DHCPv6 and MAC Address inclusion

Ted Lemon Ted.Lemon at nominum.com
Tue Jan 24 22:52:58 UTC 2012


On Jan 24, 2012, at 4:34 AM, <scott_stone at trendmicro.com>
 <scott_stone at trendmicro.com> wrote:
> Interesting - I took the approach of learning about this through the documentation for the specific implementations that I was using (isc dhcpd, rh/centos DHCP client), and other sources including other network people that I know (sadly few of whom are super familiar with IPv6) - looks like I didn't search the correct terms, however, because searching the terms that you put forth here leads me to RFC 4193 defining fc00::/7 (this hasn't been deprecated yet, has it?)

Site local was deprecated in favor of ULA.   The main difference being that ULA doesn't have to be restricted to a "site."

> As for 'intent', I forget the original source from which I read that (I thought it was an RFC) which described exactly that case of the laptop using wireless/wired as the reason behind the host-specific DUID.  A couple months back when I posted on this list about this issue I was at the time led to believe that what I wanted to do wasn't going to be do-able (not saying that anyone explicitly said that, that was just the conclusion I eventually came to after reading everyone's replies, many of which were of the "this argument has come up before, read <this archive>" which is I believe where I got the information that I was citing... anyway it was a few months ago and I have many projects on my plate.

The two interface, one address scenario is a valid use case, but the client would have to send the same IAID on both interfaces to get that functionality.   This isn't the intended behavior of all clients; it's just a behavior that we imagined might be desirable in some situations.   I don't know of any client implementations that actually support this behavior.

> Good to know that you are in the know on this/have been for a while - hope you don't mind if I bounce further questions off of you (via this list) so that I'm not getting the information second-or-third-hand.

I definitely do not mind.

> So, then, perhaps I do not quite understand how the IAID is generated or sent to the client - looking at how isc-dhcpd works, I don't see how to programatically determine the IAID of an interface from any well-known set of information that I would have ahead of time (can that be done?) - or, even, how does the client generate an IAID/is there a specific format for it/how do I find out what it is/etc?  Is this even possible with the ISC+RH server/client setup?

You should be able to get the IAID out of the IA option.   I don't know how that's done in the ISC DHCP server, however—you'd have to talk to one of the ISC guys about that.   The IPv6 code is all stuff they added after I stopped working on the ISC server about ten years ago.   :)




More information about the dhcp-users mailing list