State of resilient nexthop ECMP

[Christopher Munz-Michielin]

Hey all,

Wondering what the state of support for resilient ECMP with BIRD is?  
I'm currently using BIRD to send traffic to a number of backend servers 
using ECMP, but of course if one server leaves the group, the entire 
hash is recalculated which results in clients being potentially directed 
to a different server resulting in tcp resets.

I've been reading up on resilient nexthop which seems to solve this 
issue, and is supported natively in FRR, but I can't find any 
documentation about it's level of support in BIRD.  Wondering if anyone 
else has had experience with this technology and BIRD.