I have three domain controllers. They all are in their own subnet. One of the three however, is also the DHCP server** and instead of using a DHCP relay it connects directly to each subnet** so that's the network with the other two domain controllers and around a dozen more.
In DHCP though, the specified server addresses are those in the DC subnet so for the most part clients cross subnets to query DNS and it all works.
The problem comes when a client queries the A records for the multi-homed domain controller itself: the domain controller responds with any of its addresses instead of evaluating the client's address and responding with the interface address it has on that subnet. It's a problem because it means that I must add an unneeded gateway on each interface further complicating firewall rules.
On top of that, in the DNS chatter among domain controller I often see traffic from the multi-homed DC traversing subnets to reach the other DCs instead of using the directly attached interface on the DC network. Why!?
Why is it not self-aware? How can I make it so? Other DNS servers aren't this dumb which and Windows Server (I recently updated to 2022) is supposed to be location-aware. I realize this are policies that have to be configured but that's for public networks, isn't it? BIND, for instance, does this automatically for directly attached networks--or even statically routed, so, I assume I must've overlooked some setting.
I appreciate your advice on this. I can't use a the DHCP relay on the interVLAN router because it had been crashing, I have a switches with relays too but they automatically route they require complex rules so they don't route the traffic (creating routing loops) that I haven't had the time to learn/create.
Thanks!
**: (for all subnets where one is required)