BGP neighbors flapping in vPC scenario

[maverick0]

I am facing a BGP flap issue between C3750 and C9300 where the BGP session is tearing down because the hold timer is being increased for some reason that I couldn't find yet.

Basic topology:

C3750 <-----> N3K (vPC domain) <------- L2 circuit -----> N3K (vPC domain) <------> C9300

In the physical aspect, the C3750 is directly connected to N3K using a orphan port and the C9300 is connected to N3K using a vPC. The both nexus are in different DC and interconnected by a L2 circuit.

In the logical aspect, in the C3750 the interface is configured to use ip address and the C9300 is configured to use a VLAN Interface (Vlan100).

I am running eBGP between the devices and using a simple configuration (neighbors and authentication), nothing else. I know there is some risk to run routing protocols over vPC scanarios, but the BGP session is going down because the vPC is drop the packet internally and the BGP is being affected by this behavior.

Thanks in advance

[Geevarghese Cheria · Accepted solution]

The "hold time expired" is very often related to a different TCP maximum segment size (MSS) use by both neighbors. Please have a look at this forum discussion - https://community.cisco.com/t5/routing/bgp-peering-flapping-hold-time-expired/td-p/3389106