Route target format for EVPN

Hi. I’m configuring 2 leaf switches (SN 2410) with BGP-EVPN for VXLAN overlay. 2 servers running FRR are also configured as the EVPN downstream of the leaf switches. However, no VXLAN tunnels can be created for any VNI between the leaf switches and the servers. Do note that the BGP underlay for this setup is using EBGP for running BGP unnumbered.

Based on what I read on ipSpace blog here , it looks like FRR imports any route target by default as long as the VNI part is the same. E.g. ASN:111. I believe the default behavior for Mellanox/Nvidia Onyx is also the same since I tried configuring a SVI on both leaf switches and the SVIs can ping each other via the VNI.

I’ve configured auto VLAN to VNI mapping with the base of 10,000. So VLAN 27 would be mapped to VNI 10027. After looking at the route target generated by the leaf switches, it looks like it’s using ASN:VLAN format instead of ASN:VNI.

Because of this, the VNI part for the route target is not the same with my FRR servers which causes the VXLAN tunnel not being created for the VNI. So, is there any config that can be done to use the other route target format instead? Or I need to change the base auto VLAN to VNI mapping to 0?

Edit: Tried setting up the mapping base to 0 but unable to do so.

(config) # interface nve 1 nve vni auto-vlan-map base 0
% Invalid base 0. Valid values 1-16773120

Hi Hanis irfan,

“Tried setting up the mapping base to 0 but unable to do so”.

It is not supported , the supported value are:
Base-number Range: 1-16773120

You can create VNI-to-VLAN manual mapping.
switch (config interface nve 1) # nve vni 5000 vlan 5

https://docs.nvidia.com/networking/display/nvidiaonyxusermanualv3104408lts/vxlan+commands#src-2615838741_VXLANCommands-nvevnivlannvevnivlan

https://docs.nvidia.com/networking/display/nvidiaonyxusermanualv3104408lts/vxlan+commands

Thanks.