Valve

Node.js HTTP rate limiting plugin. Based on native http module, Frameless !!!

Installation

npm install rate-valve [--save]

Usage

• [x] 接口限流
• [x] 服务限流
• [ ] ip 隔离
• [x] 手动触发
• [x] 拦截器 - filter

Configuration

• enable - Optional, type:Boolean, 是否开启限流，默认 true
• rule - Optional, type:Object, 限流规则, 默认无上限
• rule.api - Optional, type:Array<Object>, API 限流规则
• rule.api.path - Required, type:String, API 路径
• rule.api.method - Required, type:'GET' | 'POST' | 'DELETE' | 'PUT' | 'PATCH' | 'HEAD' | 'OPTIONS', 请求方式
• rule.api.rule - Required, type:Object, 限流规则
• rule.api.rule.limit - Required, type:Number, 窗口时间内请求上限
• rule.server - Optional, type:Object, 服务限流规则
• rule.server.limit - Required, type:Number, 窗口时间内服务请求上限
• rule.ip - Optional, type:Object, IP 限流规则
• rule.ip.enable - Optional, type:Boolean, 是否开启 IP 限流，默认 false
• rule.ip.limit - Required, type:Number, 窗口时间内 IP 请求上限
• rule.ip.storage - Optional, type: String | RedisOptions, IP 限流存储方式，默认内存存储 'memory', if use redis,please see ioredis configuration
• rule.ip.whiteList - Optional, type:Array<String>, IP 白名单, 优先级高于黑名单，白名单内的 IP 不经过限流
• rule.ip.blackList - Optional, type:Array<String>, IP 黑名单, 黑名单内的 IP 强制限流
• interval - Optional, type:Number, 窗口时间，单位秒，默认 60 秒
• message - Optional, type:String, 限流时响应内容，不配置则响应默认内容
• statusCode - Optional, type:Number, 限流时响应的 HTTP 状态码，默认 429
• isSendRetry - Optional, type:Boolean, 限流时是否响应 Retry-After 响应头，默认 false
• requestPropertyName - Optional, type:String, 限流信息绑定在 request 对象上的属性名，默认 valve
• filters - Optional, type:Array<Filter>, 限流拦截器，通过则对当前请求不限流，只要通过一个拦截器即可
• logger-Optional, type:Any, 自定义 Logger，默认 Console
• performance-Optional,type:Object, 性能配置，开启自适应限流的时候需要配置
• performance.enable-Optional, type:Boolean, 是否开启自适应限流，默认 false
• performance.limitThreshold-Optional, type:Number, 超过性能限制次数，当超过该次数，将开启限流，默认 2
• performance.limit-Optional, type:Object, 自适应限流的上限配置
• performance.limit.cpu-Optional, type:Number, cpu 上限，取值 0-1, 默认 1,
• performance.limit.memory-Optional, type:String, 内存上限，支持单位：B, KB, MB, GB，默认 '1GB',
• performance.recoveryThreshold-Optional, type:Number, 低于性能限制次数，当超过该次数，将关闭限流，默认 1
• performance.recovery-Optional, type:Object, 自适应限流的下限配置
• performance.recovery.cpu-Optional, type:Number, cpu 下限，取值 0-1, 默认 0,
• performance.recovery.memory-Optional, type:String, 内存下限，支持单位：B, KB, MB, GB，默认 '0GB',
• debug-Optional, type:Boolean, 是否输出调试日志，默认 false

type Filter = (request: http.IncomingMessage, response: http.ServerResponse) => boolean;

Example:

const options = {
    enable: true,
    rule:{
        server: {
            limit: 300,
        },
        api: [
            {
                method: 'GET',
                path: '/test1',
                rule: { limit: 100 }
            },
            {
                method: 'GET',
                path: '/test2/:id',
                rule: { limit: 150 }
            }
        ],
    },
    interval: 10,
    message: 'Request sent too fast, please fill it out later',
    statusCode: 429,
    isSendRetry: false,
    filters: [
        (request, response) => {
            return request.headers['x-user-id'] == 1;
        }
    ],
    performance: {
        enable: true,
        limitThreshold: 2,
        limit: {
            cpu: 0.8,
            memory: '1GB',
        },
        recoveryThreshold: 1,
        recovery: {
            cpu: 0.2,
            memory: '0GB',
        },
    },
    debug: false,
}

limit in request

通常情况下不需要人为操作，如果想要在代码中手动去触发限流，可参考：

import { Valve } from 'rate-valve';
import * as http from 'http';

// Initialization configuration needs to be done before the http server starts
new Valve(ValveOptions);

http.createServer((request, response) => {
    console.log(request.valve.toJSON());        // { count: { total: 100, api: 10 } }

    if(request.headers['x-user-id'] === 1)
        return request.valve.limit();           // 在请求中手动执行限流

    res.write('hello world /' + req.url);
    res.end();
}).listen(2233);