FreeBSD is initiating IPv6-only validation work [IPv6-only Systems]

Bjoern A. Zeeb bzeeb-lists at
Thu Jun 9 00:11:45 UTC 2011

On Jun 6, 2011, at 4:14 PM, Cameron Byrne wrote:


I wrote that reply the hour your email came in, but didn't want
to send this out earlier to not distract people too much.

> FreeBSD is initiating IPv6-only validation work

I think that's a less scary topic to some readers so I put it into
subject. I hope you don't mind.

> My own testing of the Windows 7 shows that, at it's core, it works
> well as IPv6-only for most "web & email" functions. Layering on
> applications like Skype, things start to fall apart.

Well, that really depends on what you are doing and targeting.

I have seen plugins crash a browser if there was no IPv4 anymore,
I have seen timekeeping software to log me an error every few minutes,
I have seen "v6ready certified" multifunction devices to force me to
pretend to print before being able to scan,  I have seen content
scanners falling over, network monitoring tools, web servers not
being ready, yadda yadda yadda.

I have seen a lot of confusing error messages when v4 was tried after
v6 and that hid the actual error messages making debugging a hell.

As an open source person, for example, I could imagine to see every out
of the almost 300k open source projects is hosting and which do
networking to eventually just work .. well say the more common ones of
them.  I'd love to see soooo much perl stuff to just work with IPv6,
I'd love software updates of my backup software, that worked well
without IPv4, not to break because the new major version *oops* is
still missing parts of the IPv6 code, jdk to not trouble anymore, ....

It's the software that runs in enterprises, SMBs, at ISPs, that is used
with the little tools that if failing put you into blind flying and ruin
your week, ...

I have spent some good time the last 6 months chasing some of these
things and not unsurprisingly some other things worked fairly great
out of the box.   But I also have a list of things to fix still
and I am sure we'll find more the more people are actually looking.

Note well -- this is different to "just no IPv4 address" which still
allows you to still do certain things on AF_INET sockets. This is
returning "Protocol not supported" in that case instead and that really
triggers another series of problems.

Also note well that all these things worked in dual-stack (almost)
flawlessly and we don't want to motivate end users to go IPv6-only at
this point (the much fun it would be;).  It's you people who'll force
them to eventually.  It's embedded folks that want it already in addition
to your mobile world.

Once you stop targeting the top-100 websites and what your parents or
children do and forget about that single voice/chat/file sharing program,
but look at things that run the world these days it's getting more

I am sure we'll get there, as we are trying to get users and content onto
dual-stack currently, but I don't want us to be there in only another
15 years. Thus starting early, as you and others have done, is the key.

Given the huge number of FreeBSD based things that ``run the Internet'',
and unsurprisingly end users, I hope that it'll help the commercial
vendors, protocol developers, app writers, QA, ... as well to have their
or rather your gear, the daemons, upper layer protocols, etc. working
flawlessly w/o legacy-IP for when the time comes that others want to
reduce mgmt costs and complexity and be able to say "no inet4" as well.

This is the beginning of the journey, and we'll hopefully continue
to head straight into the direction of "done, just works" to be able
to tick that checkbox off soon;)


A lot of the this June 8th World IPv6 Day verbiage was about picking the
right color, not so much for the bikeshed of finally putting IPv6 into
use, but it was hopefully a redpill day for some bluepill people.    /bz

More information about the NANOG mailing list