[Outages-discussion] What makes "dig +trace" start with 8.8.8.8? (was Re: [outages] Ebay dns record)

[Patrick W. Gilmore]

On Oct 01, 2012, at 21:24 , Jay Ashworth <jra at baylink.com> wrote:
>> From: "Jon Radel" <jradel at vantage.com>

>> On 10/1/12 4:23 PM, Jay Ashworth wrote:
>>> When I dig +trace, I do not expect GDNS to be the first hop, and the
>>> end records to be the second. I'm not sure what is causing that, but
>>> I'm relatively sure it's something doubleplus ungood. See the two
>>> traces in my original posting, if it's not clear what I'm
>>> complaining about.
> 
>> I'd expect that if the root hints file contained GDNS rather than the
>> actual roots.
>> 
>> At best I'd have mixed feelings about putting my DNS into Google's
>> hands to that extent, but I'm having trouble deciding what the moral
>> distinction between that and forwarding everything to GDNS really
>> is...I'll have to ponder that.
> 
> Well, it breaks dig +trace rather well.
> 
> I'm assuming it's because my edge NAT router is set to forward there rather
> than FiOS's resolvers (though I didn't set that box up)... but that sort
> of means that that Westell has to be *transparent proxying* DNS, rather
> than just resolving to GDNS, I would think -- otherwise it shouldn't affect
> dig +trace, which is running on my OpenSUSE12 laptop, and whose hints file
> would be the stock one.  I think.

Wow.  There are a ton of assumptions in there, many of which are easy to check.

I expected better of you, Jay!

-- 
TTFN,
patrick