前言
相信使用过WebApiThrottle的童鞋对AspNetCoreRateLimit应该不陌生,AspNetCoreRateLimit是一个ASP.NET Core速率限制的解决方案,旨在控制客户端根据IP地址或客户端ID向Web API或MVC应用发出的请求的速率。AspNetCoreRateLimit包含一个IpRateLimitMiddleware和ClientRateLimitMiddleware,每个中间件可以根据不同的场景配置限制允许IP或客户端,自定义这些限制策略,也可以将限制策略应用在每个API URL或具体的HTTP Method上。
实践
起初是因为新做的项目中,有天查询日志发现,对外的几个公共接口经常被“恶意”调用,考虑到接口安全性问题,增加限流策略。
AspNetCoreRateLimit GayHub:https://github.com/stefanprodan/AspNetCoreRateLimit
根据IP进行限流
通过nuget安装AspNetCoreRateLimit,当前版本是3.0.5,因为实际项目中用的都是分布式缓存,在这里不用内存存储,而是结合Redis进行使用,内存存储直接参考官方的Wiki就可以了。
Install-Package AspNetCoreRateLimit
Install-Package Microsoft.Extensions.Caching.Redis
在Startup.ConfigureServices中将服务和其他依赖注入
public void ConfigureServices(IServiceCollection services) { #region MVC services.AddMvc( options => { options.UseCentralRoutePrefix(new RouteAttribute("api/")); } ).SetCompatibilityVersion(CompatibilityVersion.Version_2_2); #endregion services.AddDistributedRedisCache(options => { options.Configuration = "127.0.0.1:6379,password=123456,connectTimeout=5000,syncTimeout=10000"; options.InstanceName = "WebRatelimit"; }); //加载配置 services.AddOptions(); //从appsettings.json获取相应配置 services.Configure<IpRateLimitOptions>(Configuration.GetSection("IpRateLimiting")); //注入计数器和规则存储 services.AddSingleton<IIpPolicyStore, DistributedCacheIpPolicyStore>(); services.AddSingleton<IRateLimitCounterStore, DistributedCacheRateLimitCounterStore>(); services.AddSingleton<IHttpContextAccessor, HttpContextAccessor>(); //配置(计数器密钥生成器) services.AddSingleton<IRateLimitConfiguration, RateLimitConfiguration>(); }
在Startup.Configure启用
// This method gets called by the runtime. Use this method to configure the HTTP request pipeline. public void Configure(IApplicationBuilder app, IHostingEnvironment env) { if (env.IsDevelopment()) { app.UseDeveloperExceptionPage(); } else { app.UseHsts(); } //启用限流,需在UseMvc前面 app.UseIpRateLimiting(); app.UseMvc(); }
为了不影响appsettings.json的美观吧,可以新建一个RateLimitConfig.json,并Program中启动加载中增加
public static IWebHostBuilder CreateWebHostBuilder(string[] args) => WebHost.CreateDefaultBuilder(args) .UseStartup<Startup>().ConfigureAppConfiguration((host,config)=> { config.AddJsonFile($"RateLimitConfig.json", optional: true, reloadOnChange: true); });
RateLimitConfig.json 配置如下:
{ "IpRateLimiting": { //false则全局将应用限制,并且仅应用具有作为端点的规则* 。 true则限制将应用于每个端点,如{HTTP_Verb}{PATH} "EnableEndpointRateLimiting": true, //false则拒绝的API调用不会添加到调用次数计数器上 "StackBlockedRequests": false,
//注意这个配置,表示获取用户端的真实IP,我们的线上经过负载后是 X-Forwarded-For,而测试服务器没有,所以是X-Real-IP "RealIpHeader": "X-Real-IP", "ClientIdHeader": "X-ClientId", "HttpStatusCode": 200, "QuotaExceededResponse": { "Content": "{{"code":429,"msg":"访问过于频繁,请稍后重试","data":null}}", "ContentType": "application/json", "StatusCode": 200 }, "IpWhitelist": [ ], "EndpointWhitelist": [], "ClientWhitelist": [], "GeneralRules": [ { "Endpoint": "*:/api/values/test", "Period": "5s", "Limit": 3 } ] } }
重要配置说明:
QuotaExceededResponse 是自定义返回的内容,所以必须设置HttpStatusCode和StatusCode为200。
GeneralRules是具体的策略,根据不同需求配置不同端点即可, Period的单位可以是s, m, h, d,Limint是单位时间内的允许访问的次数;
IpWhitelist是IP白名单,本地调试或者UAT环境,可以加入相应的IP,略过策略的限制;
EndpointWhitelist是端点白名单,如果全局配置了访问策略,设置端点白名单相当于IP白名单一样,略过策略的限制;
其他配置项请参考Wiki:https://github.com/stefanprodan/AspNetCoreRateLimit/wiki/IpRateLimitMiddleware#setup
Fiddler开始测试
测试接口:http://127.0.0.1:5000/api/values/Test
[HttpGet] public object test() { return "ok"; }
调用结果:
调用次数和剩余调用次数在Head可以看到,(吃我一个链接:https://www.cnblogs.com/EminemJK/p/12720691.html)
如果调用超过策略后,调用失败,返回我们自定义的内容
在Redis客户端可以看到策略的一些情况,
其他
通常在项目中,Authorization授权是少不了了,加入限流后,在被限流的接口调用后,限流拦截器使得跨域策略失效,故重写拦截器中间件,继承IpRateLimitMiddleware 即可:
public class IPLimitMiddleware : IpRateLimitMiddleware { public IPLimitMiddleware(RequestDelegate next, IOptions<IpRateLimitOptions> options, IRateLimitCounterStore counterStore, IIpPolicyStore policyStore, IRateLimitConfiguration config, ILogger<IpRateLimitMiddleware> logger) : base(next, options, counterStore, policyStore, config, logger) { } public override Task ReturnQuotaExceededResponse(HttpContext httpContext, RateLimitRule rule, string retryAfter) { httpContext.Response.Headers.Append("Access-Control-Allow-Origin", "*"); return base.ReturnQuotaExceededResponse(httpContext, rule, retryAfter); } }
然后修改Startup.Configure,
//启用限流,需在UseMvc前面 //app.UseIpRateLimiting(); app.UseMiddleware<IPLimitMiddleware>(); app.UseMvc();
特别需要注意的坑是,在其他文章的教程中,他们会写成:
app.UseMiddleware<IPLimitMiddleware>().UseIpRateLimiting();//错误的演示 https://www.cnblogs.com/EminemJK/p/12720691.html
这些写你测试的时候会发现,
X-Rate-Limit-Remaining 递减量会变成2,而正常的递减量应该是1,举栗子,配置如下:
"Endpoint": "*:/api/values/test", "Period": "3s", "Limit": 1
表示3秒内可以访问的次数是一次,当发生调用的时候会直接返回被限制的提示,而不能正常访问接口,原因就是因为调用了两次中间件。
最后
AspNetCoreRateLimit还可以根据客户端ID进行配置策略,具体可以看一下官方的Wiki吧。
#2020-06-09 遇到的BUG,并解决
2020-05-21 02:51:42,900 [10] ERROR System.OperationCanceledException: The operation was canceled. at System.Threading.CancellationToken.ThrowOperationCanceledException() at Microsoft.Extensions.Caching.Redis.RedisCache.RefreshAsync(String key, Nullable`1 absExpr, Nullable`1 sldExpr, CancellationToken token) at Microsoft.Extensions.Caching.Redis.RedisCache.GetAndRefreshAsync(String key, Boolean getData, CancellationToken token) at Microsoft.Extensions.Caching.Redis.RedisCache.GetAsync(String key, CancellationToken token) at Microsoft.Extensions.Caching.Distributed.DistributedCacheExtensions.GetStringAsync(IDistributedCache cache, String key, CancellationToken token) at AspNetCoreRateLimit.DistributedCacheRateLimitStore`1.GetAsync(String id, CancellationToken cancellationToken) at AspNetCoreRateLimit.IpRateLimitProcessor.GetMatchingRulesAsync(ClientRequestIdentity identity, CancellationToken cancellationToken) at AspNetCoreRateLimit.RateLimitMiddleware`1.Invoke(HttpContext context) at Microsoft.AspNetCore.Authentication.AuthenticationMiddleware.Invoke(HttpContext context) at Microsoft.AspNetCore.StaticFiles.StaticFileMiddleware.Invoke(HttpContext context) at DigitalCertificateSystem.Handlers.ExceptionHandlerMiddleWare.Invoke(HttpContext context)
也不清楚是什么时候会出现这个问题,在QPS达到一定的时候,访问网站非常非常慢,同一服务器的后台管理却没有问题,最终发现是这个组件的问题。于是,改造:
一、实现 IRateLimitStore<T>,引入自己的组件
/// <summary> /// 重写使用我们的自己的redis /// </summary> /// <typeparam name="T"></typeparam> public class RateLimitStore<T> : IRateLimitStore<T> { public Task SetAsync(string id, T entry, TimeSpan? expirationTime = null, CancellationToken cancellationToken = default) { return RedisUtils.StringSetAsync(RedisKEY.UserBehaviorCache, id, JsonConvert.SerializeObject(entry), expirationTime); } public async Task<bool> ExistsAsync(string id, CancellationToken cancellationToken = default) { var stored = await RedisUtils.StringGetAsync(RedisKEY.UserBehaviorCache, id); return !string.IsNullOrEmpty(stored); } public async Task<T> GetAsync(string id, CancellationToken cancellationToken = default) { var stored = await RedisUtils.StringGetAsync(RedisKEY.UserBehaviorCache, id); if (!string.IsNullOrEmpty(stored)) { return JsonConvert.DeserializeObject<T>(stored); } return default; } public Task RemoveAsync(string id, CancellationToken cancellationToken = default) { return RedisUtils.KeyDeleteAsync(RedisKEY.UserBehaviorCache, id); } }
二、实现 IRateLimitCounterStore
public class RedisCounterStore: RateLimitStore<RateLimitCounter?>, IRateLimitCounterStore { }
三、实现 IIpPolicyStore
public class RedisIpPolicyStore : RateLimitStore<IpRateLimitPolicies>, IIpPolicyStore { private readonly IpRateLimitOptions _options; private readonly IpRateLimitPolicies _policies; public RedisIpPolicyStore( IOptions<IpRateLimitOptions> options = null, IOptions<IpRateLimitPolicies> policies = null) { _options = options?.Value; _policies = policies?.Value; } public async Task SeedAsync() { // on startup, save the IP rules defined in appsettings if (_options != null && _policies != null) { await SetAsync($"{_options.IpPolicyPrefix}", _policies).ConfigureAwait(false); } } }
四、修改 Startup
services.AddOptions(); services.Configure<IpRateLimitOptions>(Configuration.GetSection("IpRateLimiting")); services.AddSingleton<IIpPolicyStore, RedisIpPolicyStore>(); services.AddSingleton<IRateLimitCounterStore, RedisCounterStore>(); services.AddSingleton<IRateLimitConfiguration, RateLimitConfiguration>();
搞定
2020-07-09
Redis使用的是 StackExchange.Redis,帮助类封装在 https://github.com/EminemJK/Banana,
直接nuget搜索 Banana.Utility
PM>Install-Package Banana.Utility
本文已独家授权给xLong设计(ID:xlongsheji)公众号发布