OSPF link only as backup?

[pashdown1]

I have the following OSPF setup:

Area 100---(OSPF cost 1000)---Router A---(10Gb link)---Router B----(OSPF cost 0)--Area 100

"Area 100" is the same network and the same router, so this is really one big circle.  I desire to only use the link on Router A if Router B cannot reach "Area 100", yet in spite of playing with the cost of the link on Router A, it remains the only route to "Area 100" Router A knows about.  The area between Router A & B is backbone (0.0.0.0), but the OSPF route for Area 100 are not shared over the 10Gb link.  Other OSPF routes are propagating over the 10Gb link just fine.  Any ideas?

[pashdown1]

I am going to test this with the customer this week, but I think you're on the right track with an additional OSPF process to overlap the area.  The only correction I have is that 6509 isn't an area, it is my Cisco 6509 router.  The area would be 0.0.0.0 for those.

Will the routes be shared between processes?  IE: If I have process 100 for area 0.0.0.0 will the routes from process 104 for area 104 be redistributed into process 100 and subsequently area 0.0.0.0?

[tobyarnett]

Pete,

OK so the 6509 makes more sense as a switch type than area. Per your question on the router with two OSPF areas on it, they will not automatically redistribute routes between areas. They will keep a database of the each processes routes, but in order to pass routes from one process to the next you can use a redistribute command. My personal preference is to use the route-map command in your redistribution. This allows for flexibility for future use, but if you want simplicity you can always point an acl only..


Example

Router

Router ospf 100
Network x.x.x.x area 500
Network y.y.y.y area 500

Router ospf 104
Redistribute ospf 100 subnets tag 1 route-map

Route-map permit 10
Match IP address


IP access-list standard
Permit z.z.z.z wildcard


Something like that.

Toby


Sent from Cisco Technical Support Android App

[tobyarnett]

Pete,

Did you ever get resolution to your issue or do you still need help? If every thing is working could you please close the question as answered and rate any helpful posts.

Thank you,
Toby


Sent from Cisco Technical Support Android App

[pashdown1]

Unfortunately, it still isn't resolved.  I hate to keep bugging you guys on this.  Here is what I have now:

Router A:

router ospf 100

  area 200 stub no-summary

  network X.X.X.0 0.0.0.255 area 0.0.0.0

  network X.X.X.108 0.0.0.3 area 200

router ospf 200

  area 200 stub no-summary

  network X.X.X.108 0.0.0.3 area 200

  redistribute ospf 100 subnets

Router C:

router ospf 200

  area 200 stub

  network 0.0.0.0 255.255.255.255 area 200

Now the reason I still have area 200 configured in ospf 100, is that is the only thing that actually works.  If I change the ospf process on Router C to area 100, then I get the proper default route.  If I change it to 200, the neighbor appears, but no routes get exchanged.

[Gabriel Hill]

Pete,

I read through this, and although I believe I understand your topology, I am not 100% sure. Could you make a diagram and post it. It would be easier to lab such a situation up to better assist you.

[pashdown1]

The topology and diagram from above is still the same.  Translated for the snippet above:

Area 100, Router C---(network X.X.X.108)---Router A (Cisco 6509)

Area 200, Router C---(network X.X.X.108)---Router A (Cisco 6509)

Yes, they are overlapping areas.  I'm trying to start simple on this.

[smitesh kharecha]

Hi Pete,

Can you please explain why you are running two different OSPF processes ?

and also a scematic diagram would be helpful.

Also, if you could replace your IP addresses XXX with some other dummy IP addresses as it is bit confusing which XXX is in which areas; so probably dummy IP addresses can be bit more helpful.

Regards,

Smitesh

[tobyarnett]

Pete,

A couple things. Can you list out the Router process your using on Router B. I am drawing a network map to help figure this out better. Router B's config would be helpful.

So I know you can use the "network 0.0.0.0 255.255.255.255 area 200" command as an all inclusive statement, but that is not a best practices. You should really list out each network or use the redistribute connected or static commands. Both are functional and work. I have only used that network statement on a lab, but not in real life practices.

Also why are using Area 0.0.0.0 and Area 200 on Router A? Note that area 0.0.0.0 is the same as Area 0, and if you are trying to get a backbone area network you should convert that to area 0.

See this article:

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094aaa.shtml

-Toby

[pashdown1]

Ok, I'm going to try and sketch this out as much as possible.  Thank you for your patience on this.

Router A Config:

router ospf 1000

log-adjacency-changes

auto-cost reference-bandwidth 1000

area 104 stub no-summary

area 200 stub no-summary

redistribute connected subnets

redistribute static subnets

network 10.7.4.0 0.0.1.255 area 0.0.0.0

network 10.2.149.108 0.0.0.3 area 104

maximum-paths 8

!        

router ospf 200

log-adjacency-changes

area 200 stub no-summary

redistribute ospf 1000

network 10.7.4.108 0.0.0.3 area 200

Router B Config:

router ospf 1000

log-adjacency-changes

auto-cost reference-bandwidth 1000

area 104 stub no-summary

redistribute connected subnets

redistribute static subnets

network 10.7.4.0 0.0.1.255 area 0.0.0.0

network 10.2.132.108 0.0.0.3 area 104

maximum-paths 8

Customer Router Config:

router ospf 1000

log-adjacency-changes

area 104 stub

redistribute connected subnets

redistribute static subnets

network 0.0.0.0 255.255.255.255 area 104

Test Router Config:

router ospf 200

log-adjacency-changes

area 200 stub

redistribute connected subnets

redistribute static subnets

network 0.0.0.0 255.255.255.255 area 200

I setup the Test Router because I was having trouble with a second ospf process taking with the customer.  I'm still having trouble.  Although it says it has 1 interface in that area on both the test router and Router A, it is not sending any routes across in either direction.

As you recall the whole reason I have to setup another process is that I want to share routes from 104 across the 10Gb link between Router A & B, so the link between Router A and Customer Router A (6Mb) is only used if the link between Router B & Customer Router B is down.

[tony.henry_2]

Pete,

These are totally stubby areas. you only get 1 route on the customer and test routers. 0.0.0.0/0  if you want more take the no summary statements out of your router A and B.

Additionally I'm assuming customer A and Customer B are actually the one customer and thats just a drawing error. You'd also want to make sure that the customer LAN is area 104. if it isn't your going to have problems. as you won't be able to influence the network correctly.

You may want to reconsider OSPF as the routing protocol if this is the case. BGP is the best protocol to use between two autonomous (seperate management) systems. If you manage the customer LAN and routers then make sure all their networks are area 104 and it should come together properly.

HTH

Tony

[pashdown1]

Customer A & Customer B refer to different routers under the same customer.  Yes, it is all area 104.

There are no default routes being sent to 200.  When I said "not sending any routes" I meant nothing.  Nothing on sent to the Test Router and nothing sent back to Router A.

[tony.henry_2]

OK, so what does "sho ip ospf neighbors"  have from your router A it should have customer A, router B and test router.

Tony

[pashdown1]

show ip ospf neighbors:

Router A has Customer A, but no Test Router.

Router B has Customer B.

Test Router has nothing.

[tony.henry_2]

Pete,

Well thats part of your problem area 0  is broken, you need to fix this.

Is customer A able to see customer B across through area 104?

are all these devices directly connected? are they able to ping each other

Tony