FC3: IPv6 Route add failed; ping6 not working

[Michael H. Warfield]

On Fri, 2005-10-28 at 11:04 +0200, Eder Norbert wrote:
> Hello all,

> I have a Fedora Core 3 PC with a IPv6 Address.
> I will ping now a IPv6 address in a other Segment/VLAN, so I must add a
> route.
> But this fails with "RTNETLINK answers: Invalid argument" or "Error:
> either "to" is duplicate, or "gw" is a garbage."

	You're probably going to have to provide a LOT more information, like
your routing tables and interface configurations, but...  One immediate
question comes to mind.  Why are you not simply using autoconf rather
than going through this much effort?  Your router should be advertising
itself to the end nodes.  You shouldn't need to add any routes or
addressing prefixes manually to any end nodes unless you are setting up
tunnels and routers and such on them.  Contact me off-list if you are
not comfortable with sharing this level of detail in public.
 
> I tried following commands:
> ip -6 route add '16::/16'  via  51::f:1 dev eth0
> ip -6 route add '16::/16'  via  51::f:1 
> ip -6 route add '16::/16'  gw  51::f:1 dev eth0
> ip -6 route add '16::/16'  gw  51::f:1 

	What's the network you are on?

	If those really are the network addresses you are using, that's also
VERY strange network addresses.  It's not a properly allocated network.
Do you have a real v6 network assignment like this (mine):
2001:4830:3000::/48?  If not, you should be using one of the local scope
ranges.  Use the site-local range fec0::/10, if all you want to do is
play around.  Typical subnets are /64's while full networks are /48's.
Your subnet number would be in the third field of hex numbers.

> With 
> ip -6 route add '16::/16'   dev eth0  
> it is working, but when I make 
> ping6 16::f:1
> he always used the loop-back Interface and not the eth0.

	But all you've declared is that 16::/16 is ON that device.  You haven't
told it to route through anything.  What's the address of your router
and, again, why isn't it "advertising" and why aren't you autoconfing?
This isn't IPv4.

> Do you know, were the problem is??
 
	If you are going to play with IPv6, get with a tunnel broker and get a
real allocation.  Seriously.  Once your systems have IPv6 configured
they are going to default to trying to USE IPv6 preferentially for
things like smtp and http where servers have v6 addresses.  IPv6 is very
common in the DNS now.  If you don't have this set up right, your
systems are going to end up dain bramaged slow and even broken for some
things, if you don't have your setup correct.  You can get free
allocations (/64 and/or /48) from tunnelbroker.net (US and Australia),
freenet6.net (North America), occaid.org (if you know how to speak BGP -
not for amateurs), or sixxs.net (in Europe).  They've also got good
information on configuring and getting started with IPv6.  OTOH...  If
you aleady have live native IPv6 at your ISP (Europe and Asia), then set
up your routers properly with router discovery advertisements.

> Thanks.
> bye,
> Nobsi

	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: signature.asc
Type: application/pgp-signature
Size: 307 bytes
Desc: This is a digitally signed message part
URL: <http://listman.redhat.com/archives/fedora-list/attachments/20051028/e6389a76/attachment-0001.sig>