Mac OSX 10.7

[Brandon Applegate]

On Fri, 22 Jul 2011, Tore Anderson wrote:

> * Sander Steffann
>
>>> As in many things, Apple doesn't work this way. (In this case that's
>>> good, because the /etc/resolv.conf behavior is crap.) Since 10.6,
>>> Mac OS X uses mDNSResponder for all DNS lookups, and appears to have an
>>> algorithm that cycles between resolvers based on reachability:
>>>
>>> http://support.apple.com/kb/HT4030
>>>
>>> One can only hope that IPv4 and IPv6 servers are treated equally, but I
>>> don't know for sure. Maybe someone with a working dual-stack resolver
>>> and Mac OS X 10.7 can do some measurement and let us know?
>>
>> I think Tore Anderson had some info on that… Tore?
>
> There was an informative post over at Apple's IPv6 mailing list
> explaining the new behaviour in OS 10.7:
>
> http://lists.apple.com/archives/Ipv6-dev/2011/Jul/msg00009.html

So sorry to potentially tie this in a knot.  I don't know if I'm 
understanding.  I read the Apple mailing list thread above, and it seems 
to me that connections to v4 vs. v6 destinations are still essentially not 
predictable.  If the determination is made based on cached RTT, and the 
tiebreaker ( ?  same RTT ?) falls back to 3484 - how is this different or 
'fixed' ?  I know 'fixed' is a relative term.  My personal meaning echoes 
Jeroen - 'always prefer v6 over v4'.

The SSH example is the perfect test.  I checked Chrome only on my wife's 
machine (latest SL) and noticed connections always seemed to land on v6. 
I didn't check SSH.

--
Brandon Applegate - CCIE 10273
PGP Key fingerprint:
7407 DC86 AA7B A57F 62D1 A715 3C63 66A1 181E 6996
"SH1-0151.  This is the serial number, of our orbital gun."