Note: If spines are not used, then these steps are not required for a single rack deployment.
Perform the following steps to configure unnumbered BGP for the default VRF: - Configure the BGP ASN with the router bgp command. Use the same AS number that was used in Configure unnumbered BGP for the nondefault VRF.
- For the router ID, specify the IP address that is configured on interface Loopback 0.
- Configure a maximum multi exit discriminator (MED) value to send on startup. This setting helps to minimize switch failover time.
- Enable ECMP with the bestpath as-path multipath-relax command.
- Enter the address-family ipv4 unicast command.
- Specify the redistribute connected command to redistribute IPv4 routes from physically connected interfaces.
- Configure the maximum number of eBGP routes that can be used to forward packets over multiple paths to a neighbor.
- Enter the address-family l2vpn evpn command.
- Advertise all VNIs.
- Advertise the primary IP (the router ID) of the switch.
- Specify the router ID IP address of the MC-LAG peer switch.
- Disable duplicate address detection.
- Create a peer group for the spine connections.
- Configure a remote autonomous system to exchange routes through eBGP. The remote-as external command is required for unnumbered BGP.
- Configure the time (in seconds) between sending keepalive messages to a BGP neighbor, and the wait time to receive a keepalive message before considering a BGP peer to be dead.
- Configure the route advertisement interval between sending BGP route updates to neighbors.
- Enable BFD. BFD rapidly detects communication failures between two adjacent routers.
- The capability extended-nexthop command is required for unnumbered BGP. It enables BGP to negotiate the extended next hop capability with a peer.
- Enter the address-family ipv4 unicast command and activate.
- Set allowas-in to accept routes with the local AS number in updates.
- Enter the address-family l2vpn evpn command and activate.
- Set allowas-in to accept routes with the local AS number in updates.
- Put the interfaces that are connected to the spines in the spine peer group.
Leaf1A Leaf1B router bgp 65101
router-id 10.1.1.1
max-med on-startup 420
bestpath as-path multipath-relax
!
address-family ipv4 unicast
redistribute connected
maximum-paths 2
!
address-family l2vpn evpn
advertise-all-vni
advertise-pip ip 10.1.1.1 peer-ip 10.1.1.2
no dup-addr-detection
!
peer-group SPINE
remote-as external
timers 3 9
advertisement-interval 5
bfd
capability extended-nexthop
!
address-family ipv4 unicast
activate
allowas-in 2
!
address-family l2vpn evpn
activate
allowas-in 2
!
neighbor interface Eth1/53
peer-group SPINE
!
neighbor interface Eth1/54
peer-group SPINE
!
neighbor interface Eth1/55
peer-group SPINE
!
neighbor interface Eth1/56
peer-group SPINErouter bgp 65101
router-id 10.1.1.2
max-med on-startup 420
bestpath as-path multipath-relax
!
address-family ipv4 unicast
redistribute connected
maximum-paths 2
!
address-family l2vpn evpn
advertise-all-vni
advertise-pip ip 10.1.1.2 peer-ip 10.1.1.1
no dup-addr-detection
!
peer-group SPINE
remote-as external
timers 3 9
advertisement-interval 5
bfd
capability extended-nexthop
!
address-family ipv4 unicast
activate
allowas-in 2
!
address-family l2vpn evpn
activate
allowas-in 2
!
neighbor interface Eth1/53
peer-group SPINE
!
neighbor interface Eth1/54
peer-group SPINE
!
neighbor interface Eth1/55
peer-group SPINE
!
neighbor interface Eth1/56
peer-group SPINE