[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: "DNS-4to6 automatic mapping" - No more DNS-ALGs & 6then4 DNS queries



I guess the concept of just working is overrated also.

I do not think we have time to debate the finer points of eeking operational
overhead by asking vendors to not query AAAA just because a v6 stack/default
route is not enabled.... versus telling vendors where dns servers are broken
and fixing things to pave the way for adoption of v6 wholesale.

I am sorry you see me as a one size fits all mentality associated with
OpenBSD.  I simply mention it as one os vendor that does the opposite of
what you suggest.  I would be surprised if there are not others.

To me, removing AAAA queries based upon conditions adds more complexity
and does not provide a consistent user experience and is a step backwards
rather than a step forwards.

You will, however, get no complaints from me regarding BCP of not querying
A6; this was deprecated years ago...

Think about it, you want userland to query the kernel for v6 support and
further for a v6 default route before each potential AAAA query?  This seems
rather complex and fragile and has overhead that to me is worse than
unconditionally sending the AAAA requests.

You want a user to note that a host has no v6 record, on a v4 only node,
yet suddenly discover there is a v6 record on a v6 enabled node?  How insane
is this?  Suddenly starting to pull wool across eyes is what this sounds
like.  I much prefer to verify my dns is sending out v6 results even from a v4
only node than to alter something that I (and I suspect I am not alone)
expect to work today.

How can asking vendors to add more special casing to v6 code to hide v6
rather than enabling it by default be a step forward?

I for one, regardless of my os of choice, see the scenario you propose
to be a giant leap backwards.

Thanks,
-- 
Todd Fries .. todd@fries.net

 _____________________________________________
|                                             \  1.636.410.0632 (voice)
| Free Daemon Consulting, LLC                 \  1.405.227.9094 (voice)
| http://FreeDaemonConsulting.com             \  1.866.792.3418 (FAX)
| "..in support of free software solutions."  \  1.700.227.9094 (IAXTEL)
|                                             \          250797 (FWD)
 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
                                                 
              37E7 D3EB 74D0 8D66 A68D  B866 0326 204E 3F42 004A
                        http://todd.fries.net/pgp.txt

Penned by Gert Doering on 20071129 13:51.28, we have:
| Hi,
| 
| On Thu, Nov 29, 2007 at 06:19:25AM -0600, Todd T. Fries wrote:
| > Make all the distinguishing you wish.
| 
| Thanks for your well-reflected views on a one-size-fits-all reality :-)
| 
| > The bottom line is that OpenBSD userland unconditionally enables IPv6
| > regardless if the kernel supports it or not.  If there is no v6 default
| > route, v6 fails quickly and seamlessly back to v4 (telnet, ftp, ssh, etc)
| > and behaves sanely.
| 
| Aah, OpenBSD.  Now, of course, OpenBSD is always right, and the single
| point of reference on how operating systems should work.
| 
| The initial topic on this thread was "can we avoid extra DNS queries that
| have no value, but may cause operational issues".
| 
| What happens *afterward*, whether or not the connection fails "quickly" 
| or "slowly", is really of not much interest for the question under
| discussion.
| 
| Gert Doering
|         -- NetMaster
| -- 
| Total number of prefixes smaller than registry allocations:  110584
| 
| SpaceNet AG                        Vorstand: Sebastian v. Bomhard
| Joseph-Dollinger-Bogen 14          Aufsichtsratsvors.: A. Grundner-Culemann
| D-80807 Muenchen                   HRB: 136055 (AG Muenchen)
| Tel: +49 (89) 32356-444            USt-IdNr.: DE813185279