Arend, sup! While I agree that the task itself is easy, the problem is in our large codebase and templates. We currently expect that MAC address exists, there might be dragons. However you can start trying, just remove this line from nic/base.rb file:

    validates :mac, :presence => true,
              :if => Proc.new { |nic| nic.managed? && nic.host_managed? && !nic.host.compute? && !nic.virtual? }

And start testing this.

Hey Lukas! Thanks for your input, it has proven to be very helpful! (sorry for getting back so late on this subject)
I've managed to play around a bit, and the following code seems to do the trick:

validates :mac, :presence => true,
              :if => Proc.new { |nic| nic.managed? && nic.host_managed? && !nic.host.compute? && !nic.virtual? && nic.provision? && (nic.subnet.present? || nic.subnet6.present?)}

The "beef" is in both checking for it being a provisioning interface (because, well, provisioning means DHCP, DHCP means MAC-address) but also checking if there's a subnet set at all (assuming that a subnet always needs a MAC-address... could probably be filtered down further in case no DHCP/IEU-64 is used)

However, seems like a lot of checks are happening for that property? Is this the way forward, or should I not split this up in a separate validation function?

Another thing, as you mentioned, the templates regarding networking (luckily, as far as i can tell, a snippet) needs to be adjusted to take into account checking for a MAC-address

(also, writing a bit before my turn because the tests are still running; the above is based on a manual test I've performed)

Kind regards,
Arend Lapere

The change looks good, no need to split this into another method, just file a PR and maybe an extra test for this case would be required.

Templates need to be changed only when host.mac returns nil, which is only if the provisioning interface has mac set to nil. And you can go ahead and patch Foreman core first and only then make necessary changes in community-templates repo.