Hi Bert,
Your concern is very valid, apparently xp had these in some
configurations and win8 will have these everywhere (?).
Are there any flags on these dns addresses? Filtering by those would
be easier than hardcoded list/mask.
If I read the draft correctly, only the host and not user processes
should use these, and that timeout out to be very short, since these
are all site-local, it is even said that after power-on host should
use timeouts 3, 6 a and 9 seconds, presumably timeouts are much
shorter under normal conditions.
Daniel, who is the authority when it comes to c-ares name resolution?
In any case your proposed patch should to exclude fec0:0:0:ffff::1,
fec0:0:0:ffff::2, and fec0:0:0:ffff::3 specifically rather than whole
IANA-assigned site-local block (deprecated).
Dima
On 17 August 2012 02:25, Bert Belder <bertbelder_at_gmail.com> wrote:
>> http://tools.ietf.org/id/draft-ietf-ipngwg-dns-discovery-03.txt
>>
>> section 2 tells what those are. not sure what c-ares whould do with these
>> though.
>
> Well, if these addresses would generally (or even, ever) work then maybe c-ares should try to use them.
> However, before we used this patch in node we would every now and then get a bug report about IPv6 dns resolution being super slow. After we included this patch, I've never had any complaints.
>
> - Bert
>
>
Received on 2012-08-17