Hi, I’ve recently set up my box and helped a friend to set up one too, both on Linode. Both were set up nice and easy.
But my friend encountered what guys at Spamhaus call the “Linode problem”. The default IPv6 that you get on Linode can be in banned /64 range at Spamhaus (as they ban ranges and not single IPs for IPv6). Spamhaus advises in that case to ask for a personal /64 range of IPs at Linode, so one can be sure that they won’t get spamlisted if an IP next to theirs get blacklisted.
I didn’t have that problem but decided to get myself a range also and try setting a static IP from that range, just to make sure that I can if need arises.
Having spent some time on setting a static IP from a range and battling system’s attempts to use default SLAAC IP (which is not usually from a range that they give you) no matter what, I’ve encountered a stange behavior after running MIAB setup.
I have a static IP set up using netplan, and also a default route set for system to use it for outgoing connections.
I reboot several times, and
ip route get (which box uses to get private ip) and
curl -6 http://ip6only.me/api/ tell my static IP correctly
mailinabox script to update the configuration
It correctly gets public and private ip, all in well in /admin system check.
But after that, if reboot, Ubuntu now uses old SLAAC default IP for same checks, with the same before configured netplan.
Does the box setup do something with system routes that can explain the observed behavior?
(Manually changing the route with
ip route change "::/0" via "fe80::1" src "<my ipv6>" dev eth0 returns the desired behavior but only for a time, and netplan configuration for the same is somehow ignored after running setup
I also tried to disable accept-ra so the system won’t get a SLAAC ip at all, but in that case after running setup I lost ipv6 connectivity completely after first reboot)
I really want to write a guide afterwards.