Since years I am wondering what is correct: by defining ranges in the DHCP section (start address - end address), shall the fixed leases below in the config page be within these ranges or out of them?
I tried both and both works.
Note: Make sure that any IP used for a static fixed lease does not come from the ‘dynamic’ pool of IP addresses
Because this question comes up from time to time again, once more a short description of IP association ( in IPv4 ):
In a network with DHCP server there are three ways of chosing an IP address
- the client device choses its address himself ( static address )
- the DHCP server chooses an address out of a set of addresses ( dynamic lease )
- the DHCP server assigns an IP based on a set of (MAC,IP) tuples ( fixed lease )
Because ethernet addresses devices by MAC, the assignment of IPs to active devices to MACs must be inambiguous. Ranges from 1. and (2., 3.) must not overlap.
The DHCP server tries to achieve this by checking the existance of a device with a selected IP.
The selection of an IP offered to a device, identified by MAC, is roughly as follows
is there a definition for MAC in the set of fixed hosts, if yes offer this IP
is there an association MAC<-->IP in the active dynamic leases, if yes offer this IP
select the 'oldest unused' IP from the dynamic pool ( all addresses {<start>....<end>} ), check by 'ping' for usage, if no device answers offer this IP, otherwise check with the next IP.
Because of this check for usage, it is urgent that the three sets of IPs are disjunct.
The DHCP server must be sure to use a free IP. An IP potentially used by some device not active must not be chosen.
This condition can be satisfied by the disjunct selection of the three IP sets.