PPP MLG bundle

[fprime7]

Hi

I have two T1s in a Multilink 1 group

and having issues with the bundle name not always being the same and thus causing only one T1 to be able to join the bundle

debugging -------------------------

1w5d: Se1/0:0 MLP: Link not added to bundle

1w5d: Se1/0:0 MLP: Request add link to bundle

1w5d: Se1/0:0 MLP: Adding link to bundle

1w5d: Se1/0:0 MLP: Link unable to join bundle Mu1(00601d3d8ed300), name 0060.1d3d.8ed3 doesn't match

1w5d: Se1/0:0 MLP: Link not added to bundle

1w5d: Se1/0:0 MLP: Request add link to bundle

1w5d: Se1/0:0 MLP: Adding link to bundle

1w5d: Se1/0:0 MLP: Link unable to join bundle Mu1(00601d3d8ed300), name 0060.1d3d.8ed3 doesn't match

I am running IOS 12.3 on 7204VXR

tried the

ppp multilink endpoint ip

on the T1s and on the Multilink1

still same result

Any ideas

Thanks

Frank

[Richard Burts]

Frank

It would be helpful if you would post the router configuration. Perhaps also a show version from both routers.

HTH

Rick

[ilya.varlashkin]

It's curious why you had bundle name appended with extra '00'. It should have been 00601d3d8ed3. Looks like a bug so far.

With your current config, could you post 'show ppp multilink' output? And what do you see in 'debug ppp multilink ev'?

Generally you don't even need to configure (actually change) the way endpoint descriminator is selected - default hostname would work fine in your case since there are only your routers that could join the bundle.

[fprime7]

Hi

here is the output from the commands and running config

#sh ppp multilink

Multilink1, bundle name is 00601d3d8ed300

Endpoint discriminator is 00601d3d8ed300

Bundle up for 00:05:16, 1/255 load, 1 receive classes, 4 transmit classes

Receive buffer limit 12000 bytes, frag timeout 1000 ms

Receive Class 0:

0/0 fragments/bytes in reassembly list

1 lost fragments, 1 reordered

1/30 discarded fragments/bytes, 0 lost received

0x16 received sequence

Transmit Class 0:

0x66 sent sequence

Transmit Class 1:

0x0 sent sequence

Remaining Transmit Classes unused

Member links: 1 active, 1 inactive (max 3, min not set)

Se1/1:0, since 00:05:15, 1920 weight, 1496 frag size

Se1/0:0 (inactive)

Running configuration

multilink bundle-name both

!

interface Multilink1

ip address 172.1.1.105 255.255.255.252

ip helper-address 172.16.206.60

no ip redirects

no ip route-cache cef

no ip mroute-cache

peer default ip address 172.1.1.106

no keepalive

no cdp enable

ppp ipcp dns 172.16.206.42 172.16.206.42

ppp multilink

ppp multilink fragment delay 10

ppp multilink links maximum 3

ppp multilink interleave

ppp multilink group 1

ppp multilink endpoint ip 172.1.1.106

!

!

interface Serial1/0:0

no ip address

no ip redirects

ip load-sharing per-packet

encapsulation ppp

keepalive 1

no fair-queue

ppp multilink

ppp multilink group 1

ppp timeout retry 1

!

interface Serial1/1:0

no ip address

no ip redirects

ip load-sharing per-packet

encapsulation ppp

keepalive 1

no fair-queue

ppp multilink

ppp multilink group 1

ppp timeout retry 1

!

# debug ppp multilink events

1w6d: Se1/0:0 MLP: Request add link to bundle

1w6d: Se1/0:0 MLP: Adding link to bundle

1w6d: Se1/0:0 MLP: Link unable to join bundle Mu1(00601d3d8ed300), name 0060.1d3d.8ed3 doesn't match

1w6d: Se1/0:0 MLP: Link not added to bundle

1w6d: Se1/0:0 MLP: Request add link to bundle

1w6d: Se1/0:0 MLP: Adding link to bundle

1w6d: Se1/0:0 MLP: Link unable to join bundle Mu1(00601d3d8ed300), name 0060.1d3d.8ed3 doesn't match

1w6d: Se1/0:0 MLP: Link not added to bundle

[ilya.varlashkin]

The command 'ppp multilink endpoint ip 172.1.1.106' only sets how _this_ system will identify itself to the remote system. So you'd have to type this command not on this, but on remote system.

Simply remove all occurences of 'ppp multilink endpoint' - in your scenario there's no need for it, it will work (using default method of hostname).

[fprime7]

Hi Ilya,

I removed the ppp multilink endpoint commands

one thing I notice that I can ping the remote IP on the Multilink1 say 172.1.1.106

but cannot ping the router side 172.1.1.105

?

Thanks

[ilya.varlashkin]

Without seeing complete config from both sides it's difficult to say, but you can try following things:

1) enable CEF on multilink interface

2) remove all IP-related configs from serial interface (IP should be configured only on Multilink interface, serials are L2-only interfaces in this scenario)

If this doesn't help, try also 'no peer neighbor-route' on Multilink of both sides.

If still doesn't help, please post full config from both sides and output of 'sh ip cef 172.1.1.105 detail' from both routers.

[fprime7]

I think that is how I have it setup

global command -

ip cef

how is the bundle name assigned, as the T1 join the bundle

by the source router ?

or the far end router ?

my source router

======================================

interface Multilink1

ip address 172.1.1.105 255.255.255.252

ip helper-address 172.16.206.60

no ip redirects

no ip route-cache cef

no ip mroute-cache

peer default ip address 172.1.1.106

no keepalive

no cdp enable

ppp ipcp dns 172.16.206.42 172.16.206.42

ppp multilink

ppp multilink fragment delay 10

ppp multilink links maximum 3

ppp multilink interleave

ppp multilink group 1

ppp multilink endpoint ip 172.1.1.106

!

!

interface Serial1/0:0

no ip address

no ip redirects

ip load-sharing per-packet *** should I remove this ***

encapsulation ppp

keepalive 1

no fair-queue

ppp multilink

ppp multilink group 1

ppp timeout retry 1

!

interface Serial1/1:0

no ip address

no ip redirects

ip load-sharing per-packet *** should I remove this ****

encapsulation ppp

keepalive 1

no fair-queue

ppp multilink

ppp multilink group 1

ppp timeout retry 1

!

[ilya.varlashkin]

Bundle name is either ppp username of remote side if authentication is used and 'ppp multilink bundle-name authenticated' is configured on local system, or endpoint descriminator from _remote_ system of the first link to join the bundle. By default endpoint descriminator is hostname and unless you have good reason to change it, you probably shouldn't.

I'd suggest following changes to your config:

1) remove 'ppp multilink endpoint ' command on all interfaces of both sides

2) remove 'ip load-sharing' from serial interface (they don't have IP layer anyway)

3) enable 'ip route-cache cef' on Multilink and don't use 'ip load-sharing' anywhere because it's only single link from IP prospective.

4) remove 'ip ip redirects' from serials because they're not IP-layer interfaces

5) use 'ppp keepalive' on multilink interface if you want, but remove 'keepalive' from serials

For getting things moving, you don't even need to set maximum number of links (nobody going to join anyway besides your serial links). Fragmentation and interleaving makes sense if you carry some delay-sensitive traffic over your link, otherwise you can gain performance by not using fragmentation and interleaving.

What are you trying to achieve by having helper address on Multilink interface? This would affect only broadcast packets on that link.

Please let me know if this makes your link working.

[dsouzaswetha]

Hi ,

I have a query on MLPPP

I am using a CISCO 2941 DC-A router,

For the PPP testing.I was using GL card from the GL communications.

In the case of GL card it supports this below Feature

"In MLPPP Simulation, NCP can be configured to send Packets over PPP links in the bundle or on the MLPPP bundle."

My  query Is in the Cisco Router 2941-DC-A,Is their support for sending NCP  over PPP in the case of Multilink(MLPPP).Because I found out  that in  the case of MLPPP the  Cisco router sends NCP packets  over MLPPP  only.Kindly help me how to configure the cisco router to send NCP  packets over PPP in case of MLPPP.?

Kindly Reply

Thanks in advance

Swetha Dsouza