On 9/20/2025 03:06, Bjoern A. Zeeb wrote:
On Fri, 19 Sep 2025, Andriy Gapon wrote:[multihome steup] What you are describing is in no way special to DHCP. Even a manual configuration would have the same issue, wouldn't it? For IPv6 there exist a set of RFCs which have ideas on how to deal with multi-homing. The one main issue (routing and source address selection sorted), is and remains DNS as we have no default way to pick up resolvers on a per-interface or per-domain setting. You'll find that per-interface doesn't really work as you'd have to know which path you go before you do the DNS lookup. But chosing an upstream DNS with the wrong source address often won't work. At least here ISPs won't allow you to use their resolver if you are not coming from their IP range. The answer then really is to run a local resolver independent on upstream for as long as that is feasible and working (*). In that way the DHCP approach (which I think should have worked with multiple IF just fine to merge a resolv.conf) isn't that bad. Use the DNS where your default route goes. My 0.0005 cts /bz
My answer has /always /been to run a local resolver.The reason is not just what you note; it is /also /that there are frequently, on networks I'm responsible for, two different resolved addresses for a given resource depending on whether it is outside or inside the local network, at least for IPv4, because of NAT.
Port forwarding at the gateway takes care of access from the outside however you do want BIND's "rpz" or unbound's equivalent for accesses that come from inside the gateway.
-- Karl Denninger k...@denninger.net /The Market Ticker/ /[S/MIME encrypted email preferred]/
smime.p7s
Description: S/MIME Cryptographic Signature
