DDNS handling of "dumb" DHCP clients

Uwe Meyer-Gruhl dhcp_email at congenio.de
Thu Aug 9 16:28:23 UTC 2012

> Right, but you can have the DHCP server assign a static address - so from the
> device's POV it's just a regular DHCP lease (while not really a lease for
> dhcpd).  When using the DNS name in the corresponding host declaration all you
> need is change the DNS record, restart dhcpd and on next request the deivce
> will get the new IP.  Nothing the client needs to know.

You noticed the word "static" there? The whole point of DHCP is to 
assign dynamic addresses. With your approach, I do not even need DDNS at 
all, as the devices DNS entries were static as well.

As I already wrote, we can and do use "fixed-address" now - but these 
addresses are derived from a database and have to be maintained for 
printers that are in a shared network in which pools have to be altered 
because of NOC contraints. I am not talking about three or four of these 
types here. Such modifications are done almost daily here.

For clarity, what I want, is:

Clients which are given dynamic addresses from a pool and for which the 
DHCP server does the DDNS updates WITHOUT the need for the clients to 
send the FQDN option (because they are "dumb").
The DHCP server can be provided a "host" configuration entry which 
specifies the client's DNS name (no matter if "ddns-hostname", 
"hostname" or by using "use-host-decl-names" is being used), however 
using "fixed-address" is not an option, for the named reasons.

And exactly that is what the documentation says ISC DHCP can do - alas, 
specifically only for deprecated "ad-hoc" DDNS style.

More information about the dhcp-users mailing list