I've just been on a 15 day business trip. Three meetings/conferences (OpenSSL was very well done), five cities and hotels and many trains and airports.
While my Android (now GrapheneOS) phone does relatively well, despite a
number of captive portals thinking that I should COPY some URL into my main
browser, totally defeating the privacy properties of the captive
browser... my (Ubuntu 24) laptop does not do as well.
I have a difficult relationship with systemd, and systemd-resolver has just
been terrible. I've wound up uninstalling it because it breaks more often
than it works.
Mostly, I'd rather run unbound locally (with DNSSSEC), and adjust
what forwarder it uses, and this has worked in the past.
I gave that up, perhaps during an upgrade, but going that route does not
fundamentally change the problem. It even makes it worse to see if the local
cache can answer existing names, but can't resolve new ones.
So then HTTPS goes wrong. My GNOME calendar likes to go wrong in the
background, putting up a series of dialogues.
====
But, the unseeing pain in the ass is. Let me show you;
%ifconfig docker0
docker0: flags=4098<BROADCAST,MULTICAST> mtu 1500
inet 172.17.0.1 netmask 255.255.0.0 broadcast 172.17.255.255
%ifconfig -a
br-52b5bb71a559: flags=4098<BROADCAST,MULTICAST> mtu 1500
inet 172.18.0.1 netmask 255.255.0.0 broadcast 172.18.255.255
and bahn.de loves to use 172.18.0.1 or 172.18.0.2 for default route/DNS,
and a bunch of other portals will use 172.17.0.1.
(Can you see the problem? Yes, "ifconfig foobar down" solved the problem
until I wanted to do some docker or LXC things)
And sometimes they use those IPs even when the local wifi is on some
completely random other subnet.
Can't lookup connectivity-check, can't reach their (somtimes split-horizon
0nly) portal, can't enable connectivity.
Sigh. IPv6 ULA. Please? Please.
The problem of "eth0" and some local bridge conflicting in IP addresses is
unsolvedable by the IETF, but I think that the DNS conflict is perhaps
solvable. Maybe not easily deployed, because few of these operators know
CAPPORT, but I think that it's growing.
I have seen, even implemented, situations where 8.8.8.8 is set up an alias on
the local resolver, which can work for Do53, but not for DoT. Hey, I often
resort to visiting http://8.8.8.8 to kick off the portal interaction when I
can't figure out what else is broken.
My idea is to reserve an IPv4/32 from some utility prefix.
Something not routed, and never will be. *MAYBE* the 100.64.0.0/10
is the right place.... I'm sitting on the fence here.
Effectively, this v4 is an anycast "find me a Do53" resolver.
Want Trust? Use DNSSEC. Want Privacy? Use it to find your DoT host.
--
Michael Richardson <[email protected]>, Sandelman Software Works
-= IPv6 IoT consulting =- *I*LIKE*TRAINS*
signature.asc
Description: PGP signature
_______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
