Re: [v6ops] IPv4 trajectory

[Mark Andrews <[e-mail]>]

In message <[e-mail]>, Alexandru Petrescu writes:
> Le 02/04/2015 13:04, Bjoern A. Zeeb a =E9crit :
> >
> >> On 02 Apr 2015, at 06:36 , Alexandru Petrescu
> >> <[e-mail]> wrote:
> >>
> >> Sorry to insist on this, just as side note.
> >>
> >> Le 01/04/2015 19:54, Bjoern A. Zeeb a =E9crit :
> >>>> On 01 Apr 2015, at 14:07 , Alexandru Petrescu
> >>>> <[e-mail]> wrote:
> >>>>
> >>>> It is customary to see current computers having both an IPv6
> >>>> stack and an IPv4 stack on them.  But some computers only have
> >>>> IPv4, whereas the rest have IPv4 and IPv6 stacks.  There are no
> >>>> computers (to my knowledge) which have only IPv6 stacks.
> >>>
> >>> One of my desktop systems does (not have IPv4 support anymore;
> >>
> >> In general I can agree with you, in a sense where maybe a computer
> >> is not reachable on IPv4 from another computer; maybe part of IPv4
> >> support is no more there.
> >
> > On FreeBSD I compile it out of my kernel and parts of user space if
> > I want to.  We did this for World IPv6 Day almost 4 years ago:
> > http://www.prweb.com/releases/2011/6/prweb8529718.htm
> >
> > If you used FreeBSD jails for services you could run them IPv6-only
> > since FreeBSD 7.2 and you do get =93Address Family Not Supported=94 back
> > for every IPv4 operation you try to do (7.2 was released in May
> > 2009).  Gosh, been doing this for more than 5 years already.
> 
> That's good for BSD then, I am happy for it.  Does it remove the
> 127.0.0.1 from its /etc/hosts too?
> 
> In linux you can't make the IPv4 stack or any of its features into a
> module.  It's builtin.

It looks like Linux needs some work then.
 
> >>> and to my knowledge you can just disable it on Windows as well,
> >>
> >> Right - uncheck the IPv4 box, and check IPv6 on a computer Windows
> >> 7's =93Properties" of the network interface - it works.
> >
> > Wrong:
> >
> > netsh interface ipv4 uninstall and reboot
> 
> I'd like to try that.  But if it fails can I 'install' it back somehow?
> 
> (because I already experienced some hangs with checking off IPv4 while
> wireshark running - for some reasons some apps think that if IPv4 is not
> there then networking is not there although they are IPv6-capable).

Which is a application bug that should be reported.
 
> > and you=92ll not see IPv4 anymore in route print or ipconfig /all and
> > ping 127.1 will no longer work but error on you.
> 
> And Windows' hosts file containing 127?

Do you want to rip all the A records out of the DNS as well?  If
the getaddrinfo() is properly written the IPv4 addresses will be
filtered from the responses with AI_ADDRCONFIG set so it shouldn't
matter if it is still there.

> >>> so it=92s not a big deal anymore).
> >>
> >> I beg to differ, as I see it this is still far away from an ideal
> >> where only IPv6 were present in the computer.
> >
> >
> > Well, better start living it today and we can have a lot better
> > operational discussions on the end-node (client/server systems)
> > side. I have 7 IPv4 addresses left in use in total for all my
> > infrastructure (and could condense it to 2 or 3 if I wanted really
> > badly).  And the only reason I keep those is for the ipv4-only world
> > to reach NS/MX/Web.  But that=92s just me.  Leave a /27 for a SMB and
> > be good ;-)
> 
> Congratulations.
> 
> I have no BSD here, except for the proprietary derivatives (ios,
> windows, etc).  For all code that I can modify it's linux.  Given that I
> can't turn off IPv4 until linux gives a means to.

You have a editor, a compiler and the source.  You have the means
to do it.  Whether you have the skill / inclination to do it is
another matter.

> Do you know whether ios has such a means to turn off fully IPv4?
> 
> Alex
> 
> >
> > =97 Bjoern A. Zeeb                                  Charles Haddon
> > Spurgeon: "Friendship is one of the sweetest joys of life.  Many
> > might have failed beneath the bitterness of their trial  had they
> > not found a friend."
> >
> >
> >
> 
> 
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: [phone]                 INTERNET: [e-mail]