IPv6 router advertisement
ionic at ionic.de
Wed May 6 19:34:07 UTC 2009
* On 06.05.2009 20:27, Ted Lemon wrote:
> So what are the sysctl settings for net.ipv6.conf.all.accept_ra
net.ipv6.conf.all.accept_ra = 1
> and net.ipv6.conf.all.accept_ra_defrtr?
net.ipv6.conf.all.accept_ra_defrtr = 1
> And what about for the same variables that are specific to the interface?
net.ipv6.conf.eth0.accept_ra = 1
net.ipv6.conf.eth0.accept_ra_defrtr = 1
> Do you get a default route if you plug the same machine into a /64
> network with the same RA settings?
I've just changed /96 to /64, restarted radvd and gave my laptop in two
rounds 2001:41d0:1:678e:abcd::2/128 and 2001:41d0:1:678e:abcd::2/64 on
eth0, without any result. :(
> Also, you didn't answer this question:
> On May 5, 2009, at 12:53 PM, Mihai Moldovan wrote:
>> AdvRouterAddr off;
> Why is AdvRouterAddr off?
You're right of course, sorry.
I had a quick look in the manpage of radvd.conf and saw that
AdvRouterAddr is defaulting to off and just to be used in "Mobile IPv6"
setups which I don't have. Thus I thought disabling it might be a good idea.
However, just after the first question about it turned up on the ML, I
re-enabled it, just in case disabling it was wrong. :)
> Given the symptoms you're describing, my working theory is that
> Microsoft's autoconf implementation is synthesizing the default route
> using the source address of the RA message, and Linux is not.
Sorry for asking such a stupid question, but isn't this was RA is
actually meant to be used for? Or where does Linux get the router
> I suspect the Linux behavior is more compliant with the protocol spec,
> although I'm sure if this theory is correct that Microsoft had a good
> reason for doing what they did.
Or maybe not, based on the assumption that IPv6 is only to be used for
development and testing purposes in Windows.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 898 bytes
Desc: OpenPGP digital signature
More information about the dhcp-users