- Remove `DNS_SERVER` (aka DOT) option: the DNS server forwarder part is now always enabled (see below why)
- Remove `DNS_KEEP_NAMESERVER`: the container will always use the built-in DNS server forwarder, because it can handle now local names with local resolvers (see #2970), it can use the `plain` upstream type (see https://github.com/qdm12/gluetun/commit/5ed6e8292278b54bb5081de0e8ccd0d63a275b3c) AND you can use `DNS_UPSTREAM_PLAIN_ADDRESSES` (see below)
- Replace `DNS_ADDRESS` with `DNS_UPSTREAM_PLAIN_ADDRESSES`:
- New CSV format with port, for example `ip1:port1,ip2:port2`
- requires `DNS_UPSTREAM_TYPE=plain` to be set to use `DNS_UPSTREAM_PLAIN_ADDRESSES` (unless using retro `DNS_ADDRESS`)
- retrocompatibility with `DNS_ADDRESS`. If set, force upstream type to plain and empty user-picked providers. 127.0.0.1 is now ignored since it's always set to this value internally.
- Warning log on using private upstream resolvers updated
- Warning log if using a private IP address for the plain DNS server which is not in your local subnets
All in all, this greatly simplifies code and available options (less options for the same features is a win). It also allows you to specify multiple plain DNS resolvers on ports other than 53 if needed.
- this makes PMTUD TCP reliable
- this only works on kernels with the mark module
- on kernels without the mark module, the icmp pmtud mtu found is used
- Existing option `WIREGUARD_MTU` , if set, disables PMTUD and is used
- New option `PMTUD_ICMP_ADDRESSES=1.1.1.1,8.8.8.8` and `PMTUD_TCP_ADDRESSES=1.1.1.1:443,8.8.8.8:443`
- ICMP PMTUD now targets external-by-default IP addresses
- New TCP PMTUD (binary search only) as a second MTU confirmation and fallback mechanism.
- Force set TCP MSS to MTU - IP header - TCP base header - "magic 20 bytes" 🎆
- Fix#3108
- Specify fallback ICMP IP addresses
- Defaults changed from 1.1.1.1 to 1.1.1.1,8.8.8.8
- Small periodic check cycles through addresses as it fails and moves to retry
- Specify fallback addresses
- Defaults changed from cloudflare:443 to cloudflare:443,github.com:443
- Startup check runs a parallel dial to each of the addresses specified with a global 6s timeout
- Full periodic check cycles through addresses as it fails and moves to retry
- `-proton-username` flag for cli update
- `-proton-password` flag for cli update
- `UPDATER_PROTONVPN_USERNAME` option for periodic updates
- `UPDATER_PROTONVPN_PASSWORD` option for periodic updates
- For example: `{"auth":"basic","username":"me","password":"pass"}`
- For example`{"auth":"apiKey","apikey":"xyz"}`
- For example`{"auth":"none"}` (I don't recommend)
- Migrate `DOT` to `DNS_SERVER`
- Migrate `DOT_PROVIDERS` to `DNS_UPSTREAM_RESOLVERS`
- Migrate `DOT_PRIVATE_ADDRESS` to `DNS_PRIVATE_ADDRESSES`
- Migrate `DOT_CACHING` to `DNS_CACHING`
- Migrate `DOT_IPV6` to `DNS_UPSTREAM_IPV6`
- Cloudflare's 1.1.1.1 seems more reliable than the VPN server public IP address you connect to
- This can still be changed back to 0.0.0.0 to use the VPN server IP address if needed
- You should really leave it to `on` ⚠️
- Turn it to `off` if you have trust issues with the healthcheck. Don't then report issues if the connection is dead though.
- New option: `HEALTH_ICMP_TARGET_IP` defaults to `0.0.0.0` meaning use the VPN server public IP address.
- Options removed: `HEALTH_VPN_INITIAL_DURATION` and `HEALTH_VPN_ADDITIONAL_DURATION` - times and retries are handpicked and hardcoded.
- Less aggressive checks and less false positive detection
- Reduce silly image pull rate limiting from docker hub registry
- still rely on docker hub registry to pull golang and alpine images since these are not on ghcr.io
- `PUBLICIP_API` accepts a comma separated list of ip data sources, where the first one is the base default one, and sources after it are backup sources used if we are rate limited.
- `PUBLICIP_API` defaults to `ipinfo,ifconfigco,ip2location,cloudflare` such that it now has `ifconfigco,ip2location,cloudflare` as backup ip data sources.
- `PUBLICIP_API_TOKEN` accepts a comma separated list of ip data source tokens, each corresponding by position to the APIs listed in `PUBLICIP_API`.
- logs ip data source when logging public ip information
- assume a rate limiting error is for 30 days (no persistence)
- ready for future live settings updates
- consider an ip data source no longer banned if the token changes
- keeps track of ban times when updating the list of fetchers
- No point periodically fetch the public IP address. Could not find anything mentioning why this was added.
- Simplification of the publicip loop code
- `PUBLICIP_ENABLED` (on, off) can be set to enable or not public ip data fetching on VPN connection
- `PUBLICIP_PERIOD=0` still works to indicate to disable public ip fetching
- `PUBLICIP_PERIOD` != 0 means to enable public ip fetching
- Warnings logged when using `PUBLICIP_PERIOD`
- Faster start up
- Clearer error messages
- Allow for more Gluetun-specific customization
- DNSSEC validation is dropped for now (it's sort of unneeded)
- Fix#137
- `STORAGE_FILEPATH=` disables storing to and reading from a local servers.json file
- `STORAGE_FILEPATH` defaults to `/gluetun/servers.json`
- Fix#2074
- Split `VPN_ENDPOINT_IP` in `OPENVPN_ENDPOINT_IP` and `WIREGUARD_ENDPOINT_IP`
- Split `VPN_ENDPOINT_PORT` in `OPENVPN_ENDPOINT_PORT` and `WIREGUARD_ENDPOINT_PORT`
- Fixes bad usage of Wireguard config file endpoint for OpenVPN #2347
Quentin McGaw