A Python tool to automatically build (and test) feature-rich configurations for BGP route servers.

How it works

  1. Two YAML files provide general policies and clients configurations options:

      rs_as: 999
      router_id: ""
      add_path: True
          policy: "same-as"
        policy_ipv4: "rewrite-next-hop"
      - asn: 111
        - ""
        - "2001:db8:1:1::11"
            - "AS-AS111MAIN"
  2. ARouteServer acquires external information to enrich them: bgpq3 for IRRDb data, PeeringDB for max-prefix limit, ...

  3. Jinja2 built-in templates are used to render the final route server’s configuration file.

    Currently, BIRD (1.6.3) and OpenBGPD (OpenBSD 6.0 and 6.1) are supported.

Validation and testing are performed using the built-in live tests framework: Docker instances are used to simulate several scenarios, and more custom scenarios can be built on the basis of the user’s needs. More details on the Live tests section.


  • Path hiding mitigation techniques (RFC7947 section 2.3.1).
  • Filtering features (most enabled by default):
    • NEXT_HOP enforcement (strict / same AS - RFC7948 section 4.8);
    • minimum and maximum IPv4/IPv6 prefix length;
    • maximum AS_PATH length;
    • reject invalid AS_PATHs (containing private/invalid ASNs);
    • reject AS_PATHs containing transit-free ASNs;
    • RPKI-based filtering (RFC6811);
    • reject bogons;
    • prefixes and origin ASNs enforcing via RPSL/IRRdb AS-SETs (RFC7948 section 4.6.2);
    • max-prefix limit based on global or client-specific values or on PeeringDB data.
  • Blackhole filtering support:
    • optional NEXT_HOP rewriting;
    • signalling via BGP Communities (BLACKHOLE and custom communities);
    • client-by-client control over propagation.
  • Control and informative communities:
    • prefix/origin ASN present/not present in IRRDB data;
    • routes RPKI validity state;
    • do (not) announce to any / peer;
    • prepend to any / peer;
    • add NO_EXPORT / NO_ADVERTISE to any / peer.
  • Optional session features on a client-by-client basis:
  • Automatic building of clients list:

A comprehensive list of features can be found within the comments of the distributed configuration file on GitHub.

More feature are already planned: see the Future work section for more details.


Beta testing, looking for testers and reviewers.

Anyone who wants to share his/her point of view, to review the output configurations or to test them is more than welcome!

Bug? Issues?

But also suggestions? New ideas?

Please create an issue on GitHub or drop me a message.


Pier Carlo Chiodi -

Blog: Twitter: @pierky