ACI Multi-Pod BGP Peer idle

[Kobus Raath]

Hi,

I have new Multi-Pod deployment and have an issue with the BGP peering between the spine nodes.

Looking at all the spine nodes they all peer with each other except for spine 1022.  No one peers with it and the peers remain idle.  I cannot seem to find any logs related to this and do not see any errors in the fabric.

Any advice to rectify or troubleshoot this further?

Each spine has two links to the respective IPN nodes and they are up. 

VLAN 4 is used on the subinterfaces and MTU correct.

LLDP discovery works and the OSPF neighborship established on all the links.

Loopback addresses defined and are used as the OSPF IDs

All nodes were discovered correctly and are registered with the fabric.

Regards

Kobus

[Beau Poehls · Accepted solution]

Is this spine in pod-2 or a pod other than 1? If so, can you ping the 1022 VTEP address (lo0) when sourcing from the VTEP address (lo0) of a spine in pod-1? This the network between spine 1022 and the IPN device defined in the fabric external routing profile under tenants > infra > policies > protocol > fabric external connection policies?