DHCP "static" assignments

[Peter Rathlev]

On Thu, 2013-08-08 at 08:08 +0100, Simon Hobson wrote:
> Personally I think the MS way sucks (as does the lack of capability in
> their software) - but I can see that for someone "brought up on it" it
> will seem natural.

I agree wholeheartedly with you regarding MS DHCP capabilities, but one
use case for "their" method that comes to mind is if you have e.g.
printers or other devices where a "static" address (still served via
DHCP of course) makes sense, but where you don't want the hassle of
pre-registering the device. In that case it would be nice to just have
the device pull any address and then reserve this afterwards. You can do
this with OMAPI of course, but OMAPI is still black magic for many
people. Maybe some kind soul would one day write a few scripts that
makes it easy to accomplish simple tasks like "dhcp-reserve-address
[<mac-address>|<ip-address>]" or something like that.

As an alternative, which is what we are trying to convince the printer
staff to accept, is to just ignore everything about static assignments
and let dynamic DNS do its thing. Every reference to the equipment, e.g.
printers, would then be by name. That would make the specific address
assigned irrelevant. Unfortunately some management suites for printers
are unable to use DNS names, we're told. The fault is clearly with those
tools, but sometimes you have what you have.

You could also rely on ISC DHCP following the RFC and (mostly) always
assigning the same address to a device over time. On the other hand I
have seen our servers not do this; we have had clients whose lease-time
was exceeded doing a DISCOVER and receive a new address, even though the
lease file contained an expired lease for the same MAC address / client
identifier. If the lease-time is long enough this doesn't happen. We
never found out exactly why, but this change of addresses is contrary to
both what the ISC DHCP documentation says will happen and what the RFC
says should happen if I read them correctly.

-- 
Peter