ESI Multihoming with EBGP only underlay on Nexus

If you are patient you may miss the below pre while configuring ESI multihoming on Nexus switches;

If eBGP is used with VXLAN EVPN multi-homing, the administrative distance for local learned endpoints must be lower than the value of eBGP. The administrative distance can be changed by entering the fabric forwarding admin-distance distance command.

My switches are connected to SRV with SVI configured on both of them;

Switch side configurations which are same on both them;

interface port-channel9
switchport mode trunk
switchport trunk allowed vlan 100,200
ethernet-segment 9
system-mac 0000.0000.2011
spanning-tree port type edge

interface Vlan100
no shutdown
vrf member FW_ZONE_X
ip address 10.1.100.0/31
fabric forwarding mode anycast-gateway

 

When you look at the ip route table;

S1-BL2# sh ip route 10.1.100.1 vrf FW_ZONE_X
IP Route Table for VRF “FW_ZONE_X”
‘*’ denotes best ucast next-hop
‘**’ denotes best mcast next-hop
‘[x/y]’ denotes [preference/metric]
‘%<string>’ in via output denotes VRF <string>

10.1.100.1/32, ubest/mbest: 1/0
*via 192.168.1.98%default, [20/0], 1d11h, bgp-61099, external, tag 61201, segid: 100100 tunnelid: 0xc0a80162 encap: VXLAN

via 10.1.100.1, Vlan100, [190/0], 00:08:55, hmm

This causes the traffic for 10.1.100.1 loop between them as each will have route pointing to each other. Thus after lowering the admin distance of the local routes;

fabric forwarding admin-distance 19

S1-BL2# sh ip route 10.1.100.1 vrf FW_ZONE_X
IP Route Table for VRF “FW_ZONE_X”
‘*’ denotes best ucast next-hop
‘**’ denotes best mcast next-hop
‘[x/y]’ denotes [preference/metric]
‘%<string>’ in via output denotes VRF <string>

10.1.100.1/32, ubest/mbest: 1/0, attached
*via 10.1.100.1, Vlan100, [19/0], 00:00:26, hmm
via 192.168.1.98%default, [20/0], 1d11h, bgp-61099, external, tag 61201, segid: 100100 tunnelid: 0xc0a80162 encap: VXLAN