I was just hoping to iron out some issues I am having with setting up DHCP Failover with OES Linux.

For a long time we were using our Netware boxes for DHCP at our two main sites. Each site had two DHCP servers.
Without using failover (which Netware didn't have) or clustering, I split the pool of addresses into two even ranges and assigned one range to each server.
And that worked perfectly well.

A while back we replaced one of the servers with OES Linux and all I did was replicate the settings of one of the address ranges to the OES Linux side.
This was not perfect as I could not easily move address ranges between the two servers as they were now two seperate OSes and configs.

Recently we replaced the other box with OES Linux and now I have two OES Linux servers.
I could keep the two seperate address pools like I did in Netware, but I have learned about this new thing called FAILOVER!

One other thing to note about our environment, we use a number of VLANs. So each DHCP Service has multiple subnets, and some of the subnets have multiple pools.

I was reading over the documentation and it seems to have a few holes. So here are my questions:

  1. If, like in the examples, I have a subnet with a single pool that is defined on both servers and failover is working correctly. Do both server respond to DHCP requests or is it a Master/Slave relationship?
  2. When setting up the TCP ports in the Failover Object, what an I suppose to put here? The documentation says I can use the same port for everything, but all the examples I have seen, people choose different ports. (for example I used port 6767 for primary and secondary in both objects)
  3. Is failover something that needs to be setup on every subnet and pool? Can I first setup failover on just one pool in one subnet before expanding the config to everything?