OSPF Network Design Solutions

Frequently Asked Questions

Q—Are there any outstanding issues with routing OSPF over IP unnumbered interfaces?

A—Yes, Cisco bug ID: CSCdi46217 discusses problems with IOS 11.3(8) and 11.0(3) where OSPF cannot route packets over ISDN with IP unnumbered on the interface. This problem has been fixed in later versions of the IOS. Consult the TAC if you have any further questions or if you think the IOS is not running OSPF properly.

Q—I am using OSPF Debugging and the router is giving me an error message: authentication-key match fails clear text. What does this mean?

A—This message is being generated by OSPF because it is detecting a neighboring router that has a different authentication key than the router giving the message.

Q—I have OSPF configured on a X.25 network and it is not routing correctly when looking at the OSPF database in the router. I can see all the correct networks but when executing a show IP route, there is no routing table.

A—Make sure that the command IP OSPF Network Broadcast is applied to all interfaces connected to other OSPF routers.

Q—I am redistributing routes into OSPF and my subnetted routes are not being redistributed. Why?

A—When redistributing routes into OSPF, if the keyword subnets is not specified in the router running OSPF, they will not be redistributed. The default in the redistribution command is to only allow unsubnetted routes through.

Q—When you force OSPF to generate a default route (default-information originate always metric 50 route-map abc), how does OSPF know the IP address of the default gateway?

A—OSPF does not care. Default route is announced as external route in OSPF and the only information you have about an external route is which router is announcing it (plus the metric).

Q—Will OSPF work well in an FDDI environment?

A—Yes. FDDI is what a lot of people use as their backbone (area 0).

Q—Do we have to manually set up adjacencies for routers on the SMDS cloud with the OSPF neighbor subcommand?

A—In Cisco IOS 9.1, you need the OSPF neighbor command to make OSPF work on SMDS. In Cisco IOS 9.21 and later, you use the multicast capability. Note that neighbors are no longer needed as of IOS 10.3

Q—Does the offset-list subcommand work for OSPF, or is it implemented only for IGRP, RIP, and Hello?

A—It is implemented only for IGRP, RIP, and Hello. It does not work with OSPF.

Q—For OSPF, is there a suggested method of determining the lowest possible ospf dead-interval to use (and corresponding ospf hello-interval to use with the dead-interval)? We are currently using the default values and would like to reduce the convergence time.

A—The default value for the dead-interval is four times the hello interval. In Cisco’s implementation, that is 40 seconds for broadcast networks and two minutes for nonbroadcast networks. However, adjusting these parameters might not be sufficient enough to meet your needs, and things like network topology, traffic loading, and keepalive timers on the interfaces also play key roles in determining the speed of the convergence time. As this request can’t be answered in a single sentence, you should contact your local support for assistance so that a detailed evaluation can be performed on your network before any OSPF parameter is adjusted. However, if you have decided to do it yourself, keep in mind that if you lower the dead-interval timer, be sure to lower the hello-interval timer as well. In addition, the hello-interval timer must be the same for all nodes on a specific network. This rule also applies to the dead-interval timer.

Q—Will a Cisco router transform subnet mask information correctly if I redistribute OSPF into IGRP? My users are clamoring for VLSM. I’d like to give them the ability to use OSPF (or other routing protocol of their choosing) in the Ethernet side of their router, while still maintaining IGRP across the WAN. Will this work? We use a 9-bit mask on the WAN side. The mask on the LAN side varies between 4 and 12 bits.

A—No, it won’t work because IGRP doesn’t support VLSM. When you are redistributing the VLSM routes into IGRP, some or all of the VLSM routes will be lost. If VLSM is definitely required, then you will have to run either OSPF or EIGRP on your WAN side to accommodate the VLSM routes from the remote sites. You could run IS-IS, but it is more complicated.

Q—I have a Cisco 7010 router that is running OSPF and IGRP. OSPF is used to route within my network, and IGRP is used to my service provider. I redistribute between them. I also have two static routes. I have a requirement to advertise one of my static routes out all of my OSPF interfaces and my IGRP interface. The second static route needs to be advertised out all of my OSPF interfaces but needs to be filtered from being advertised to my service provider via IGRP. I think I need to use the distribute-list and access-list commands in my IGRP process, but I am not sure how to do it.

A—The best way is to use route-map. Here is an example:

! router ospf 333 redistribute static subnets route-map ospf-static network 128.30.0.0 0.0.255.255 area 0.0.0.0 ! router igrp 187 redistribute static subnets route-map igrp-static network 128.99.0.0 ! ip route 128.30.20.0 255.255.255.0 128.30.28.21 ip route 128.30.26.0 255.255.255.0 128.30.28.22 ip route 128.101.0.0 255.255.0.0 128.99.15.2 ! access-list 7 permit 128.30.20.0 0.0.0.255 access-list 7 permit 128.30.26.0 0.0.0.255 access-list 7 deny any access-list 8 permit 128.101.0.0 0.0.255.255 access-list 8 deny any ! route-map ospf-static permit 10 match ip address 7 ! route-map igrp-static permit 10 match ip address 8 In the preceding example, there are three static routes defined. The first two static routes will be redistributed into the OSPF domain only, and the very last static route will be redistributed into the IGRP domain only.

Q—Can EIGRP or OSPF give me full connectivity for IP and IPX in a partially-meshed Frame Relay network? Or will I have to configure subinterfaces?

A—OSPF has a feature called point-to-multipoint interfaces to easily allow full connectivity over a partially-meshed Frame Relay network in Cisco IOS 11.0 and later. It could be done before Cisco IOS 11.0, but it required the hub router to be the designated router and some map statements on each spoke router to all the rest of the spoke routers through the hub router. By the way, OSPF only supports IP. EIGRP does allow it, too, but requires careful configuration, taking the non-fully-meshed nature on the network into consideration.

Q—I have both OSPF and EIGRP running with default administrative distances. Will the OSPF routing table be ignored because it has a higher cost? If I then add static routes, will the OSPF and EIGRP be ignored until I remove the static routes?

A—Yes. The default administrative distances are: Static 1; EIGRP 90; OSPF 110. Therefore, static is always preferred over an EIGRP/OSPF route to the same destination. The administrative distance is a measure of the trustworthiness of a routing information source. The default administrative distance can be altered using the distance command, but this is not recommended.

Q—What is the purpose of the ip-ospf-transmit-delay command?

A—If the delay is not added before transmission over a link, the time in which the LSA propagates over the link is not considered. The default value for this command is one second. This parameter has more significance on very low-speed links, where it can alleviate some bandwidth constraint.

Q—In Cisco IOS 9.1, is the neighbor command required when running OSPF over X.25 networks?

A—You need the neighbor command to make OSPF work on X.25 in Cisco IOS 9.1. In 9.21 and later, at OSPF level, an X.25 network can be configured to be a broadcast network, and OSPF treats X.25 as a broadcast network only. X.25 maps with the broadcast keyword are be needed to make it work.

Q—What does %OSPF-3-DBEXIST mean?

A—This message appears when the router receives an LSA older than the one in its own database. No action is necessary. A DDT (CSCdi48981) exists to document this message.

Q— Can I run OSPF with IOS 9.1? I want to change my routing protocol from RIP to OSPF. I have a WAN network with IGS routers running Cisco IOS 9.1 to Cisco 7513s running Cisco IOS 11.0. The network consists of 20 routers.

A—Yes, running OSPF in 9.1 software is fine. Sites with Cisco IOS 11.0 and Cisco IOS 9.1 together are all running OSPF without any problems.

Q—I need help to understand Cisco’s OSPF network statements. I have a router that I am configuring for OSPF. The router has two interfaces:

e0 - 10.120.175.254 mask 255.255.255.128 s0 - 10.120.173.114 mask 255.255.255.252 (numbered serial, 8 address subnet) Both interfaces belong in the same area. However, when I use the following commands, OSPF does not work:

network 10.120.175.128 0.0.0.128 area 0.0.0.2 network 10.120.173.112 0.0.0.252 area 0.0.0.2 Cisco IOS appears to have problems resolving the VLSM that I’ve defined. What am I missing?

A— Your network statements should have been:

network 10.120.175.128 0.0.0.127 area 0.0.0.2 network 10.120.173.112 0.0.0.3 area 0.0.0.2 OSPF uses inverse masks for defining the VLSM:

* 0.0.0.127 is the inverse mask for 128 (127 + 128 = 255) * 0.0.0.3 is the inverse mask for 252 (252 + 3 = 255)

Q—What is the difference between Type 1 and Type 2 external routes in OSPF?

A—Type 2 uses only the external cost as set in the ASBR. The cost doesn’t change as the route is propagated. Type 1 adds internal OSPF costs as the route is propagated. The default is Type 2, but this can be overridden.

Q—Does OSPF automatically load balance or are additional commands needed? (I have parallel T1 serial links going between an OSPF Area 0 router and an Area 1 router.)

A—The router will load balance over two T1 links if it sees equal OSPF costs to a given destination. The default is up to four parallel paths, and in your case, it should be two. You can verify this by doing a show ip route and look for two paths to a given destination network across the WAN links. Also, if IP fast-switching is turned on, the router will load balance on a per-destination basis, and if IP fast-switching is turned off, the router will load balance on a packet-by-packet basis.