OSPF and different MTU

[Konstantin Dunaev]

Hello,

I will now try to ask here

I have some problems with stability of my OSPF session between 3 devices. the topology is simple:

cisco7200 --- catalyst3560 --- cisco2600

the cisco7200 and catalyst3560 are speaking MTU 1546 (because of QinQ and MPLS)

cisco2600 speaks only MTU 1500.

The OSPF between cisco7200 and catalyst3560 works perfect, but the OSPF betwenn catalyst3560 and cisco2600 is not stable and doesn't come correctly to Full adjacency. The reason is quite stupid - catalyst3560 doesn't support differet MTU on different Interfaceses and tries to send OSPF packets with MTU 1546 to cisco2600, which obviously drop them.

The comman "ip ospf mtu-ignore" IS CONFIGURED on the both sides  but doesn't help, and I know that MTU value should be the same alone the  whole path, but cisco2600 is just a "spoke" router and doesn't forward  any backbone traffic and cisco2600 doesn't support lager MTU anyway.

What would be the best soluition or work-around? static routing is not really an option.

[cadet alain]

Hi,

I have a workaround to propose: run EIGRP between 3560 and 2600 and then redistribute routing processes on the 3560.

Regards.

Alain.

[Konstantin Dunaev]

Hi Alain,

thank you for the answer.

I'll think about second routing protocol, but the topology will become more complex, from other side the cisco2600 is not key part of my network.

But I suppose EIGRP will face the same problem with MTU as OSPF, if catalyst3560 sends EIGRP packet of 1546 bytes (as MTU defines) it will be droped by cisco2600

[cadet alain]

Hi,

But I suppose EIGRP will face the same problem with MTU as OSPF

Yep you're right,  forget my workaround

Regards.

Alain.

[Tomas Fidler]

Are you aware of "system mtu routing" global configuration command?

As I understand you are using bigger MTU for MPLS and dot1Q reasons. For IP packet routing packet size are still limited to 1500B.

It may be solution of your problem. Try to investigate if this command  don't break you network and everything works as expected (nogiants, no  packet drops).

Tomas

[Konstantin Dunaev]

Hello Tomas,

thank you for the advise, but we've already discussed this command in "LAN,routing,switching"  community and unfortunatly it doesn't help for Gigabit ports at all, and it makes changes only for routed interfaces (no switchport) and not for SVI's.

[Tomas Fidler]

I've looked thru mentioned discusion. and it seems that the topology is different than you draw it.

What devices are ospf neighbors? How is c3560 involved in routing and switching.

For example if 7200 and 2600 are ospf neighbors, it is important that they must have the same MTU on interface/subinterface because there will be possibility of droping Giants otherwise.

[Konstantin Dunaev]

Tomas,

the topology (simlified) looks like:

cisco7200 <--OSPF--> catalyst3560 <--OSPF--> cisco2600

cisco7200 and catalyst3560 are OSPF neighbours

cisco2600 and catalyst3560 are OSPF neighbours

cisco2600 and cisco7200 dont's speak OSPF with each other.

or may be more simple:

cisco7200 are backbones and in OSPF area 0

catalyst3560 are edge devices and are in their own OSPF area

cisco2600 are stub routers only for management purposes in catalyst3560's OSPF area.

FIDLAFIDLA wrote:
For example if 7200 and 2600 are ospf neighbors, it is important that they must have the same MTU on interface/subinterface because there will be possibility of droping Giants otherwise.

and it's exactly what happens between catalyst3560 and cisco2600

[manish arora]

Hi,

how are the 3560 and 2600 series connected ?

Manish

[Konstantin Dunaev]

Manish,

Ethernet, back to back, catalyst3560 has different VLANs and SVIs for link between cisco7200 and cisco2600

[manish arora]

I am assuming the Link between 3560 & 2600 is point to point , Have never done something like this & it's purely an idea :-

1> Place the 2600 in a different area but still allow the P2P to be propagated from 3500 series.

2> Create a gre tunnel between 7200 and 2600 , modify the MTU on tunnel interfaces , connect that new area with area zero using a virtual link inside the gre tunnel.

3> Place filters to avoid routing loops & may be need few static routes for tunnel endpoint reachabilty.

Then again It would be complex, but still serves the purpose.

Manish

[Konstantin Dunaev]

manisharora111 wrote:
I am assuming the Link between 3560 & 2600 is point to point , Have never done something like this & it's purely an idea  :-
1> Place the 2600 in a different area but still allow the P2P to be propagated from 3500 series.
2> Create a gre tunnel between 7200 and 2600 , modify the MTU on tunnel interfaces , connect that new area with area zero using a virtual link inside the gre tunnel.
3> Place filters to avoid routing loops & may be need few static routes for tunnel endpoint reachabilty.
 Then again It would be complex, but still serves the purpose.

GRE could be a soilution, but... I don't have cisco7200 in all locations, only in a few of them, it would meen that I need 5-6 GRE tunnels on each cisco7200, but I'll really think about it, as it's a really good  idea.

[manish arora]

Hi,

I am not exactly sure about your topology, but if you consider placing 2600 in it's own Area , then you will just need one Tunnel with a 7200 router to connect that New area with AREA0.

Note: Gre with routing protocols comes with their own issues, test it before you start anything on Production .

Manish

[rtannertwc]

I may be thinking too simplistically here but can you put both connecting interfaces into a passive-interface state while you look for a permanant solution?  That should keep them from looking for neighbors but still advertise connected subnets.

[Konstantin Dunaev]

Ryan,

rtannertwc wrote:
I may be thinking too simplistically here but can you put both connecting interfaces into a passive-interface state while you look for a permanant solution?  That should keep them from looking for neighbors but still advertise connected subnets.

that would be an idea, but I can't follow your: how cisco2600 will get routing information and advertise its own routes if there is no OSPF adjacency between catalyst3560 and cisco2600?