IPv6 in FC4 - How

Sat Nov 5 18:42:57 UTC 2005

On Sat, 2005-11-05 at 14:10 +0000, Timothy Murphy wrote:
> Tim wrote:

> > I hadn't noticed until recently that ifconfig showed IPv6 addresses on
> > my PC.  I had set up the NICs manually, and IPv6 wasn't ticked on, and
> > thought nothing more about it.  Later I noticed the addresses in the
> > ifconfig output, so I tried turning IPv6 on and off in the network
> > configuration, but didn't notice any change in the output of ifconfig.
> > 
> > I haven't noticed any other effects.  Should I bother to do anything
> > about it?  I don't need IPv6 internally.

> I was going to include IPv6 in the kernel I'm just compiling - 2.6.14 -
> but after reading this thread it sounded as though
> it was likely to cause unwanted problems.

	The only unwanted problems I'm aware of seem to occur when it's enabled
but not properly configured.  Even then, it SHOULD work fine most of the
time.  Some people have reported unusual delays in connections which I
suspect are due to the underlying libraries and applications attempting
to access IPv6 sites when you are not connected to IPv6.  I've never
personally experienced any of these problems, myself, though.  There are
already a number of web sites and DNS servers and mail servers (and ntp
servers and ftp servers and irc servers, etc, etc, etc) that are
supporting IPv6.  The client applications (and resolver libraries) will
generally try IPv6 first, when the server has an IPv6 address advertised
and IPv6 is present on the client system.  If the routing tables are set
up correctly, you should get an immediate error when you attempt to
access IPv6 sites and are not connected to the global space.  It seems
to be the case where there are no routing tables setup that seems to be
the hang up and delays.  If you compile in IPv6 (as a module would
probably be wiser) just make sure you enable it in the configurations so
all the default routing table structure is initialized properly.  Or get
connected to IPv6 for real.  :-)

> As far as I am concerend IPv4 works perfectly -
> I was only thinking of IPv6 to keep up with the Jones's.
> It's a bit like selinux - great in theory,
> but in practice likely to rasise more problems than it solves.

	Well...  Let's look at a few number from my BGP servers...

IPv4:
Number of routes in the routing tables:		171859
Total number of unique host addresses:		1441100864

IPv6:
Number of routes in the routing tables:		631
Total number of unique NETWORK addresses:	1512195986

	Hmmm...  So the IPv4 core gateways currently know how to route 1.44
billion IPv4 host addresses.  The IPv6 core gateways know how to route
1.51 billion IPv6 networks (that's a production /48 v6 network
supporting 65536 subnets each).  More v6 networks routed in the core
than v4 host addresses.  Yeah, not all of those networks are populated
(it does, at least, route to an ISP) and it's harder to tell if you've
got a real network at the end of that IPv6 route than it is to tell if
you got a real host at the end of an IPv4 route.  An IPv6 core route is
suppose to be a minimum of 65,536 networks (but there are a few strays
that don't adhere to the standard).  But not near all of those IPv4
addresses are populated either.

	Also...  If you look at the "DNS Server Survey" from early last year,
you might notice an interesting item near the bottom.  One of the DNS
servers they found was "totd" (the Trick Or Treat Daemon).  About 600
instances of totd service a couple thousand domains.  Guess what...
Each instance of totd represents a minimum of one or more IPv6-Only
networks - no IPv4 present.  That's what totd is used for.  Proxying,
caching, and mapping IPv4 DNS requests and response over to an IPv6 only
network.  I haven't seen a more recent survey and the use of IPv6 has
just exploded over the last 18 months.  So, no telling now...

	That's a lot of potential Jones out there...

> Or is that nonsense?

	I wouldn't call it nonsense.  Just inaccurate.  I've found that, from
personal experience, IPv6 has actually solved more problems (especially
when I'm roadwarrior at shows and such) than it creates.  In fact, I
don't know of any problems it has created for me.  But then I'm fully
configured, connected, and actually using it.  So, I guess, I'm one of
the Jones you would be keeping up with.  :-)

> -- 
> Timothy Murphy  
> e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie
> tel: +353-86-2336090, +353-1-2842366
> s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

	In Europe you should have good access to IPv6 natively from a lot of
ISPs and I know SixXS has a good presence there in the British Isles.

	Mike
-- 
 Michael H. Warfield    |  (770) 985-6132   |  mhw at WittsEnd.com  
  /\/\|=mhw=|\/\/       |  (678) 463-0932   |  http://www.wittsend.com/mhw/
  NIC whois:  MHW9      |  An optimist believes we live in the best of all
 PGP Key: 0xDF1DD471    |  possible worlds.  A pessimist is sure of it!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 307 bytes
Desc: This is a digitally signed message part
Url : http://lists.fedoraproject.org/pipermail/users/attachments/20051105/e5343193/attachment-0002.bin