MLPP routing issue

Unanswered Question
Feb 17th, 2009

There is something I am not understanding here. See attached for my config on my 1841. I am accessing the internet via bonded T1's using MLPPP. The ISP has assigned me the address block I have fastethernet0/0 attached to my local ( network.

When the ppp connects, for example, it is assigned from ipcp the ip of with a gateway of Now connecting to FE0/1 everything is perfect. If I connect from my local network (FE0/0)it will work for a few minutes, very slow, and then tell me destination unreachable.

What am I missing here? How do I get my local network to route over the PPP?

Here is my route table:

Gateway of last resort is to network is subnetted, 2 subnets

C is directly connected, Multilink1

C is directly connected, Multilink1 is subnetted, 1 subnets

C is directly connected, FastEthernet0/1

C is directly connected, FastEthernet0/0

S* [1/0] via



I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Edison Ortiz Tue, 02/17/2009 - 11:42

Modify the following statement:

ip nat inside source list 1 interface FastEthernet0/1 overload


ip nat inside source list 1 interface Multilink1 overload




Richard Burts Tue, 02/17/2009 - 11:57


I agree with Edison that there is a problem in the way that NAT is configured. But I do not believe that changing the NAT so that it overloads on the multilink will solve the problem. The address learned on the multilink is also RFC1918 address space and will not produce addresses that can be used in the Internet.

I believe that there is a conceptual problem here that we need to work through. The ISP assigned you a block of addresses. Those addresses should be the source address of traffic that you send to the Internet. But you assigned them to a FastEthernet interface. I believe that this is not the right thing to do. I would believe that you need to create a NAT pool with those addresses and to do the translation using the pool instead of doing NAT overload on an interface address.



Paolo Bevilacqua Tue, 02/17/2009 - 12:14

I differ with both Edison and Rick.

I believe the config is correct, multink interface uses a private address as happens more and more these days.

Please let us know the exact IOS used.

Richard Burts Tue, 02/17/2009 - 12:18


Differences of opinion make for better discussion. In the original post he specified that his network is on the FastEthernet0/0 interface and that he is routing to the Internet over the multilink. And that the ISP had assigned him a block of addresses. Given that the traffic originates on FA0/0 and is routed out the multilink, how is assigning the ISP block on FA0/1 going to work?



Paolo Bevilacqua Tue, 02/17/2009 - 12:33

Hi Rick,

It works because it Fa0/1 is not to be natted, it has a globally routable address. My understanding is that this is what the OP wants.

I correct my previous statement, Mu1 uses a public address as the OP clearly specified. On this one, overloaded NAT for FA0/0 is done

micrinservices Tue, 02/17/2009 - 14:51

The IOS ver:

Cisco IOS Software, 1841 Software (C1841-ADVIPSERVICESK9-M), Version 12.4(9)T4, RELEASE SOFTWARE (fc2)

Note: If I down FA0/1 with the public ip nothing will get routed out at all - When it is up and I do a trace from the private network, it doesnt appear to touch my public IP's. Next hop shows as strange that is.



Paolo Bevilacqua Tue, 02/17/2009 - 15:05

Hi, I would recommend you try either 12.4(9)T7, or more conservatively, 12.4(3j).

Your traceroute is normal, next-hop is ISP router.

micrinservices Wed, 02/18/2009 - 07:34

Unfortunately I do not have access to other version images. Is it your position that this is a bug and not a configuration issue?



Paolo Bevilacqua Wed, 02/18/2009 - 07:46

Yes, I've seen at least a bug like this in the past - many NAT entries caused memory leak and slowdown.

Note you cannot professionally operate cisco networks without at least one smartnet contract for downloads.

Paolo Bevilacqua Wed, 02/18/2009 - 10:10

I've looked at this further and found that you'reu sing MLPPP over frame-relay. That is not a very commonly used configuration and you may find issues in doing that.

Basically one reason more to try some different IOSlike the mentioned 12.4(3j).

micrinservices Wed, 02/18/2009 - 11:41


I tried Ricks suggestion and created a NAT pool. It is working much better. The download speed is pretty much right on but upload is very low. I am going to compare results by doing a test connection via fa0/1 and compare results.

I am working on getting smartnet up and going.




This Discussion