lines

MSK>what was causing IMQ to fail if you know.

I use it and it's work OK for me
Traffic at some router up to 30-40 Mbit

IMQ has one trouble
Don't assing address to imq interface becase kernel crash it you do this.

--
Best regard, Aleksander Trotsai aka MAGE-RIPE aka MAGE-UANIC My PGP key at ftp://blackhole.adamant.ua/pgp/trotsai.key[.asc] Big trouble - ..disk or the processor is on fire.

--__--__--

Message: 5
From: =?iso-8859-15?q?Raúl Alexis Betancort Santana?<[email protected]>
Organization: =?iso-8859-15?q?Dimensión Virtual?= S.L.
To: [email protected]
Date: Wed, 28 Jan 2004 10:01:56 +0000
Subject: [LARTC] Problems with multipath routing.

Quoted 3 lines

Hi all, I have setup two multipath route tables on my system for doing failover routing, What I want it's that if GW at route1 of the MP is dead, traffic goes by route2, for doing that I have created the multipath routes

as

Quoted 7 lines

follows:

ip route add table mail.traffic proto static nexthop via ${GW1} dev eth1 weight 1 nexthop via ${GW2} dev eth1 weight 250

But it does not run as I espected, I want that most (all if posible) the traffic goes by GW1, and if it is down (DgD patches are applied) traffic

must

Quoted 1 line

goes by GW2, the kernel it's not taking into account the weight parameter

or

Quoted 15 lines

maybe I'm doing it wrong.

Any hit will be apreciated ...

Best regards

--__--__--

Message: 6
Date: Wed, 28 Jan 2004 16:03:33 +0200
From: Adrian Coman <[email protected]>
Organization: NoBug Consulting
To: [email protected]
Subject: [LARTC] small netwok traffic shaping

Quoted 3 lines

Hi,

First of all I must say that I'm a newbie in the network adimistration

domain.

Quoted 2 lines

I have the following situation: a network composed of ~10 computers which

are connected to

Quoted 1 line

the internet through a gateway. The connection speed is 128kbps for the

addresses outside

Quoted 3 lines

my country, and 10mbps for the addresses in my country.

I would like to set-up a traffic shaper on the router machine with the

following features:

Quoted 2 lines

- if all the users are browsing/downloading in the same time, the

bandwidth must be

Quoted 2 lines

equally shared
- if the users are using file sharing software, their bandwidth for such

applications must

Quoted 1 line

be limited to maximum 10kbps if noone else is requesting bandwidth for

normal http

Quoted 2 lines

transfers, else the bandwidth must be 0 for such applications.
- i want also that some computers from the network to get priviledged

access, with no

Quoted 4 lines

restrictions

If possible i wold also like that:
- the users who are using network scanning software to have their

bandwidth cut to 0 for a

Quoted 3 lines

period

The router has 2 NIC's eth0 and eth1. eth0 is connected to the outside

world and eth1 to

Quoted 3 lines

the internal network.

Can you help me with examples? What solutions do you advise me to

implement?

Quoted 2 lines

I know I can read the manuals (as most of my friends say), but it's very

difficult for me

Quoted 56 lines

to get it right form the first time.

Thanks,
Adrian

--__--__--

Message: 7
Date: Wed, 28 Jan 2004 15:55:36 +0100
From: Michael Renzmann <[email protected]>
To: Adrian Coman <[email protected]>
Cc: [email protected]
Subject: Re: [LARTC] small netwok traffic shaping

Buna ziua, Adrian :)

Quoted 4 lines · Adrian Coman

- if the users are using file sharing software, their bandwidth for such applications must be limited to maximum 10kbps if noone else is requesting bandwidth for normal http transfers, else the bandwidth must be 0 for such applications.

There are two things that might be interesting for your work:

1. http://l7-filter.sf.net
That's a facility for either the QoS framework or iptables that enables to distinct between several application layer (iso layer 7, hence the name) protocols such as http and ftp. You could use that to apply special marks to packets which then help you to classify the packets.

2. http://rnvs.informatik.uni-leipzig.de/ipp2p/index_en.html
That's an extenstion to iptables which allows to mark connections that belong to common peer-to-peer applications (which would be helpful to apply the above quoted rule).

Unfortunately I can't give you any more pointers, as I'm myself new to the whole QoS-stuff. But I hope this will help you a little.

La revedere.
Mike

--__--__--

Message: 8
Date: Wed, 28 Jan 2004 16:03:45 -0800
To: [email protected]
From: Scott Baker <[email protected]>
Subject: [LARTC] Burst Rate?

If I'm using the following to very simply police my incoming bandwidth:

Quoted 7 lines
ingress {
   $p = bucket(rate 7Mbps, burst 1000kB, mpu 200B);
   class (1) if (conform $p && count $p) || drop;
}

How should i be calculate the burst rate? Cisco has their own special algorithm for calculating the correct burst rate, is there a similar

method

Quoted 18 lines

I should be using to calculate burst on a linux box? Also are there any improvement in the 2.6.x kernel with regards to traffic shaping?

Scott

Scott Baker - Network Engineer - RHCE
bakers @ web-ster . com - 503.266.8253

--__--__--

Message: 9
Date: Thu, 29 Jan 2004 01:06:29 +0100
From: Michael Renzmann <[email protected]>
To: "[email protected]" <[email protected]>
Subject: [LARTC] Question(s) for the programming gurus

Quoted 4 lines

Hi all.

I'm quite new to the concepts of the "traffic control" framework, and I've got a programming-related question. Hopefully someone has the

answer...

Quoted 53 lines

Is it possible, either for the device driver itself or for a userspace program, to get information about how many packets are currently queued for a given network interface?

Let's describe it a little more in detail:

I have a network interface eth0 in my linux box. Now I apply traffic shaping to that interface, for example the outgoing traffic is shaped down to 1 MBit/s. There is an application that creates packets which are meant to be sent out via eth0, and the application creates its packets with a much higher rate than 1 MBit/s. This would result in the shaper enqueuing packets for eth0 and, sooner or later, in dropping some of the packets if the queue is full.

So I want to slow down the rate at which the application creates its packets. The easiest way would be to take a look at the "traffic control" queues for eth0 and check their current state. When the queue is filled up to a specified level, the application should stop creation of new packets until the queue has been emptied. (*)

So, is there any way for my application to check the state of the eth0-queues? Or is this possible for the driver of eth0 (as I'm in control of this driver, I could implement a way to pass the needed information down to the application if necessary)?

Next question: if I understood the concepts of the "traffic control" system correctly, one could add several queues to a single device. Is there any way to simply get the total amount of packets that are waiting in all attached queues? Or would I need to check each queue and sum up the values?

And last question: what kind of information can I get about the currently enqueued packets? Just the amount of packets that are enqueued, or only the amount of enqueued bytes, or both?

I'd appreciate any kind of help very much. Pointers to existing documentation welcome - I didn't find the answers in the docs I found, but maybe I just didn't search well enough (or in the wrong places)?

Thanks in advance.

Bye, Mike

(*) In other words: I want to have the effects of slowing down the traffic generation of my application without having to care about the actual implementation of the traffic shaping. In my special case this makes sense and would save me a lot of work.

--__--__--

Quoted 8 lines

_______________________________________________
LARTC mailing list
[email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc

End of LARTC Digest

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/