[c-nsp] BGP "keep alives throttled do to tcp", MTU mismatch?

[Scott Granados]

Hi, I have a puzzler.

I have two Cisco 7206VXr routers one with a G1 and the other with a 400 processor.  
Between them terminated on a gig interface on each side is an ILEC provided metro E connection that’s pretty vanilla using VLANs to address different connections.

I have a /30 set up on a common VLAN between the two with the BGP peer set up on that /30.
    After 3 minutes, my session will reset I’m assuming do to lack of keepalives / problems with MTU.  Lots of googling as pointed me at various MTU mismatch situations but I’m not sure what I’m doing wrong possibly assuming something is set “in an interesting way” between the sites on the ILEC network.

On each interface I have the following

int gig x/x
mtu 1522

Assuming 1500 for standard Ethernet and 22 for the VLAN tags.
I also have the following set in global mode
ip tcp path-mtu-discovery

On both VXR devices code 12.4-25C is being run.

    I would appreciate any pointers to investigate.  What have I missed?


Thanks
Scott