Threat Intel & GeoIP Settings
Elchi enriches the traffic that API Discovery observes with two kinds of reference data that you load as an administrator:
- Threat-intel feeds: lists of IP addresses and CIDR ranges (a Spamhaus DROP list, an AbuseIPDB export, your own red-team list). A request from a listed address is tagged with the feed name and raises the critical
threat_intel_hitrisk flag. - GeoIP databases: a City and an ASN database in MaxMind MMDB format. They resolve each client IP to continent, country, city, ASN and network owner for the geo, consumer and risk views.
Both are managed under Settings → API Discovery, in the Threat Intel and GeoIP sections below the collector configuration. Only the Owner and Admin roles can change them. This page is the administrator's checklist; how the collector applies the data, and the tags it writes, is in Threat Intel & GeoIP in API Discovery.
Scope: platform-wide
Feeds and databases are global. They are stored once (the feeds in the MongoDB document api_collector_threatintel, the databases in the MongoDB GridFS bucket geoip) and apply to every project. Every collector replica picks up a change on its next poll, about every two minutes, without a restart, so one upload covers the whole fleet.
Load GeoIP databases
The GeoIP section shows one card each for city and asn, with the size, source, uploader and SHA-256 of the database in use. For each card you can:
- Download db-ip Lite: the backend fetches the free db-ip.com Lite database for the current month (falling back to the previous month) and stores it. The backend needs outbound HTTPS to
download.db-ip.com. - Upload .mmdb: bring your own file, for example a paid MaxMind GeoIP2 or a GeoLite2 database, up to 256 MiB. The backend checks that it is a valid MaxMind database of the right type (City for
city, ASN forasn) before storing it. - Delete: remove the database.
Downloads and uploads can take a while for large files; the backend allows them up to five minutes. Without a database, GeoIP is simply off: traffic is still marked internal or external, but gets no country or ASN. The ASN database also drives the hosting/VPN signal of the impossible_travel detector.
Plan to refresh the databases regularly (db-ip publishes its Lite databases monthly), since IP allocations change over time.
Load threat-intel feeds
Click Upload feed, give the feed a name (for example spamhaus-drop) and drop a text file with one IPv4 or IPv6 address or CIDR range per line. Blank lines and # comments are ignored. Lines that do not parse are skipped and counted; one bad line never rejects the whole file.
| Limit | Value |
|---|---|
| Entries per feed | 250,000 |
| Entries across all feeds | 600,000 |
Uploading a feed under an existing name replaces it. In the feed list you can view a feed's entries, mute it with the Enabled switch without losing its entries, or delete it. When an address is on several feeds, the event lists all of them.
To keep a feed current, re-upload it on a schedule, either by hand or by calling the API from a job.
Automating updates
The same operations are available on the controller API for an Owner or Admin session:
| Task | Call |
|---|---|
| List feeds | GET /api/v3/setting/threat-intel |
| Upload or replace a feed | POST /api/v3/setting/threat-intel/upload (multipart: name, file) |
| Enable, disable or replace all feeds | PUT /api/v3/setting/threat-intel |
| Delete a feed | DELETE /api/v3/setting/threat-intel/{name} |
| Show GeoIP databases | GET /api/v3/setting/geoip |
| Upload a database | POST /api/v3/setting/geoip/upload (multipart: kind = city or asn, file) |
| Fetch db-ip Lite | POST /api/v3/setting/geoip/download (body {"kind": "city"}) |
| Delete a database | DELETE /api/v3/setting/geoip/{kind} |
Request and response shapes are in Threat Intel & GeoIP in API Discovery.
Detection, not blocking
These feeds tag and score traffic in API Discovery; they do not block anything. To drop requests from bad addresses at the edge, use Shield's IP reputation engine, which enforces inline in the Envoy ext_proc sidecar and has its own address lists.
Without GeoIP and threat data, the geo, consumer, and risk views still work but show less context. Loading these databases makes them far more actionable.