I had an issue with a 6.40.8 HAP ac lite talking to a Ubiquiti EdgeRouter Pro 1.8.5 via a Netonix switch. The mikrotik was seeing hellos from the ER-Pro. The ER-Pro was not seeing the hellos from the Mikrotik. OSPF logging wasn't helping on the Mikrotik. I enabled "debug ospf" on the ER-Pro. Nothing useful in the logs. I upgraded the HAP AC lite to 6.42.3, same symptoms. I downgraded to 3.93.3, same symptoms. So, I went back to the bugfix channel. I was about to kick the Netonix in the head, but I could ping perfectly from the HAP to the ER-Pro and vice versa.
I eventually ran "sudo tcpdump -i eth3 proto 89 " on the ER-Pro. That clearly showed that the HAP-lite was sending hellos using the 10.115.0.1 src IP on that interface rather than the 204.x.x.x/30 globally routeable IP which Routing OSPF Interfaces showed was selected for that interface. The 10.115.0.1 IP is not covered by either of the two Routing OSPF Networks subnets.
I disabled the 10.115.0.1/24 IP address and my neighbor adjacency came up immediately. Re-enabling the the IP address seemed to continue to work. I've seen other devices where the router uses the "primary IP" on an interface. Since I can't set primary/secondary on a MikroTIk, I just deleted the 10.115.0.1/24 IP. I then added it again, hoping that would leave a first added/primary status on the 204.x.x.x/30 address. After a test reboot, it was back in the broken state.
We use a 10.x.x.x/29 to talk to radio bridge devices on links between towers. We run OSPF on routable.x.x.x/30 addresses to keep traceroute working from out of network. In this particular case, I don't have radio bridges on this interface. I'm just using the 10.115.0.1/24 as src address for managing devices that shouldn't be accessible from off network. I moved it to the loopback bridge for now. This problem is going to bite me somewhere as I convince mangement to allow me to use MikroTiks.