different devices getting the same ip address

Brian De Wolf bldewolf at csupomona.edu
Thu Dec 1 21:31:34 UTC 2011


On Wed, 30 Nov 2011 10:44:10 -0800
Simon Hobson <dhcp1 at thehobsons.co.uk> wrote:

> Okeanos.net AS wrote:
> 
> >>Because the devices are using the same client ID, they are the same 
> >>device as far as DHCP is concerned.
> 
> >Is the decision on assuming that the interfaces are on the device 
> >solely based on the uid?
> >Is that the default for all versions of the isc dhcp server?
> 
> Yes, not only is it the default, but I don't think there is a way to 
> change it either. IIRC there was talk of making a config option that 
> would allow it, but I don't think that's made it off the wishlist yet.
> 

If you're willing to run a patched dhcpd, I wrote a patch that adds an
"ignore-client-uids" option that can be enabled where you want to
discard UIDs back in July:
https://lists.isc.org/pipermail/dhcp-users/2011-July/013440.html

Note that this is implemented by completely disabling parsing of UIDs
for the clients where it is enabled.  So, if you enable this option for
existing clients, they aren't going to match the leases they previously
had (without nuking the UIDs from the leases file or something).

So far we've been running this patch with it enabled in a handful of
subnets where dualbooting occurs without issue.

I've sent it to dhcp-suggest@ but haven't really received a response
(from any humans, anyway). I can understand why, as the ideal
implementation of this feature would record the UIDs and just not use
them when matching leases. However, that ideal implementation would be
much more invasive, and would have to come from someone much more
involved with dhcpd than I.



More information about the dhcp-users mailing list