provide geographic information about mail senders.
Use MaxMind's GeoIP databases to report geographic information about senders.
This plugin supports several geoip modules:
* [maxmind](https://github.com/runk/node-maxmind) 0.6 (maxmind v1) * [maxmind](https://github.com/runk/node-maxmind) 1.0+ (maxmind v2) * [geoip-lite](https://github.com/bluesmoon/node-geoip)
geoip-lite is published to npm separately as haraka-plugin-geoip-lite.
The maxmind module requires the manual download of the GeoIP databases. The npm module maxmind-geolite-mirror will download the files for you and also keep them up-to-date if you run it periodically.
mkdir -p /usr/local/share/GeoIPnpm install -g maxmind-geolite-mirror/usr/local/bin/maxmind-geolite-mirror
GeoIP results are stored in connection.notes.geoip and the haraka-results object at
connection.results.geoip. The following information is typically available:
continent: NA, country: US,
If the GeoIP city database is available, the following may also be available:
region: CA, city: San Francisco, ll: [37.7484, -122.4156], distance: 1539 // in kilometers range: [ 3479299040, 3479299071 ],
geoip also adds entries like this to your logs:
[geoip] US [geoip] US, WA [geoip] US, WA, Seattle [geoip] US, WA, Seattle, 1319km
Calculating the distance requires the public IP of this mail server. This may
be the IP that Haraka is bound to. If not, make sure that
can figure it out (automatically via STUN or manually specified in
Performs geodesic distance calculations. Calculates the distance "as the crow flies" from the remote mail server.
This calculation requires a 'from' IP address. This will typically be the
public IP of your mail server. If Haraka is bound to a private IP, net_utils
will attempt to determine your public IP. If that doesn't work, edit
config/smtp.ini and set
Set a connection result to true if the distance exceeds this many kilometers.
show city data in logs and headers. City data is less accurate than country.
- show.region in logs and headers. Regional data are US states, Canadian provinces and such.
SPAM PREDICTION WITH DISTANCE
"For ham, 90% of the messages travel about 4,000 km or less. On the other hand, for spam, only 28% of messages stay within this range."
Observations in 2014-2016 suggest that geodesic distance continues to be highly correlated with spam.
The distance calculations are more concerned with being fast than accurate. The MaxMind location data is collected from whois and is of limited accuracy. MaxMind offers more accurate data for a fee.
For distance calculations, the earth is considered a perfect sphere. In reality, it is not. Accuracy should be within 1%.