Most of the fedoraproject.org hostnames (including wildcard.fedoraproject.org) have two AAAA records. Neither of those servers are actually responding on port 80. So if a client has working IPv6 connectivity, every HTTP request to a fedoraproject.org website takes 120+ seconds to timeout and fall back to IPv4.
Reproducer: Find a computer with global IPv6 connectivity, try to go to any fedoraproject.org website, watch the page not load for a very long time.
The ideal resolution would be to fix those two servers so they actually serve content over HTTP on those IPv6 addresses / tcp/80. If that's not possible, at least pull the AAAA records from DNS so IPv6 people don't have to suffer. :)
I'm proxying this report for rwg who is on the CC for this ticket.
% host -t AAAA fedoraproject.org
fedoraproject.org has IPv6 address 2001:4178:2:1269::11
fedoraproject.org has IPv6 address 2610:28:200:1::fed0:1
% nc -6vv 2001:4178:2:1269::11 80
nc: connect to 2001:4178:2:1269::11 port 80 (tcp) failed: Operation timed out
% nc -6vv 2610:28:200:1::fed0:1 80
nc: connect to 2610:28:200:1::fed0:1 port 80 (tcp) failed: Operation timed out
(And just to prove this machine actually has global IPv6 connectivity...)
% nc -6vv www.google.com 80
Connection to www.google.com 80 port [tcp/http] succeeded!
% nc -6vv www.ntt.net 80
Connection to www.ntt.net 80 port [tcp/http] succeeded!
% nc -6vv www.he.net 80
Connection to www.he.net 80 port [tcp/http] succeeded!
puppet template had extraneous characters. mmcgrath and dgilmore fixed it.
to comment on this ticket.
Copyright © 2014-2017 Red Hat
3.10.1 — Documentation