On Fri, 12 Jan 2018 06:15:44 +0000 Francis Dupont <fdup...@isc.org> wrote:
> I found at least 2 easy ways to fix your problem: > - the first one is to have a second relay which fills the link address: > the code skips all link addresses which are undefined or link local > as it wants to be able to identify the relay. BTW the interface-id > option must be added by the first (closest to client) relay. This is the current plan, which I may have mentioned. > - the second is from the ISC DHCP code which has a comment about L2 relays: > the idea is to use the receiving interface and client classes matching the > interface-id option. With 1.2 you have to know the number of relays > (1 in your case) to evaluate the right relay option > (syntax: relay6[0].option[18].hex == 'a/b/c') and to restrict each > subnet to a class matching an interface-id value. This is interesting enough that I might try it even if I don't ultimately use it. > With 1.3 you get the two ways to count relays (1.2 / positive from > the server, negative from the client). With 1.4 you'll get classes > for pools which is far easier to use in particular when migrating > from ISC DHCP... Well, this is my first DHCP server, so I don't really have a history to fall back to, but this still sounds like an interesting feature. Thanks! --TimH _______________________________________________ Kea-users mailing list Kea-users@lists.isc.org https://lists.isc.org/mailman/listinfo/kea-users