We primarily use Cisco on our network core, but are deploying a CCR1072-1G-8S+ as a PPPoE end-user concentration box. As such, it needs to speak OSPF with the rest of the core network.
The design is a fairly straightforward dual-uplink; each uplink network having two core routers as DR/BDR.
It's the same two core routers connected to both uplink networks.
Everything's fine on IPv4, but OSPFv3 is only forming adjacencies on one of the uplinks at a time. I think it's because we use custom link-local IPv6 addresses on our routers to make it easier to identify them in routing tables, etc.
The primary core router is fe80::1 on both sides, and the secondary core router is fe80::2 on both sides. Our Cisco access routers have no trouble with this, but the CCR is logging info-level messages as follows:
Discarding packet: no neighbor with this RouterId
RouterId=1.1.1.1
Source=fe80::1
Discarding packet: no neighbor with this RouterId
Routerid=2.2.2.2
Source=fe80::2
etc...
Both routers 1.1.1.1 and 2.2.2.2 are active in the neighbors table on the "A" uplink, but the "B" uplink shows no neighbors.
I've also been able to get them active on "B" with "A" refusing to establish adjacency.
The Cisco routers show the CCR as being stuck in state INIT/DROTHER (i.e. it never reaches 2WAY/DROTHER)
I suspect that the CCR is getting confused by the fact that the same router has the same routerID and same link-local address on both interfaces.
Is this the case?
p.s. It sure would be nice if I could manually assign fe80::809 as link-local addresses on the Mikrotik....