OSPF Route Priority

Unanswered Question
Nov 21st, 2008


I am looking for some help with configuring OSPF path priorities for portions of our WAN network. I have attached a sample drawing showing the current network. Our Core_Rtr-A(7506)is directly connected to our upstream provider. All other routers attached to the network backbone Area 0 receive their default route from the adjacent routers.

I need to have Rtr-H(L3 3550)route through Rtr-G as its preferred path out, but OSPF injects the default-route to Rtr-H from Rtr-N. The only time Rtr-H can route through Rtr-G is if the path to Rtr-N goes down. From a redundancy standpoint, all works fine, but the paths are not as optimal as they should be. I have tried to setup Rtr-H with a static default-route to Rtr-G, but this causes all traffic on Rtr-H and the other connected Rtr's in AREA 5 to stop passing traffic altogether and become unreachable.

This network is a mix of Cisco and Mikrotik routers. From an OSPF routing standpoint, all works very well, but not as desired. Any suggestions would be helpful.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Giuseppe Larosa Sat, 11/22/2008 - 01:52

Hello Jesse,

the picture shows a ring topology in OSPF Area 0.

OSPF default routes can be of two types:

O E2 : only the seed metric is considered

O E1: path cost to ASBR is added to seed metric.

you should use the second type of default route.

on the router(s) that originate use

default-originate metric-type 1

if you do so you can influence the route installed on RH by simply increasing OSFF cost of link of RH to RN so that the default route from RG is chosen and installed in the routing table.

From the picture you can see that RG is 6 router hops away from Border Router Core_RtrA but the path via RG has a DS3 link and on the other side the penultimate hop has a 100 Mbbps link to Core_RtrA.

If all ring links are FE and with default settings that DS3 link has OSPF cost 2 while all FE links has cost 1.

This should explain what is happening here.

Simply put ip ospf cost 3 (or more if needed desired) on RG on link RG to RN.

Hope to help


Jesse Hottle Sat, 11/22/2008 - 09:56


That does exaplain a lot here. The only router where I am using the default-originate command it on the Core_Rtr-A. Do I need to redistribute on all routers?

Giuseppe Larosa Sat, 11/22/2008 - 10:20

Hello Jesse,

the default route should be already propagated in OSPF but every router decides how to go to the default route based on the OSPF cumulated cost to reach your only one ASBR Core_Rtr-A.

I had started to think of a more complex solution.

I would just change to use a metric type1 on Core_Rtr-A and then you can influence RG choices as I've suggested in previous post.

In other words path to the default route follows the path to reach ASBR Router IP addresse(s)

Hope to help


Jesse Hottle Sat, 11/22/2008 - 11:06


Thanks for the clarification. I already have our Core_Rtr-A setup with the metric type-1 command.

With this being the case, are you saying I can just change the cost value on the interface on Rtr-H that points to Rtr-G to make it the preferred route?

Giuseppe Larosa Sat, 11/22/2008 - 13:12

Hello Jesse,

the cost is used outgoing direction: increase the cost in the direction you don't want to take on the router you want to change behaviour.

With default values an FE has cost 1 so you cannot lower it so you need to increase on the other side/interface

Hope to help


Jesse Hottle Sat, 11/22/2008 - 16:30


I changed the cost on the FE interface pointed to Rtr-N to Cost=2. This dropped all connectivity to the router on both interfaces. I was able to login to one of the adjacent routers and telnet from there to change the cost back to 1. Could this have something to do with this Rtr-H and Rtr-N connecting to another OSPF area?

Giuseppe Larosa Sun, 11/23/2008 - 01:33

Hello Jesse,

it is difficult to say what happened.

I would suggest you to post a filtered version of your configuration.

Also with unchanged OSPF costs you should provide

sh ip route

with yoursource.net I mean where is your PC.

also it would be useful to see

sh ip ospf database external


sh ip ospf neighbors

sh ip ospf

I'm sorry that my suggestion caused this problem, there is some other aspect that we need to take in account.

RG is a router that is part of Area 0 or not ?

I supposed it had at least two interfaces in area0.

are the neighbors in area 0 or in another area ?

The show commands I've asked help to answer to these questions.

Hope to help


Jesse Hottle Sun, 11/23/2008 - 10:32

I was changing the cost on Rtr-H, this is the router I want to control the paths out on.

All neighbors are in Area 0, except for routers I and M, they are ABR's between Area0 and Area 5. I have attached what you requested.

Thanks for the help,


Giuseppe Larosa Sun, 11/23/2008 - 11:56

Hello Jesse,

the default route is an O E1 and this is good.

About the change in ip ospf costs I think for the ring nature of your topology that is better to increase the costs on both RH and RN on their common link to a common higher value.

In this way you should move the "dead" link from RH to RG to RH to RN and this could provide the desired change in moving traffic coming from RI and area 5 to RG.

I would suggest to use an high value like ip ospf cost 500 on both sides so that the link becomes a backup route between RH and RN.

You can experience an OSPF convergence of up to 40-45 seconds.

If you can access RH by a telnet session on RG to be sure to be able to control it.

A maintanance window of 1 hour should be enough.

Hope to help


Jesse Hottle Mon, 11/24/2008 - 13:03

Cool. I am going to try this tonight. I will let you know what happens.

Thanks again.



This Discussion