ISC DHCP Failover in DHCPD v3.0.5

Alon Livshits alonnetmail at
Thu Jul 12 17:32:40 UTC 2007

I'm trying to implement DHCP Failover with DHCPD v3.0.5.  I had some
questions about the split statement function.  From what I understand, the
split index will specify how frequently a primary DHCP server will answer
versus the secondary from its pool of addresses.
For example, a pool of 254 addresses and a split of 128 would evenly
distribute client requests to both servers.  The pool of IP addresses would
also be split between the primary and secondary servers with each pool being
127 addresses.

Now from my understanding the split should not affect the pool distribution
which should always remain as close to 50% between the two servers.  Following
this logic what would a split setting of 0 and 255 do?

If I were to guess, a split setting of 0 would force the secondary server
answer the majority of the DHCP requests until all of the IP addresses in
the pool (127 addresses) are used up.  After this occurred, the remaining
127 addresses on the primary server would never be used because the split
statement is effectively saying that the secondary server must answer all
the time.

Moreover, I would also assume that the reverse would occur if the split
value is set to 255.  In this case the primary will answer all of the times
until its pool of 127 addresses are used up.

If anyone could provide me with more insight on how the split mechanism
works and let me know if the above assumptions are correct, I would greatly
appreciate it.

More information about the dhcp-users mailing list