[Dnsmasq-discuss] using DHCP with virtual interfaces

[Tom Metro]

Simon Kelley wrote:
> The behaviour WRT to MAC addresses and client-identifiers is mandated by
> the DHCP standards, which don't allow the server to assume identify
> between two hosts with different MAC addresses or client-ids based on
> the hostname it supplies.

Ah, OK. I wasn't aware of that. I suspect most casual users of DHCP 
wouldn't be either. Superficially the two attributes seem like they 
should be interchangeable (in terms of their impact on the IP address 
selection decision logic), and if I recall correctly the man page 
doesn't say otherwise.

I'd like to offer a documentation patch to correct for that, but I'm not 
sure if tacking another paragraph onto the dhcp-host documentation is 
the best place for it. I'm wondering if something more like a "theory of 
operation" section, which could then document the full set of steps 
Dnsmasq uses to get from discover to offer, would be the better way to 
go. Sure, it's going to be largely redundant with what's in the RFC, but 
  it'll be documenting things specific to the Dnsmasq implementation as 
well.

I see the man page already has a few paragraphs near the end covering 
network-id, which is a portion of the DHCP decision logic. This patch 
would probably involve reorganizing the notes section, renaming it 
something like OPERATION, giving DNS and DHCP their own sub-headings, 
and putting the signal behaviors under their own SIGNALS heading.

Would this approach make sense? I could take a stab at a first draft for 
it, using the RFC as guidance, but I'd be dependent on you to refine it 
to be Dnsmasq-specific.

  -Tom

-- 
Tom Metro
Venture Logic, Newton, MA, USA
"Enterprise solutions through open source."
Professional Profile: http://tmetro.venturelogic.com/