Leaderboard

Well, define "best". It's down to people's preference. I'd enable v4 and v6 both on router and in Eddie and prefer v6 wherever possible, others get spooked everytime they get a v6 address assigned, see below. This is one of the things you see people telling you all the time, and when asked, the primary, sometimes the only reason, is the possibility for IPv6 addresses to be assigned using the unique hardware address of an interface, which represents the wet dream of a tracking service, if I may write it like that. It needs to be mentioned, though, that no hardware for the private sector does that by default these days. Heck, newer versions of iOS and Android even randomize that hardware address by default. Thank you for seeing it this way. IPv6 itself grew out of puberty a long, long time ago, so I don't see the reason to shut it out, even in a VPN context. If we shut out v6 now (so people don't use it, get spooked about trifles, etcetc) and not think about how to make it more friendly for paranoids, v4 will never die. And v4 needs to die. We're still coming up with ridiculous ways of keeping a protocol alive and make it so it fits today's web of devices in one space, blatantly ignoring the simple fact it will never fit today's global web of devices in one space. It's like keeping CRT displays alive, for freakin' what? Buy a flatscreen, you'll get thrice your money's worth today. Would it be better? Well, for whom? For you there's no real difference between v4 and v6, both work. But for those building networks out there v4 vs v6 is like CRT vs. flatscreen, and you don't hear them asking the market for CRTs. On a side note, people constantly get angry at carriers for using Carrier-Grade NAT or Dual-Stack lite on their networks to provide v4 connectivity but very rarely ask themselves why tf this is the case…

After a bit more research I think I might know what is going on. It baffled me a bit to see the netmask being 255.255.255.255, but I realized it makes some sense for a point-to-point connection. Your host could be a dedicated server operating in a VLAN, or the host is directly connected to a router. Something like that. Now, a little brainstorming again. "Solutions" I can think of: Prevent OpenVPN from setting any routes, then set your own in a manner that still routes all traffic through the VPN (also see next point). Directives would be pull-filter ignore "redirect-gateway" and any number of route. The latter is subject to some serious thought. Play with the redirect-gateway and redirect-private directives. The latter does not change the default gateway, so one could think about adding custom routes routing the world through the default gateway, similar to how Eddie for Android adds exceptions to routing (which goes like this: 0.0.0.0/1 for the range 0.0.0.0-127.255.255.255, then 128.0.0.0/4 for the range 128.0.0.0-143.255.255.255, then 144.0.0.0/10 for the range 144.0.0.0-144.63.255.255 and so on; so basically routing one specific network, inversed) Set the subnet of the host to /17 or /18. This should create a route like I wrote prior by itself, if not, set it yourself. Then OpenVPN can do its job triggered by the redirect-gateway directive and not fail. Hopefully. A little more dangerous: Enable DHCP on eth0. You will lose connectivity if there is no DHCP server available (and it's likely you will, but hey, we're brainstorming). Contabo offers VNC, so you could still get the VNC address from customer panel, then connect using a VNC software, and once you're there, you can fix it by simply setting the static default route again. Think of VNC as access to the machine as if you connected a monitor to it. A complicated matter, really. Input from others is greatly appreciated.

This option controls whether you connect to an AirVPN server using IPv4 or IPv6. This option controls whether you want to route v4 only, v6 only or both v4 and v6 through the VPN. Definition of Dual Stack is not "v4 or v6" but "v4 and v6 at the same time". I'd leave both enabled, but some people might object and say "but you can be tracked with v6 on". As if they can't be tracked with v4. It doesn't need to match, though, it just needs to be enabled for systems to be able to use them. Meaning, if there is no IPv6 connectivity in your router, you'll have a hard time connecting to AirVPN using IPv6.

Yes Bluetit deamon starts with boot. But Goldcrest is not running. (update) Found this? localhost:~ # journalctl | grep bluetit Apr 22 07:31:44 localhost bluetit[2107]: Starting Bluetit - AirVPN OpenVPN 3 Service 1.1.0 RC3 - 16 April 2021 Apr 22 07:31:44 localhost bluetit[2107]: OpenVPN core 3.7 AirVPN linux x86_64 64-bit Apr 22 07:31:44 localhost bluetit[2107]: Copyright (C) 2012-2020 OpenVPN Inc. All rights reserved. Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Control process exited, code=exited status=1 Apr 22 07:31:44 localhost bluetit[2107]: Hummingbird client is running or it did not exit gracefully. Exiting. Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Unit entered failed state. Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Failed with result 'exit-code'. And this... (update) test with Bluetit - AirVPN OpenVPN 3 Service 1.0.0 - 7 January 2021 all fine! * bluetit.service - AirVPN Bluetit Daemon Loaded: loaded (/etc/systemd/system/bluetit.service; enabled; vendor preset: disabled) Active: failed (Result: exit-code) since Thu 2021-04-22 07:31:44 CEST; 38s ago Process: 2107 ExecStart=/sbin/bluetit (code=exited, status=1/FAILURE) Apr 22 07:31:44 localhost systemd[1]: Starting AirVPN Bluetit Daemon... Apr 22 07:31:44 localhost bluetit[2107]: Starting Bluetit - AirVPN OpenVPN 3 Service 1.1.0 RC3 - 16 April 2021 Apr 22 07:31:44 localhost bluetit[2107]: OpenVPN core 3.7 AirVPN linux x86_64 64-bit Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Control process exited, code=exited status=1 Apr 22 07:31:44 localhost systemd[1]: Failed to start AirVPN Bluetit Daemon. Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Unit entered failed state. Apr 22 07:31:44 localhost systemd[1]: bluetit.service: Failed with result 'exit-code'.