I've had problems with Hetzner and v6 also. When I was configuring v6 sub-subnets from the /64 they give out, on containers, I would get the same behavior. From what I could tell because the container bridges also use the fe80::1 link local route for the sub-subnets hetzners next-hop would get confused and just stop pushing any v6.
They also have a bunch of Layer2 switch monitoring stuff they do to watch for mainly v4 routes and IP's they are not expecting to exit from a particular ports on their switches. They call it 'locking' in their documentation and there isn't very precise description of what exactly it is they are monitoring. -Joel On 22 May 2015 at 15:15, Thomas Bohl <thomas-b...@gmx.de> wrote: > Am 22.05.2015 um 23:55 schrieb Heiko Zimmermann: > >> # route delete -inet6 default >>> # route add -inet6 default fe80::1%re0 >>> >> >> It worked one time until reboot. >> > > That is a problem I have too: > http://marc.info/?l=openbsd-misc&m=142249632125559&w=2 > > You first need to ping fe80::1%re0 in order to get it into the address > mapping table. I tried static entries without luck. Try to add > > @reboot sleep 10 && ping6 -c 10 fe80::1\%re0 > /dev/null > > into your crontab. Works for me. > > > Could you advise me hostname.re0 and mygate ? >> > > I don't use mygate. Just > > !route add -inet6 default fe80::1%re0 > > in hostname.re0
