Solved: OSPF problem

rvr_76bg · ‎04-04-2008

Hello,

I have recently started in networking department in my company. I am attaching 2 txt files, one with sh run output , second with sh ip route output. Interfaces s0/0/0 and s0/1/0 are connected to 2xT1 to the other site of the company. The configuration of the other site is the same. I really think that the config of the interfaces above is wrong because we have 2 interface on this device on the same subnet with /29 mask so, IP/s 192.168.254.157 and 153 are from the same subnet (192.168.254.152/29). Can OSPF be confused by this configuration?? I see (sh ip route.txt) that I have 192.168.254.152/29 directly connected but but I see two additional routes for these connected interfaces with /32 masks which is very strange for me. Is this normal?? I see two running OSPF processes 151 and 152. What I know we need only one OSPF process per device because this is less calculation for the CPU of the router and it is good practice to keep the same OSPF process in entire domain in area 0. Am I right?

Under both interface's configs I dont see bandwidth command neither ip nbar protocol-descovery. However Auto QoS is used. I think bandwidth command is mandatory for that kind of QoS config.

Any help will be highly appriciated. Can you refer me to an official Cisco document?

Regards,

Edison Ortiz · ‎04-05-2008

How can I know which OSPF process exactly is running and forming ajecansies with the neighbors?

You can type show ip ospf int bri and look under the PID column for the process ID number.

As Rick stated, the show ip ospf also provides this information under "Number of interfaces in this area".

View solution in original post

rvr_76bg · ‎04-04-2008

No one can help???

Edison Ortiz · ‎04-04-2008

I really think that the config of the interfaces above is wrong because we have 2 interface on this device on the same subnet with /29 mask so, IP/s 192.168.254.157 and 153 are from the same subnet (192.168.254.152/29). Can OSPF be confused by this configuration??

Yes, they are in the same subnet and it must be corrected ASAP.

I see (sh ip route.txt) that I have 192.168.254.152/29 directly connected but but I see two additional routes for these connected interfaces with /32 masks which is very strange for me. Is this normal??

Yes, that's normal when using PPP as the encapsulation protocol. You can disable this feature with the command no peer neighbor-route and then clear the routing table.

I see two running OSPF processes 151 and 152. What I know we need only one OSPF process per device because this is less calculation for the CPU of the router and it is good practice to keep the same OSPF process in entire domain in area 0. Am I right?

I don't see the practical reason for running 2 OSPF process as this is not a common practice. You must find out from your networking group what was the idea behind that design.

As for area 0, this is another network design consideration. It all depends on the amount of LSAs you have in your network. Based on the show ip route, you don't have many...

Under both interface's configs I dont see bandwidth command neither ip nbar protocol-descovery. However Auto QoS is used. I think bandwidth command is mandatory for that kind of QoS config.

You are right. As they are using 'percent' values rather than 'whole' values, a bandwidth command is often recommended under the interfaces to avoid the inherited bandwidth value from the interface. However, I don't see a problem on the current configuration unless you are running a fractional T1.

HTH,

__

Edison.

mattcalderon · ‎04-04-2008

Excellent reply Edison! You really did take some time to help out this new network engineer. Bravo!

Richard Burts · ‎04-04-2008

Rumen

Perhaps I can add a little to this discussion. As far as both serial interfaces being in the same subnet: if these were LAN interfaces then it would be a problem and I would agree with Edison that it would need to be corrected. But with point to point serial interfaces it is not a problem. It is not a common configuration but it does in fact work. If they were reconfigured to be unique /30 subnets then it would be a more common configuration. But I do not agree that having them in a common /29 is a problem.

The configuration of the 2 OSPF processes is quite odd and I suspect that it does not work in the way that the person who configured it expected that it would work. Of the 8 network statements under each process 7 are exactly the same under both processes. But in OSPF an interface can be in only a single OSPF process. So the 7 networks will be in OSPF 150 and not in OSPF 151. The difference between the processes is that OSPF 150 includes the serial interfaces while OSPF 151 includes the loopback interface. The result is that OSPF neighbors will form on the serial interfaces but will not advertise the address/subnet of the loopback. And OSPF 151 will not have any neighbors to communicate with. I do believe that someone needs to re-evaluate the OSPF configuration and probably consolidate into a single process.

HTH

Rick

HTH

Rick

Edison Ortiz · ‎04-04-2008

Matt,

I appreciate the kind words, thanks a lot.

rvr_76bg · ‎04-05-2008

Guys,

These were extremely helpful. Thank you very much! One last question:

How can I know which OSPF process exactly is running and forming ajecansies with the neighbors? I supose they cannot form ajecancies both in the same time with the same neighbors.

Regards,

The entire post was for a problem with OSPF going crazy and the router is down for a while.

Can this problem be caused by the two running processes in OSPF??

Richard Burts · ‎04-05-2008

Rumen

the command show ip ospf would be a good place to start. It will show you what OSPF is doing including how many processes and what they are doing.

You are correct that they cannot form ajecancies both in the same time with the same neighbors. Neighbors and adjacencies will be formed on an interface in one process or the other but not in both.

It is hard to know if running 2 processes caused the problem or not. If some of the neighbors formed on interfaces in process 150 needed reachability to the loopback interface in process 151 then that could have been a problem. It is possible that some interfaces got into process 150 and some other interfaces into process 151 and that could have caused a problem.

It is possible that 2 processes contributed to a problem. I do not see any good reason to run 2 processes like that - especially with so many interfaces configured in both processes. Unless someone there knows of a good reason why it was set up that way I would certainly recommend that the configuration be changed and the OSPF be consolidated into a single process.

[edit] in re-reading my response I would like to make clear that running 2 OSPF processes is not necessarily a problem, but having interfaces configured in both processes is more the problem.

HTH

Rick

HTH

Rick

Edison Ortiz · ‎04-05-2008

How can I know which OSPF process exactly is running and forming ajecansies with the neighbors?

You can type show ip ospf int bri and look under the PID column for the process ID number.

As Rick stated, the show ip ospf also provides this information under "Number of interfaces in this area".

rvr_76bg · ‎04-07-2008

Thanks guys,

It really worked.

Regards,