Bug #123

The issues here are fairly large. Because isc-dhcp is targetted at large providers (100s of queries/sec), and not the home (20 queries/day), and the most current isc-dhcpv6 includes all of bind9, I ripped it out and at least temporarily replaced it with just using dnsmasq as a dhcp server, where it has full GUI support, and is quite smaller.

0) There is no need for dhcpv6 inside the home. Stateless autoconfiguration + dhcpv4 leveraging the generated EUI-64, suffices for 99% of all ipv6 capable devices.

1) The dnsscripts that leveraged dhcpv4 addresses and turned them into stateless autoconfig ipv6 addresses, and inserted them into bind were dependent on evan’s dual domain system, and not quite as view capable as they needed to be, where you would typically want to publish SOME addresses, but not all

2) They didn’t speak dnsmasq’s syntax for it’s plugin script, but isc-dhcp’s.

3) They didn’t pick up on the dynamically assigned 6to4 or 6in4 addresses generated by openwrt.

4) In the shorter term, I’d like a dhcp server that ran out of xinetd. That would allow for locking out misbehaving, worm-ridden, devices from even getting on the network. I found one that looked pretty good.

http://inetdxtra.sourceforge.net/

5) In the longer term, merging the functionality of AHCPD (the mesh network ip address distribution daemon) with dhcpv4 and dhcpv6 seems the right way to go, to get a common lease table and address distribution method that works both with conventional systems and mesh ones.

This work is already taking place elsewhere. I have no idea what schedule it is on.

AHCPD:

http://www.pps.jussieu.fr/~jch/software/ahcp/

As I don’t think any of these issues can be fixed in the time remaining, I’m scheduling these for the next cerowrt release, and further discussion of approaches.