route advertisment problem (OSPF over Frame)

chloi · ‎03-11-2005

Hi,

I had a problem with route advertisment for OSPF over Frame-Relay.

let say I have 2 router, A & B. A gets route advertisment from B but, B does Not get anything from A.

Here is some of my show output.

Show ip route (router A)

170.10.0.0/16 is variably subnetted, 7 subnets, 3 masks

C 170.10.129.0/30 is directly connected, BRI0

D 170.10.35.0/24 [90/2195456] via 170.10.23.2, 00:36:59, Serial1

O IA 170.10.49.0/30 [110/128] via 170.10.29.2, 00:00:42, Serial0

C 170.10.23.2/32 is directly connected, Serial1

C 170.10.22.0/24 is directly connected, Ethernet0

C 170.10.23.0/30 is directly connected, Serial1

C 170.10.29.0/30 is directly connected, Serial0

show ip route (router B)

170.10.0.0/16 is variably subnetted, 3 subnets, 2 masks

C 170.10.49.1/32 is directly connected, Serial1

C 170.10.49.0/30 is directly connected, Serial1

C 170.10.29.0/30 is directly connected, Serial0

sh ip ospf nei (router A)

Neighbor ID Pri State Dead Time Address Interface

170.10.129.2 1 FULL/DR 00:00:35 170.10.29.2 Serial0

sh ip ospf nei (Router B)

Neighbor ID Pri State Dead Time Address Interface

170.10.129.1 1 FULL/BDR 00:00:34 170.10.29.1 Serial0

170.10.49.1 0 FULL/ - 00:00:31 170.10.49.1 Serial1

sh run (Router A)

interface Serial0

ip address 170.10.29.1 255.255.255.252

encapsulation frame-relay

ip ospf network broadcast

frame-relay map ip 170.10.29.2 17 broadcast

frame-relay interface-dlci 17

router ospf 1

log-adjacency-changes

redistribute eigrp 1

network 170.10.29.1 0.0.0.0 area 0

network 170.10.129.1 0.0.0.0 area 0

network 172.10.22.0 0.0.0.255 area 0

!

sh run (Router B)

interface Serial0

ip address 170.10.29.2 255.255.255.252

encapsulation frame-relay

ip ospf network broadcast

frame-relay map ip 170.10.29.1 17 broadcast

frame-relay interface-dlci 17

router ospf 1

log-adjacency-changes

network 170.10.29.2 0.0.0.0 area 0

network 170.10.49.2 0.0.0.0 area 1

network 170.10.129.2 0.0.0.0 area 0

what would be the possible causes?

thanks

Kevin Dorrell · ‎03-11-2005

There is an error in your OSPF configuration on router A. The line:

network 172.10.22.0 0.0.0.255 area 0

should read:

network 170.10.22.0 0.0.0.255 area 0

Hope this helps.

Kevin Dorrell

Luxembourg

Richard Burts · ‎03-11-2005

Catching the typo was a good catch for Kevin. An easy way to find this kind of error is to do show ip ospf interface on each of the routers and look to see if all the interfaces that you expect to be there are actually showing up. In this case you would expect the 170.10.22.0 on Ethernet 0 to show up and it would not.

But I am not sure that this is the only problem. I suggest that we think about what each router knows and should advertise to the other.

Router A has three networks configured under OSPF. Kevin has found why the Ethernet subnet is not being advertised. One of those networks is the network that connects router A to router B. But there is a third network (170.10.129.1) under OSPF which does show up as a connected network in the show ip route. So why is that network not being advertised?

Router B has three networks configured under OSPF (170.10.29.2, 170.10.49.2, and 170.10.129.2). But only two of them show up in the show ip route table as connected interfaces. And the show ip ospf neighbor on router B shows that there is a neighbor with RID of 170.10.129.1 connected on the serial 0 interface. So we need to check on 170.10.129.2 and see what is happening. If the 170.10.129.0 subnet is actually known on router B, then the explanation for why you are not seeing advertisements from A to B is that A does not know anything to advertise to B that B does not already know about. That would change after you correct the typo and router A starts to advertise the Ethernet subnet.

HTH

Rick

HTH

Rick

Kevin Dorrell · ‎03-11-2005

I think part of the problem here is that the 170.10.129.0/30 is actually his ISDN backup link. See that Router A has it directly connected on BRI0. Now, I think that router A is 170.10.129.1 and router B is 170.10.129.2 on that ISDN link. Unfortunately, both routers seem to be using that as a RID, as seen from the neighbor tables. Which (probably?) means that the ISDN was actually up at the time the OSPF was started.

I think you would do well to define a loopback interface on each router at least to fix the RID, or configure an explicit RID. At least that would avoid some of the confusion.

http://www.cisco.com/univercd/cc/td/doc/product/software/ios123/123cgcr/iprrp_r/ip2_o1g.htm#wp1038293

I don't understand why 170.10.129.0/30 is not showing up in the routing table of router B, as I reckon it could well be directly connected on its BRI. Is the BRI down on router B? But router B has used the address as its RID, so it must be quite a recent version of the IOS if it can use the IP address of a "down" interface as its RID.

Kevin Dorrell

Luxembourg

mhussein · ‎03-11-2005

This is really odd, look at router A:

Show ip route (router A)

C 170.10.129.0/30 is directly connected, BRI0

interface Serial0

ip address 170.10.29.1 255.255.255.252

Why is Serial0's conifgured subnet appears as directly connected to BRI0?

Is there a typo in the configs? ip unnumbered/DDR dialer still up/etc ...? What happens if you shutdown BRI0?

Could you post your complete configs please?

Regards,

Mustafa

Richard Burts · ‎03-11-2005

It is not the same network on serial 0 and BRI 0. Look at the third octet and the serial is 29 and the BRI is 129.

They are close enough to be confusing but not the same.

HTH

Rick

HTH

Rick

Kevin Dorrell · ‎03-11-2005

Mustafa,

I think I can explain that:

The link between the routers is on S0 on both, and that is 170.10.29.0/30, with .1 on RouterA, and .2 on routerB.

The RIDs of the routers are 170.10.129.1 and 170.10.129.2 respectivly, just 'cos that happens to be the highest numbered interface in each box.

So it is quite logical that Router A should see RID 170.10.129.1 through S0, and vice versa. Don't forget that the RID is not actually an IP address, it is just a "random" number identifier derived from the IP addresses. RIDs do not need to routable.

It would have been easier to read if we had set a specific router-id.

Kevin Dorrell

Luxembourg

mhussein · ‎03-11-2005

Rick, Kevin thanks for the headsup and explanation.

I made mistake there, I guess when I used the phrase "look at" I wasn't lookin myself.

So far what I understand from your replies, router B's RID is the BRI ip 170.10.129.2, and the BRI itself is MIA. The router shouldn't be using the ip of a down interface as RID.

Is that correct?

Hey TGIF!

Regards,

Mustafa

Kevin Dorrell · ‎03-11-2005

Mustafa,

I think some later versions of IOS will use the IP address of a "down" interface, but I'm not 100% sure about that.

It used to be that if you derived the RID from an interface IP address, and then you changed its address, the RID would change, and the whole network would start recalculating its routes. Well, they fixed that in later releases of IOS - once the RID is fixed, it's fixed.

But I cannot remember exactly what happens in later releases if the highest interface is down at the time OSPF is started. In any case, it is best to do it with a loopback or with and explicit router-id - that avoids the problem.

Kevin Dorrell

Luxembourg