Hi,
I'm asking my question again
the first part:
I have some c3560 with system MTU set 1546 with interface VLAN10 whose MTU size is 1546 and there is no possibility to change it to another value. and we have some cisco 2600 where I can't set MTU bigger as 1500.
I have a problem in establishing the OSPF adjacency between cisco 2600 abd 3560 , the command "ip ospf mtu-ignore" is set on both side but it dosn't help - the OSPF packets which are sent by c3560 are simply lager as 1500 bytes and are dropped by cisco2600.
what can I do in this situation? Will the command "system mtu routing" on c3560 be usefull here?
Second part:
a network consists of c3560 with MTU 1546, some c7200 NPE-G1/G2 with MTU 1546 and c7200NPE400 with MTU 1500 (it can't be set lager value), There are a couple of "old" c3560 with MTU 1500 somewhere in the middle of network and it not possible to restart them in order to change the MTU value.
c7200 are speaking BGP and OSPF with each other.
the problem is that sometimes c7200 losses their BGP session, I would say in most cases it happens between NPE400 and NPE-G1/G2 whit error message like "session closed by a peer x.x.x.x" after some seconds BGP session goes again UP , and then after some minutes again DOWN .
I would suggest that it can be MTU problem, as the traffic passes those c35660 with MTU1500. The neighbour status showes that "transport tcp path-mtu-discovery" is enabled an all neighbours but it seems doesn't help.
The curious thing about it - if I disable the path-mtu-discovery on the neighbours - the BGP session between them stays stable.
Please help and advice!
