Re: DHCPv6 support in Network Manager isn't RFC compliant

[Dan Williams <dcbw redhat com>]

On Fri, 2011-07-29 at 16:12 +0200, Tore Anderson wrote:
> * Paul Wouters
> 
> > Stopping the firewall did not help me on ietf-v6ONLY though. I still got
> > not DNS entry in /etc/resolv.conf and on top of that my routing seemed to
> > not have a working default route.
> > 
> > [...]
> > 
> > [paul@thinkpad ~]$ ip -6 ro li default
> > default dev wlan0  proto static  metric 1024  expires 2147157sec mtu
> > 1500 advmss 1440 hoplimit 0
> 
> This route is bogus, it basically says the entire IPv6 internet is
> directly attached to the layer 2 LAN segment you're on. I've seen NM
> create such a route before, but under different circumstances, and
> besides that bug should be long fixed - see
> <https://bugzilla.redhat.com/show_bug.cgi?id=588560>.
> 
> > default via fe80::212:1e00:70e7:bc00 dev wlan0  proto kernel  metric
> > 1024  mtu 1500 advmss 1440 hoplimit 64
> > default via fe80::205:8500:708e:3c00 dev wlan0  proto kernel  metric
> > 1024  expires 1716sec mtu 1500 advmss 1440 hoplimit 64
> 
> Here you have another two default routes via two different next-hops.
> Likely one of them is bogus, perhaps caused by a rogue RA. I find it
> curious that only one of them is displaying a lifetime counter, too.
> 
> Are you certain that NM adds all of these? One way to try is to set the
> IPv4 mode to link-local only, IPv6 mode to disabled/ignored, and then
> connect to the network. The connection should then «succeed», and you'll
> be able to see if anything IPv6-related is going on outside of NM's control.

Running NM with --no-daemon --log-level=debug is a great way to figure
out *exactly* what NM is doing with addressing and the routing table,
which people can use as a basis for diagnosing these issues...

Dan