On Sun, Jul 12, 2026 at 07:12:15PM +0000, Jack Ma wrote:
> FDB nexthops let a VXLAN fdb entry point at a group of remote VTEPs, with the
> kernel flow-hashing across the group (commit 1274e1cc4226 ("vxlan: ecmp
> support
> for mac fdb entries")). Each leg carries its own remote IP, but the UDP
> destination port is always taken from the VXLAN device (vxlan->cfg.dst_port)
> and cannot be set per leg.
>
> Some deployments pack several receivers behind a single underlay IP and tell
> them apart by UDP destination port. To spread flows across such receivers
> they
I don't understand the "tell them apart" phrasing. Aren't all of these
receivers interchangeable given you are load balancing between them?
> need a nexthop group whose legs share the remote IP but differ in UDP port,
> which is not currently expressible.
I think you will need to provide more details about the use case given
you are asking us to accept new uAPI which we will need to maintain
forever. The FDB nexthop groups make sense when you want to spread
overlay traffic between several distinct VTEPs. In the EVPN-MH use case
these VTEPs are different TOR switches. In your use case, why can't the
load balancing happen at the target host (e.g., using XDP / TC /
SO_REUSEPORT / flow classification)?
To be clear, I'm not strictly against the idea (we can talk about the
implementation later), but I would like to understand why your use case
can't be addressed using existing mechanisms (or why it's undesirable).
Thanks
