Failover and pools with single machines
tina at zool.unizh.ch
Fri Dec 15 14:01:24 UTC 2006
Am 14.12.2006 um 16:52 schrieb Simon Hobson:
> Tina Siegenthaler wrote:
>>> Failover won't work with such small pools. However, since you will
>>> only have one client/pool then you could simply replicate the
>>> and NOT use failover. The client will get the same address
>>> of which server it took the lease from, and no other client will
>>> the same address from the other server.
>>> You would still need to use failover on the general use pools
>> Does this mean I can just use the same config file (or that part of
>> it with the single-host pools) on 2 dhcp servers at the same time?
>> Won't they get each other in the way?
> Ordinarily, having two independent servers offering the same
> addresses would be a disaster looking for somewhere to happen. It
> will work in this case because you have pools with exactly one
> address which can be leased to exactly one client - so you can't have
> the same address offered to two different clients.
>> So, if I understand this
>> correctly, a host will issue a broadcast and then take the IP address
>> from the first dhcp server that responds. And it doesn't matter if
>> there is a second dhcp server which would also be able to assign the
>> host an address, since it's already got one. Right?
> Near enough - technically the client is allowed to see what offers it
> gets and choose between them (how is not defined, but for example it
> could choose the one that gives it the most options it asked for). In
> practice, I don't think anyone has found a client that doesn't just
> accept the first offer !
OK, I set the servers up that way and it seems to be working fine
(well - running for five minutes now, but I hope it'll stay that
way). I just have one more minor question about the failover peer
failover peer "abc": Is "abc" the name of the peer of the server, or
the name of the server itself? Somehow this isn't clear from the
More information about the dhcp-users