We have a system where we authenticate DHCP clients based on DHCP option 82. This allows us to set a static IP for a service, which is not reliant on the CPE MAC or some other value which might change.
On our RADIUS server, we have taken advantage of the Framed-Route value so that we can assign additional subnets to be routed to the customer. The Mikrotik DHCP server takes the Framed-Route value and creates a "dynamic active static" route for those subnets with a gateway of the DHCP client IP. It's a fantastic feature that makes for less manual configuration.
But only for non-Mikrotik CPEs. Mikrotik CPEs lose their default route, and only add routes for those Framed-Route values.
As far as I can gather/remember (it's been a while since I investigated), the Mikrotik DHCP server also sends those Framed-Route values on to the client in the DHCP offer, as part of DHCP Option 121. As Chupaka eloquently stated in another post:
The issue is that 'other brands' does not follow RFC. RFC3442 clearly states: "If the DHCP server returns both a Classless Static Routes option and a Router option, the DHCP client MUST ignore the Router option."
From what I have seen, an all Mikrotik environment would end up with those additional routes being routed to the CPE, and the CPE would only add (broken) routes for those subnets.
My apologies if I have made any mistakes in amongst this - it's been a while since I investigated this behaviour, and only thought to post here after discussing a recent RouterOS release.
Has anyone else seen this behaviour, or gotten this working the way we intend to make it work?