The official image requires you to write and mount a complete "config.yaml" before the container will even start. This image auto-generates the config from environment variables on first start, prints a pre-auth key to the logs so you can connect a device immediately, and includes an "hs_manage" helper for common admin tasks. It's also Alpine-based (~20–30 MB vs ~85 MB for the official Debian image).
It has the same Headscale binary underneath, just with a different out-of-the-box experience consistent with my other Docker images for WireGuard, OpenVPN, and IPsec VPN.
Hi, I am not sure what "false VPN propaganda" you are referring to? In the GitHub README there is no "propaganda" and it contains mainly just the facts about the project. If you would like to clarify your comment please reply.
You may optionally add "--net=host" to the "docker run" command to let the container use the host's network stack directly. That should eliminate the overhead I think.
I haven't looked into this yet, but I think it could work on the latest Raspbian 8 [1] which is based on Debian Jessie. You are welcome to clone the source repo on GitHub and give it a try.
Author here. To add or manage VPN users, you can modify "run.sh" and build a new Docker image from the source repo on GitHub. Please refer this README [1] for more info.
The reason I chose Libreswan [1] (over Openswan) is that it is more actively developed with recent patches [2]. I have not tried strongSwan as of yet.
Regarding the firewall rules: The VPN scripts are intended to be run on a freshly installed Linux system. By adding those IPTables rules, the OS would be better protected from network attacks. For example, the L2TP port UDP 1701 is closed for traffic other than those via IPsec.
It has the same Headscale binary underneath, just with a different out-of-the-box experience consistent with my other Docker images for WireGuard, OpenVPN, and IPsec VPN.