Thanks, Guus, for your reply. I just wanted to confirm that the workaround
up ip -6 route add default via fe80::1 dev $IFACE onlink works flawless for me. I will upgrade to Stretch with one of the next point releases. Thanks again, Rainer Am Samstag, 15. Juli 2017, 21:30:27 CEST schrieb Guus Sliepen: > On Sat, Jul 15, 2017 at 07:31:29PM +0200, Rainer Dorsch wrote: > > > I added a IPv6 support to a KVM virtualized Jessie system following the > > > instructions by the hoster > > > > > > https://www.netcup-wiki.de/wiki/Zus > > > %c3%a4tzliche_IP_Adresse_konfigurieren#IPv6%7C > > That should still work, but I see it uses the ancient interface alias > notation (like eth0:1). The example given there can be rewritten to a > more modern form: > > allow-hotplug eth0 > iface eth0 inet static > address 37.100.195.4/22 > gateway 37.100.192.1 > > iface eth0 inet static > address 46.38.240.15 > > iface eth0 inet6 static > address 2a03:4000:2:11c5::1/64 > gateway fe80::1 > > Note that the netmasks and broadcast addresses are gone, and you can > have multiple inet stanzas for the same interface. > > > > root@netcup:~# /sbin/route -6 > > > Kernel IPv6 routing table > > > Destination Next Hop Flag Met Ref > > > Use If > [...] > > > > ::/0 fe80::1 UGDAe 1024 3 > > > ::0 eth0 > The flags field contains the letter 'e', which means that this route > expires. This is added by the kernel, most likely because the gateway > address is not inside 2a03:4000:2:11c5::/64, and will cause it to remove > the route after a while. This looks like the same issue reported in > Debian bug #378506, which is fixed in stretch. > > The workaround is to remove the gateway fe80::1 statement, but to add > this instead: > > up ip -6 route add default via fe80::1 dev $IFACE onlink -- Rainer Dorsch http://bokomoko.de/
