cetanu / steward

An implementation of the Lyft rate-limit service

steward

Mission Statement

Steward is an implementation of the Lyft Rate-Limit service.

Features

• Load rate limit configs from HTTP or a local file

Building and testing

Prerequisites

• Docker
• Docker-compose
• Make

Optional prerequisites for building locally

• Rust toolchain

Building locally

Simple execute cargo build --release to create a binary which, when run, will start the rate limit service as a gRPC server.

Running the environment

The environment can be brought up with make run.
It includes an envoy proxy, the rate limit service, a redis database, a mock configuration server, and a httpbin backend.

Running tests

The project uses tavern HTTP integration tests.
They can be executed with make test

Configuration

The path to local configuration can be specified using the environment variable STEWARD_CONFIG_PATH.
The default location is steward.yaml in the current working directory.

Example configuration file:

listen:
  addr: 0.0.0.0
  port: 5001
rate_limit_configs:
  http: http://mock_config:8000/api/rate_limits
redis_host: redis
redis_connections: 8
default_ttl: 10

rate_limit_configs

This parameter allows specifying a location for the service to lookup various rate limit configurations.

Either a Http or File location can be specified.

Example of what the service expects the location to contain:

{
    "domain": [
        {
            "key": "descriptor_key",
            "value": "descriptor_value",
            "rate_limit": {
                "unit": "<seconds|minutes|hours|days|months|years>",
                "requests_per_unit": 12345
            }
        }
    ]
}

There can be any number of domains and descriptors.