lordzar -
Not to be critical - by why the heck do you want to bridge everything anyway?
NORMALLY you would give your servers a private IP and use 1:1 nat’ing for them on a particular private IP block, and then do masq for clients from another private IP block. And/or if you HAVE to have public IPs on the servers then carve out a smaller - say /27, out of your class ‘C’ (/24) network, put a public gateway IP on ether3 and point all of your servers to that IP with an IP in the /27 that you defined a few moments ago…
Then you have a purely routed/nat’d network that does not need bridging at all…you can also use all the ‘power’ of ROS to filter, nat, policy route, everything…
Just a suggestion - that’s all - as it IS YOUR network…
That setup describes the “drop-in” mode of a Watchguard Firebox. Not my favorite setup; but it is sometimes useful. Especially if you need to have a device on a public IP (as in actually in the device and not NATd into it), behind a firewall and the ISP controls the subnet. It also allows the firewall to be inserted/removed without reconfiguring the devices on static public addresses.