OSPF Link and MTU Modification

[Chaker BCHIR]

I have 2 Nexus 5Ks connected with an OSPF Link implemented. The MTU value is 1500 by default.

The neighborship is OK.

Now, I have modified the MTU value of the 2 extremities of the OSPF Link to 9216. The Neighborship is KO and the state is stucking in an EXCHANGE State.

Do you have any idea of this behaviour ? I did a rollback.

Remark : I don't want to use the "ip ospf mtu ignore" command.

Thanks in advance.

[Julio E. Moisa]

Hi

OSPF must have the same MTU value under the interface between routers to create the adjacency. Have you tried ignoring the MTU? it should be applied where the MTU was not changed.

interface gx/x

ip ospf mtu-ignore

Hope it is useful

:-)

[IBMeR]

Hmm..I don't know exactly what do you have, but it might be this:

https://quickview.cloudapps.cisco.com/quickview/bug/CSCtl46093

[Chaker BCHIR]

Yes I have exactly this Log message !

%KERN-3-SYSTEM_MSG: packet_sendmsg: packet size 9246 > MTU 9230 - kernel

But in the workaround it says that the known fixed release is 5.1(3)N1(1) while I'm now with 7.0(7)N1(1) version and it is recommanded by Cisco.

[IBMeR]

It looks like a bug - I think NXOS just didn't add OSPF header lenght.

But you might wait somebody from Cisco answer(or open case in TAC) or just decrease mtu in policy-map by 16 bytes.