I work in an enterprise environment. We have many locations, each with many subnets, and both Linux and Windows everywhere. DHCP servers are centrally located for everything (Or rather, we have 2 each handling half the scope's IPs).
Now I want to get prepared for a fullblown recovery in all locations. So basically I want the DHCP servers to stay in place, but I want multiple PXE servers to be queried. I can see that I can add n different IP helper addresses for the PXE servers. So I should be able to get the DHCP Request to all of them.
In many cases I assume I would spin up a local PXE server, and have all subnets use its pre-allocated IP address as one of the IP helpers. But until that one is running, I would like the location to use the central PXE server.
But, how do I configure the PXE servers to not hand out IP addresses, but only bootp server and filename ? I assume the PXE server would be running both DHCP and TFTP, and I would like the PXE DHCP server to send only option 66+67.
It is difficult to find an explanation of this process. Supposedly a PXE booting device will keep sending out DHCP Discover packets after getting an IP. But how do I identify the packets where I need to answer with 66+67 only ? And how do I configure that ?
In the end, I want both Linux and Windows PXE servers centrally.. So if one platform is hit, it can PXE boot from the other.