In article <[EMAIL PROTECTED]> (at Tue, 21 Dec 2004 09:05:28 +0100), Lo�c Minier <[EMAIL PROTECTED]> says:
> Hans Ulrich Niedermann <[EMAIL PROTECTED]> - Tue, Dec 21, 2004: > > > If you have IPv6 enabled on your system, but don't have IPv6 > > connectivity, then IPv6 enabled software (such as Firefox, dig, etc.) > > will go look for a AAAA DNS record and try to connect to the IPv6 > > address it gets there. This will not work, but take some time. > > I've seen this on a newly installed sarge with a slight difference: the > IPv6 packets would not leave the box, but the IPv6 would try to > autoconfigure itself and send packets on its subnets. Recent kernels (say 2.6.9 or something) do not support the "all-on-link assumption" (*) any longer. Now, if you don't have default route the the destination, connect() immideately fails and your application will fall back to ipv4. The changes has not backported to 2.4.x; will do. *: very common reason of delay > > The only way to avoid that is to either > > a) get IPv6 connectivity > > b) prevent the "ipv6" kernel module from being loaded on boot > > I think (at least in my case) deactivating autoconfiguration when none > is possible is enough, and this is probably possible via > /etc/sysctl.conf. Yes, but you don't need to do this in usual cases. BTW, if you have any comments on degration by introducing IPv6, please contact at <[EMAIL PROTECTED]>. --yoshfuji @ IPv6-Fix Project <http://v6fix.net>
