Skip to content

Latest commit

 

History

History
84 lines (68 loc) · 3.42 KB

README.md

File metadata and controls

84 lines (68 loc) · 3.42 KB

Limitador (server)

Docker Repository on Quay

By default, Limitador starts the HTTP server in localhost:8080, and the grpc service that implements the Envoy Rate Limit protocol in localhost:8081. That can be configured with these ENVs: ENVOY_RLS_HOST, ENVOY_RLS_PORT, HTTP_API_HOST, and HTTP_API_PORT.

Or using the command line arguments:

Rate Limiting Server

Usage: limitador-server [OPTIONS] <LIMITS_FILE> [STORAGE]

STORAGES:
  memory        Counters are held in Limitador (ephemeral)
  disk          Counters are held on disk (persistent)
  redis         Uses Redis to store counters
  redis_cached  Uses Redis to store counters, with an in-memory cache

Arguments:
  <LIMITS_FILE>  The limit file to use

Options:
  -b, --rls-ip <ip>
          The IP to listen on for RLS [default: 0.0.0.0]
  -p, --rls-port <port>
          The port to listen on for RLS [default: 8081]
  -B, --http-ip <http_ip>
          The IP to listen on for HTTP [default: 0.0.0.0]
  -P, --http-port <http_port>
          The port to listen on for HTTP [default: 8080]
  -l, --limit-name-in-labels
          Include the Limit Name in prometheus label
  -v...
          Sets the level of verbosity
      --tracing-endpoint <tracing_endpoint>
          The endpoint for the tracing service
      --validate
          Validates the LIMITS_FILE and exits
  -H, --rate-limit-headers <rate_limit_headers>
          Enables rate limit response headers [default: NONE] [possible values: NONE, DRAFT_VERSION_03]
  -h, --help
          Print help
  -V, --version
          Print version

When using environment variables, these will override the defaults. While environment variable are themselves overridden by the command line arguments provided. See the individual STORAGES help for more options relative to each of the storages.

The OpenAPI spec of the HTTP service is here.

Limitador has to be started with a YAML file that has some limits defined. There's an example file that allows 10 requests per minute and per user_id when the HTTP method is "GET" and 5 when it is a "POST". You can run it with Docker (replace latest with the version you want):

docker run --rm --net=host -it -v $(pwd)/examples/limits.yaml:/home/limitador/my_limits.yaml:ro quay.io/kuadrant/limitador:latest limitador-server /home/limitador/my_limits.yaml

You can also use the YAML file when running locally:

cargo run --release --bin limitador-server ./examples/limits.yaml

If you want to use Limitador with Envoy, there's a minimal Envoy config for testing purposes here. The config forwards the "userid" header and the request method to Limitador. It assumes that there's an upstream API deployed on port 1323. You can use echo, for example.

Limitador has several options that can be configured via ENV. This doc specifies them.

Limits storage

Limitador can store its limits and counters in-memory, disk or in Redis. In-memory is faster, but the limits are applied per instance. When using Redis, multiple instances of Limitador can share the same limits, but it's slower.