On 24/12/14 17:08, Frank Bulk wrote:
Except queries from 96.31.0.5 and 199.120.69.24 reliably return the AAAA
while queries from 96.31.0.20 do not. And we're all the same ISP, and in
the one case, from the same /24. I don't think Google is that granular. And
we do have good IPv6 connectivity.
Yes, Google are that granular. See:
http://www.google.com/intl/en_ALL/ipv6/statistics/data/no_aaaa.txt
...which currently lists:
96.31.0.20/32 # AS53347 United States
96.31.0.31/32 # AS53347 United States
Google have been, AFAICT, silent on the specifics of how they generate
their blacklisting. Presumably it's the fairly standard approach of
correlating a web-bug to a unique hostname embedded in the google search
page, received http requests, and received DNS queries. Google
undoubtedly then do some stats magic - that is their thing, after all -
and down-score resolvers which make the AAAA query but whose clients
don't finish the HTTP request, above some threshold.
We had persistent problems in the past with our resolvers appearing in
the Google blacklist, despite having excellent IPv6 connectivity. Google
were unwilling to provide us any details that would have allowed us to
identify the cause(s).
We eventually stopped paying any attention to it, and the problem went
away by itself.
Possibly there are one or more clients with broken IPv6 using your
resolvers, but without Google specifying the criteria which gets a
resolver blacklisted, it's hard to know.
Frankly, I wish Google would ditch the AAAA blacklist. It is hiding
problems that responsible operators would like to see and fix, just so
that Google don't lose eyeballs (and ad revenue) :o/
_______________________________________________
Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe
from this list
bind-users mailing list
bind-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/bind-users
