Another big one.
- SOCKS5/HTTP proxy integration via "proxy mode"
- Kill switch (which was hacky and leaky) has been replace by a robust "lockdown mode" which keeps a dummy vpnservice tunnel active, capturing all device traffic and fowarding it to a netstack/gvisor tunnel via the SOCKS5 proxy. If a tunnel requires DNS resolution for peer endpoints, we punch a hole by bypassing a socket for DoH resolution to keep things secure and private.
- DoH support for peer endpoint resolutions for proxy, lockdown, and regular vpn mode (not kernel, yet) with support for cloudflare and Adguard DoH providers.
This is a big one.. oops.
Main changes:
- Make ping monitor more robust and global, with ping target overrides of the default cloudflare fallback target per tunnel (for full tunnels, otherwise we ping the internal tun ip)
- Include ping restart recovery to prevent tun being down if dns failures happen after a bounce
- Ping monitoring itself remains per tunnel and works without auto tunnel active, but moves the restart feature back to be managed by and integrated with auto tunnel to prevent inconsistencies and conflicts
- Ping statistics can be optionally included to be displayed with tun statistics
- Adds the beginnings of monitoring logs for handshake and data packet failures for userspace tuns (to be incorporated with restarts/tun status later)
- Improve tun error notifications, adds ping restart notifications
- Major refactor of auto tunnel logic to make it more modular and extensible for new auto tunnel conditions
- A bunch of other stuff..
zaneschepke