On Fri, 2009-06-05 at 15:59 +0300, Stoyan Angelov wrote:
> Sebastien Roy wrote:
> > On Fri, 2009-06-05 at 14:49 +0300, Stoyan Angelov wrote:
> >> i am trying to configure several zones as ipv6-only hosts. i do not want 
> >> to have the ipv4 stack enabled at all.
> > 
> > There is not a separate "stack" for IPv4 and IPv6.  It is one networking
> > stack, and IPv4 or IPv6 is used depending on the interfaces and
> > addresses configured.
> > 
> >> i have been using dual-stack 
> >> configuration for a long time but now i want to have pure ipv6 machine.
> >> is this possible ?
> > 
> > Yes, simply don't configure any IPv4 addresses.  Note that you should
> > keep the IPv4 loopback address, as local applications may depend on
> > being able to use it.
> 
> hi Sebastien,
> 
> i have configured the zones in question that way, however i thought the 
> ipv4 loopback address could be removed.

You can remove it manually, but IMO there is no benefit to doing that
(why would it matter?).  It is always created when the ip module is
loaded (and everytime a zone boots for non-global zones), so if you're
really adamant about it, you'll have to remove it at every boot.  I also
reiterate that I don't know what will break if you do that.

> also the interactive installer refuses to continue if you don't specify 
> a correct ipv4 address. should this be considered a bug ?

Yes, absolutely; anything that doesn't work with IPv6 should be
considered a bug, and a fairly high-priority bug at that.  There has
been a "big rule" for Solaris for about a decade that everything must
work with IPv6.

While it's not a strictly a bug that pkg.opensolaris.org and other
existing repositories don't resolve to AAAA addresses (it's a general
deployment problem with the entire OpenSolaris networking
infrastructure), it is certainly be a bug if the software responsible
for running repositories, as well as the client-side installation and
packaging, didn't allow for such deployments to be possible.

Can you file one at defect.opensolaris.org?

Thanks,
-Seb


_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to