Bridge MAC and Static DHCP

Our network has a couple RB751U-2HnD as CPE. Wireless backhaul to our router, and all interfaces are bridged. We serve DHCP not only to client devices, but to the CPE as well, using a static DHCP lease.

The problem is, if the MAC of the bridge changes, the static DHCP entry no longer matches, and the device gets a different IP. And–at least for some versions of ROS–the MAC changes depending on which Ethernet ports are in use.

If ROS uses the lowest MAC of all ports, that would be consistent. If it uses the lowest MAC of all active ports, then the value changes each time someone connects or disconnects a line to ether1. That means the IP changes the next time the license is renewed, and Dude reports it offline.

The ideal solution to this problem would be for the MAC of a bridge to be based on all ports, whether active or inactive.

Until that happens, I’m thinking about setting the bridge’s admin-mac to that of the lowest port (typically ether1) so existing static rules keep working. Are there any caveats to this approach? If anyone else is doing this, could you post an example of the bridge settings?

Until that happens, I’m thinking about setting the bridge’s admin-mac to that of the lowest port (typically ether1) so existing static rules keep working. Are there any caveats to this approach? If anyone else is doing this, could you post an example of the bridge settings?

That’s exactly how should be done and why that setting exist, proceed without fear…

Okay, it’s working as expected (after setting auto-mac=no). But why would MT see any advantage to having the MAC address of a bridge change according to which ports are active?

Guess the alternative is either to use an active port mac, or use a completely “made-up” one; the non active port mac shouldn’t be used as it’s not active but already present in the system and would cause issues…

Not sure what issues using the MAC of a non-active port would cause; that’s exactly what I just did with the admin-mac setting.

I set fixly the mac of the first port of the bridge. Have not seen any problems with it regardless the port is active or not.

I mean, if mikrotik is doing that, I guess the reason could be in that direction…

I’m wondering if this behavior is intentional, or if it might have been an inadvertent change at some point. The manual reads simply, “the smallest MAC address will be chosen automatically”; it makes no mention of interfaces having to be active. I’ll see if someone at support can provide the history/rationale, and I’ll summarize it here in case anyone is interested.