Hi, I understand that it might be a bit confusing at first, but this is indeed the expected and documented behavior of the BFD proto (see https://bird.nic.cz/doc/bird-3.4.0.html#bfd). Although one might expect that having multiple BFD protocol instances for the same VRF would match the stated interfaces like in other protocols, this is not the case. This is to allow for the most common case of just having empty BFD protocol that works for all eligible interfaces with the protocol defaults. Although, I agree that we should at least warn the user that the other BFD instance is not used to make it less confusing. Will put this in our backlog. Thanks, David David Petera (he/him) | BIRD Tech Support | CZ.NIC, z.s.p.o. On 11/3/26 19:44, Miao Wang wrote:
Hi,
Before commit f9deedf1, BFD was implemented allowing only one instance of BFD protocol. The implementation was not prepared for multiple BFD protocol instances. In that commit, to support VRF, the check for multiple BFD instances was removed, which results in the possibility that multiple BFD protocol instances are configured for one VRF. `bfd_add_request()` will success only if given `bfd_proto` and `bfd_request` belong to the same VRF. So for each VRF, only the first BFD protocol instance will take effect. This can be reproduced by the following configuration with the latest commit in branch master (4bee51c82):
router id 192.168.0.1;
ipv6 table main;
protocol bfd bfd1{ interface "eth1" { interval 5s; idle tx interval 5s; }; };
protocol bfd bfd2{ interface "eth2" { interval 10s; idle tx interval 10s; }; };
protocol device { };
protocol static { ipv6 { table main; }; route 2001:db8::/32 via fe80::1%eth1 bfd via fe80::2%eth2 bfd ; };
This configuration specifies interval 10s for eth2 and 5s for eth1. Actually, only the configuration in the protocol bfd1 block takes effect:
% birdc sh bfd sess BIRD 2.20.0 ready. bfd1: IP address Interface State Since Interval Timeout fe80::1 eth1 Down 18:30:36.395 5.000 0.000 fe80::2 eth2 Down 18:29:39.393 1.000 0.000
bfd2: IP address Interface State Since Interval Timeout
It can be seen that the interval for fe80::2 on eth2 is 1s instead 10s, which is beyond expectation.
Cheers,
Miao Wang