sliding-window-rate-limiter

sliding-window-rate-limiter

Sliding window rate limiter with Redis >= 3.2 backend or in-memory backend.

Requirements

This module requires ES6 with Node >= 16.

Redis >= 3.2.0 is required for the Redis backend.

Installation

npm install sliding-window-rate-limiter

Additionally for Typescript:

npm install -D @types/node @types/ioredis

Usage

const {SlidingWindowRateLimiter} = require("sliding-window-rate-limiter")

Typescript:

import SlidingWindowRateLimiter from "sliding-window-rate-limiter"
// or
import {SlidingWindowRateLimiter} from "sliding-window-rate-limiter"

constructor

const limiter = SlidingWindowRateLimiter.createLimiter(options)

Options:

• interval is a number of milliseconds in a sliding window
• redis is an instance of ioredis or URL string to Redis server (only for Redis backend)
• operationTimeout is the time in milliseconds after Redis operation is canceled (for Redis and SafeRedis backends, optional)
• safe: true (only for SafeRedis backend) reuseRedisAfter is the time (milliseconds) to reconnect to the Redis server after connection failure (only for SafeRedis backend, default value: 2000 milliseconds)

If redis parameter is a string then a new ioredis object is created with retryStrategy set to 1 second and maxRetriesPerRequest set to 1.

Example:

const limiter = SlidingWindowRateLimiter.createLimiter({
  interval: 60000,
})

or

const limiter = SlidingWindowRateLimiter.createLimiter({
  interval: 60000,
  redis: new Redis({
    host: "redis-server",
    retryStrategy: _times => 1000,
    maxRetriesPerRequest: 1,
  }),
  safe: true,
})

check

const result = await limiter.check(key, limit)
const {usage, reset} = result

Checks current usage for key. If usage is equal to or above limit, additionally sets reset time in milliseconds.

reserve

const result = await limiter.reserve(key, limit)
const {token, usage, reset} = result

Makes a reservation and returns token with a reservation. If usage is equal to or above limit, additionally sets reset time in milliseconds. Throws an error if has occurred.

cancel

const result = await limiter.cancel(key, token)
const {canceled} = result

Cancels a reservation for token and returns the number of `canceled`` tokens. It is a zero if no token previously was reserved or it was expired.

destroy

limiter.destroy()

Frees resources used by limiter (timers and Redis connection if created by limiter itself).

Errors

If reserve or usage methods return an error:

ERR Error running script (call to f_8ff6a0f745b738fe1d9fa74079c4c13d032e9947): @user_script:1: user_script:1: attempt to call field \'replicate_commands\' (a nil value)

then check if Redis has the proper version (>= 3.2.0).

Backends

Memory

This backend holds all data in memory.

Redis

This backend requires Redis 3.2 to work. The main advantage is that the state of the limiter can be shared between many clients.

SafeRedis

There is an extended version of the limiter, which behaves gracefully when the Redis server is unavailable for any reason. In case of Redis connection failure, SafeRedis backend will always return a positive response object and will try to use again the Redis server after reuseRedisAfter.

License

Copyright (c) 2017-2024 Piotr Roszatycki piotr.roszatycki@gmail.com

MIT

Based on Lua script from https://github.com/3hedgehogs/inredis-ratelimiter

Copyright (c) 2017 Serguei Poliakov serguei.poliakov@gmail.com

Inspired by ClassDojo blog note https://engineering.classdojo.com/blog/2015/02/06/rolling-rate-limiter/ and rolling-rate-limiter module.