Hello I've done similar before, to add ipv6 addresses to ipv4 only vps's The ipv6 addresses have come from either the /56 native ipv6 from my isp or from /48 6to4 subnets from dual stack vps's with a limited number of ipv6 addresses, sometimes just one. This has worked ok without having to look at proxy_ndp or ip neigh proxy.
With a new OVH server myself I have also got this working with proxy_ndp set to 1, using ip neigh in tinc-up and some tweaking to the ipv6 route table. A /48 of 6to4 addresses is also available to you if you want to test this. On Sun, 2013-10-06 at 15:49 +0200, Ismael Bouya wrote: > Hi, > I finally solved my problem (I planned to send an email "later", but since > you answered to me I'll make an effort :D ) : > > The server (who "owns" the /64 addr) needs more than just "forwarding" set > to 1, I also need to specify proxy_ndp to 1 (in > /proc/sys/net/ipv6/conf/***/proxy_ndp) and to do > > ip neigh add proxy ***ipv6*** dev eth0 > > for each address that will go through the router (including the one on the > vpn6 interface of the server) > > Note that if you follow strictly the doc at > http://www.tinc-vpn.org/examples/ipv6-network/ > you only need to run the above command for 2001:db8:beef::{2,3,4} on > routera > and then on each router{b,c} you'll have to worry about their own subnet > only > > > Maybe it should be tried somewhere else. I think it's a problem due to the > fact that I have "only" a /64, and thus all the ip addresses I can address > are in the "interface" scope of the ip6 address. Since I don't have access > to a /48 I cannot run more tests about this asumption, but I'd be quite > interested in knowing whether it is true or not (maybe someone there has > this kind of network and can check both the value of > /proc/sys/net/ipv6/conf/***/proxy_ndp on "routera" and > run ip -6 neigh show proxy > to check the necessity of this in larger network?) > > > Hm, that's indeed strange. However, the example on the website is a bit > > complicated, maybe you could simplify your setup. Does home need its own > > /64 or > > does it need only a /128? Do you plan to add more nodes or not? > > Each "node" only needed a /128, the /96 was a "bonus", but none of the > methods I tried worked > > > > I also put /proc/sys/net/ipv{4,6}/conf/all/forwarding to 1 everywhere, > > > but without success. > > > > > > Did I miss anything? > > > > Make sure you don't have firewall rules blocking the forwarding of IPv6 > > packets. Also, what does the routing table look like on home? > > I also checked that, and the routing table at home was correct. The example > documentation is perfect but for the proxy_ndp problem. > > Maybe you should put a note about it in the example? > > Anyway, thanks for all (and especially for this wonderfull program :D ) > > Best regards, > _______________________________________________ > tinc mailing list > [email protected] > http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc _______________________________________________ tinc mailing list [email protected] http://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc
