小伙伴们，你们好呀！我是老寇！一起学习学习gateway限流和熔断降级
一、限流
思考：为啥需要限流？
在一个流量特别大的业务场景中，如果不进行限流，会造成系统宕机，当大批量的请求到达后端服务时，会造成资源耗尽【CPU、内存、线程、网络带宽、数据库连接等是有限的】，进而拖垮系统。
1.常见限流算法
- 漏桶算法
- 令牌桶算法
1.1漏桶算法（不推荐）
1.1.1.原理
将请求缓存到一个队列中，然后以固定的速度处理，从而达到限流的目的
1.1.2.实现
将请求装到一个桶中，桶的容量为固定的一个值，当桶装满之后，就会将请求丢弃掉，桶底部有一个洞，以固定的速率流出。
1.1.3.举例
桶的容量为1W，有10W并发请求，最多只能将1W请求放入桶中，其余请求全部丢弃，以固定的速度处理请求
1.1.4.缺点
处理突发流量效率低（处理请求的速度不变，效率很低）
1.2.令牌桶算法（推荐）
1.2.1.原理
将请求放在一个缓冲队列中，拿到令牌后才能进行处理
1.2.2.实现
装令牌的桶大小固定，当令牌装满后，则不能将令牌放入其中；每次请求都会到桶中拿取一个令牌才能放行，没有令牌时即丢弃请求/继续放入缓存队列中等待
1.2.3.举例
桶的容量为10w个，生产1w个/s，有10W的并发请求，以每秒10W个/s速度处理，随着桶中的令牌很快用完，速度又慢慢降下来啦，而生产令牌的速度趋于一致1w个/s
1.2.4.缺点
处理突发流量提供了系统性能，但是对系统造成了一定的压力，桶的大小不合理，甚至会压垮系统（处理1亿的并发请求，将桶的大小设置为1,这个系统一下就凉凉啦）
2.网关限流（Spring Cloud Gateway + Redis实战）
2.1.pom.xml配置
xml
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-data-redis-reactive</artifactId>
</dependency>
<dependency>
<groupId>org.springframework.cloud</groupId>
<artifactId>spring-cloud-starter-gateway</artifactId>
<exclusions>
<exclusion>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-starter-web</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>org.apache.httpcomponents</groupId>
<artifactId>httpclient</artifactId>
</dependency>2.2.yaml配置
yaml
spring:
application:
name: laokou-gateway
cloud:
gateway:
routes:
- id: LAOKOU-SSO-DEMO
uri: lb://laokou-sso-demo
predicates:
- Path=/sso/**
filters:
- StripPrefix=1
- name: RequestRateLimiter #请求数限流，名字不能乱打
args:
key-resolver: "#{@ipKeyResolver}"
redis-rate-limiter.replenishRate: 1 #生成令牌速率-设为1方便测试
redis-rate-limiter.burstCapacity: 1 #令牌桶容量-设置1方便测试
redis:
database: 0
cluster:
nodes: x.x.x.x:7003,x.x.x.x:7004,x.x.x.x:7005,x.x.x.x:7003,x.x.x.x:7004,x.x.x.x:7005
password: laokou #密码
timeout: 6000ms #连接超时时长（毫秒）
jedis:
pool:
max-active: -1 #连接池最大连接数（使用负值表示无极限）
max-wait: -1ms #连接池最大阻塞等待时间（使用负值表示没有限制）
max-idle: 10 #连接池最大空闲连接
min-idle: 5 #连接池最小空间连接2.3.创建bean
java
@Configuration
public class RequestRateLimiterConfig {
@Bean(value = "ipKeyResolver")
public KeyResolver ipKeyResolver(RemoteAddressResolver remoteAddressResolver) {
return exchange -> Mono.just(remoteAddressResolver.resolve(exchange).getAddress().getHostAddress());
}
@Bean
public RemoteAddressResolver remoteAddressResolver() {
// 远程地址解析器
return XForwardedRemoteAddressResolver.trustAll();
}
}3.测试限流（编写java并发测试）
java
@Slf4j
public class HttpUtil {
public static void apiConcurrent(String url,Map<String,String> params) {
Integer count = 200;
//创建线程池
ThreadPoolExecutor pool = new ThreadPoolExecutor(5, 200, 0L, TimeUnit.SECONDS, new SynchronousQueue<>());
//同步工具
CountDownLatch latch = new CountDownLatch(count);
Map<String,String> dataMap = new HashMap<>(1);
dataMap.put("authorize","XXXXXXX");
for (int i = 0; i < count; i++) {
pool.execute(() -> {
try {
//访问网关的API接口
HttpUtil.doGet("http://localhost:1234/sso/laokou-demo/user",dataMap);
} catch (IOException e) {
e.printStackTrace();
}finally {
latch.countDown();
}
});
}
try {
latch.await();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
public static String doGet(String url, Map<String, String> params) throws IOException {
//创建HttpClient对象
CloseableHttpClient httpClient = HttpClients.createDefault();
String resultString = "";
CloseableHttpResponse response = null;
try {
//创建uri
URIBuilder builder = new URIBuilder(url);
if (!params.isEmpty()) {
for (Map.Entry<String, String> entry : params.entrySet()) {
builder.addParameter(entry.getKey(), entry.getValue());
}
}
URI uri = builder.build();
//创建http GET请求
HttpGet httpGet = new HttpGet(uri);
List<NameValuePair> paramList = new ArrayList<>();
RequestBuilder requestBuilder = RequestBuilder.get().setUri(new URI(url));
requestBuilder.setEntity(new UrlEncodedFormEntity(paramList, Consts.UTF_8));
httpGet.setHeader(new BasicHeader("Content-Type", "application/json;charset=UTF-8"));
httpGet.setHeader(new BasicHeader("Accept", "*/*;charset=utf-8"));
//执行请求
response = httpClient.execute(httpGet);
//判断返回状态是否是200
if (response.getStatusLine().getStatusCode() == 200) {
resultString = EntityUtils.toString(response.getEntity(), "UTF-8");
}
} catch (Exception e) {
log.info("调用失败:{}",e);
} finally {
if (response != null) {
response.close();
}
httpClient.close();
}
log.info("打印：{}",resultString);
return resultString;
}
}
说明这个网关限流配置是没有问题的
4.源码查看
Spring Cloud Gateway RequestRateLimiter GatewayFilter Factory文档地址
工厂 RequestRateLimiter GatewayFilter使用一个RateLimiter实现来判断当前请求是否被允许继续。如果不允许，HTTP 429 - Too Many Requests则返回默认状态。
4.1.查看 RequestRateLimiterGatewayFilterFactory
java
@Override
public GatewayFilter apply(Config config) {
KeyResolver resolver = getOrDefault(config.keyResolver, defaultKeyResolver);
RateLimiter<Object> limiter = getOrDefault(config.rateLimiter, defaultRateLimiter);
boolean denyEmpty = getOrDefault(config.denyEmptyKey, this.denyEmptyKey);
HttpStatusHolder emptyKeyStatus = HttpStatusHolder
.parse(getOrDefault(config.emptyKeyStatus, this.emptyKeyStatusCode));
return (exchange, chain) -> resolver.resolve(exchange).defaultIfEmpty(EMPTY_KEY).flatMap(key -> {
if (EMPTY_KEY.equals(key)) {
if (denyEmpty) {
setResponseStatus(exchange, emptyKeyStatus);
return exchange.getResponse().setComplete();
}
return chain.filter(exchange);
}
String routeId = config.getRouteId();
if (routeId == null) {
Route route = exchange.getAttribute(ServerWebExchangeUtils.GATEWAY_ROUTE_ATTR);
routeId = route.getId();
}
// 执行限流
return limiter.isAllowed(routeId, key).flatMap(response -> {
for (Map.Entry<String, String> header : response.getHeaders().entrySet()) {
exchange.getResponse().getHeaders().add(header.getKey(), header.getValue());
}
if (response.isAllowed()) {
return chain.filter(exchange);
}
setResponseStatus(exchange, config.getStatusCode());
return exchange.getResponse().setComplete();
});
});
}4.2.查看 RedisRateLimiter
java
@Override
@SuppressWarnings("unchecked")
public Mono<Response> isAllowed(String routeId, String id) {
if (!this.initialized.get()) {
throw new IllegalStateException("RedisRateLimiter is not initialized");
}
// 这里如何加载配置?请思考
Config routeConfig = loadConfiguration(routeId);
// 令牌桶每秒产生令牌数量
int replenishRate = routeConfig.getReplenishRate();
// 令牌桶容量
int burstCapacity = routeConfig.getBurstCapacity();
// 请求消耗的令牌数
int requestedTokens = routeConfig.getRequestedTokens();
try {
// 键
List<String> keys = getKeys(id);
// 参数
List<String> scriptArgs = Arrays.asList(replenishRate + "", burstCapacity + "", "", requestedTokens + "");
// 调用lua脚本
Flux<List<Long>> flux = this.redisTemplate.execute(this.script, keys, scriptArgs);
return flux.onErrorResume(throwable -> {
if (log.isDebugEnabled()) {
log.debug("Error calling rate limiter lua", throwable);
}
return Flux.just(Arrays.asList(1L, -1L));
}).reduce(new ArrayList<Long>(), (longs, l) -> {
longs.addAll(l);
return longs;
}).map(results -> {
// 判断是否等于1，1表示允许通过，0表示不允许通过
boolean allowed = results.get(0) == 1L;
Long tokensLeft = results.get(1);
Response response = new Response(allowed, getHeaders(routeConfig, tokensLeft));
if (log.isDebugEnabled()) {
log.debug("response: " + response);
}
return response;
});
}
catch (Exception e) {
log.error("Error determining if user allowed from redis", e);
}
return Mono.just(new Response(true, getHeaders(routeConfig, -1L)));
}
static List<String> getKeys(String id) {
String prefix = "request_rate_limiter.{" + id;
String tokenKey = prefix + "}.tokens";
String timestampKey = prefix + "}.timestamp";
return Arrays.asList(tokenKey, timestampKey);
}思考：redis限流配置是如何加载？
其实就是监听动态路由的事件并把配置存起来
4.3.重点来了，令牌桶 /META-INF/scripts/request_rate_limiter.lua 脚本剖析
lua
-- User Request Rate Limiter filter
-- See https://stripe.com/blog/rate-limiters
-- See https://gist.github.com/ptarjan/e38f45f2dfe601419ca3af937fff574d#file-1-check_request_rate_limiter-rb-L11-L34
-- 令牌桶算法工作原理
-- 1.系统以恒定速率往桶里面放入令牌
-- 2.请求需要被处理，则需要从桶里面获取一个令牌
-- 3.如果桶里面没有令牌可获取，则可以选择等待或直接拒绝并返回
-- 令牌桶算法工作流程
-- 1.计算填满令牌桶所需要的时间（填充时间 = 桶容量 / 速率）
-- 2.设置存储数据的TTL（过期时间），为填充时间的两倍（存储时间 = 填充时间 * 2）
-- 3.从Redis获取当前令牌的剩余数量和上一次调用的时间戳
-- 4.计算距离上一次调用的时间间隔（时间间隔 = 当前时间 - 上一次调用时间）
-- 5.计算填充的令牌数量（填充令牌数量 = 时间间隔 * 速率）【前提：桶容量是固定的，不存在无限制的填充】
-- 6.判断是否有足够多的令牌满足请求【 (填充令牌数量 + 剩余令牌数量) >= 请求数量 && (填充令牌数量 + 剩余令牌数量) <= 桶容量 】
-- 7.如果请求被允许，则从桶里面取出相应数据的令牌
-- 8.如果TTL为正，则更新Redis键中的令牌和时间戳
-- 9.返回两个两个参数（allowed_num：请求被允许标志。1允许，0不允许）、（new_tokens：填充令牌后剩余的令牌数据）
-- 随机写入
redis.replicate_commands()
-- 令牌桶Key -> 存储当前可用令牌的数量（剩余令牌数量）
local tokens_key = KEYS[1]
-- 时间戳Key -> 存储上次令牌刷新的时间戳
local timestamp_key = KEYS[2]
-- 令牌填充速率
local rate = tonumber(ARGV[1])
-- 令牌桶容量
local capacity = tonumber(ARGV[2])
-- 当前时间
local now = tonumber(ARGV[3])
-- 请求数量
local requested = tonumber(ARGV[4])
-- 填满令牌桶所需要的时间
local fill_time = capacity / rate
-- 设置key的过期时间（填满令牌桶所需时间的2倍）
local ttl = math.floor(fill_time * 2)
-- 判断当前时间，为空则从redis获取
if now == nil then
now = redis.call('TIME')[1]
end
-- 获取当前令牌的剩余数量
local last_tokens = tonumber(redis.call("get", tokens_key))
if last_tokens == nil then
last_tokens = capacity
end
-- 获取上一次调用的时间戳
local last_refreshed = tonumber(redis.call('get', timestamp_key))
if last_refreshed == nil then
last_refreshed = 0
end
-- 计算距离上一次调用的时间间隔
local delta = math.max(0, now - last_refreshed)
-- 当前的令牌数量（剩余 + 填充 <= 桶容量）
local now_tokens = math.min(capacity, last_refreshed + (rate * delta))
-- 判断是否有足够多的令牌满足请求
local allowed = now_tokens >= requested
-- 定义当前令牌的剩余数量
local new_tokens = now_tokens
-- 定义被允许标志
local allowed_num = 0
if allowed then
new_tokens = now_tokens - requested
-- 允许访问
allowed_num = 1
end
-- ttl > 0，将当前令牌的剩余数量和当前时间戳存入redis
if ttl > 0 then
redis.call('setex', tokens_key, ttl, new_tokens)
redis.call('setex', timestamp_key, ttl, now)
end
-- 返回参数
return { allowed_num, new_tokens }4.4.查看 GatewayRedisAutoConfiguration 脚本初始化
java
@Bean
@SuppressWarnings("unchecked")
public RedisScript redisRequestRateLimiterScript() {
DefaultRedisScript redisScript = new DefaultRedisScript<>();
redisScript.setScriptSource(
// 根据指定路径获取lua脚本来初始化配置
new ResourceScriptSource(new ClassPathResource("META-INF/scripts/request_rate_limiter.lua")));
redisScript.setResultType(List.class);
return redisScript;
}
@Bean
@ConditionalOnMissingBean
public RedisRateLimiter redisRateLimiter(ReactiveStringRedisTemplate redisTemplate,
@Qualifier(RedisRateLimiter.REDIS_SCRIPT_NAME) RedisScript<List<Long>> redisScript,
ConfigurationService configurationService) {
return new RedisRateLimiter(redisTemplate, redisScript, configurationService);
}思考：请求限流过滤器是如何开启？
1.通过yaml配置开启
yaml
spring:
cloud:
gateway:
server:
webflux:
filter:
request-rate-limiter:
enabled: true2.GatewayAutoConfiguration自动注入bean
java
@Bean
@ConditionalOnBean({ RateLimiter.class, KeyResolver.class })
@ConditionalOnEnabledFilter
public RequestRateLimiterGatewayFilterFactory requestRateLimiterGatewayFilterFactory(RateLimiter rateLimiter,
KeyResolver resolver) {
return new RequestRateLimiterGatewayFilterFactory(rateLimiter, resolver);
}重点来了，真正加载这个bean的是 @ConditionalOnEnabledFilter 注解进行判断
java
@Retention(RetentionPolicy.RUNTIME)
@Target({ ElementType.TYPE, ElementType.METHOD })
@Documented
@Conditional(OnEnabledFilter.class)
public @interface ConditionalOnEnabledFilter {
// 这里value是用来指定满足条件的某些类，换一句话说，就是这些类都加载或注入到ioc容器，这个注解修饰的自动装配类才会生效
Class<? extends GatewayFilterFactory<?>> value() default OnEnabledFilter.DefaultValue.class;
}我们继续跟进代码，查看@Conditional(OnEnabledFilter.class)
众所周知，@Conditional可以用来加载满足条件的bean，所以，我们分析一下OnEnabledFilter
java
public class OnEnabledFilter extends OnEnabledComponent<GatewayFilterFactory<?>> {}我分析它的父类，这里有你想要的答案！
java
public abstract class OnEnabledComponent<T> extends SpringBootCondition implements ConfigurationCondition {
private static final String PREFIX = "spring.cloud.gateway.server.webflux.";
private static final String SUFFIX = ".enabled";
private ConditionOutcome determineOutcome(Class<? extends T> componentClass, PropertyResolver resolver) {
// 拼接完整名称
// 例如 => spring.cloud.gateway.server.webflux.request-rate-limiter.enabled
String key = PREFIX + normalizeComponentName(componentClass) + SUFFIX;
ConditionMessage.Builder messageBuilder = forCondition(annotationClass().getName(), componentClass.getName());
if ("false".equalsIgnoreCase(resolver.getProperty(key))) {
// 不满足条件不加载bean
return ConditionOutcome.noMatch(messageBuilder.because("bean is not available"));
}
// 满足条件加载bean
return ConditionOutcome.match();
}
}5.优化限流响应[使用全限定类名直接覆盖类]
小伙伴们，有没有发现，这个这个响应体封装的不太好，因此，我们来自定义吧，我们直接覆盖类，代码修改如下
java
@Getter
@ConfigurationProperties("spring.cloud.gateway.server.webflux.filter.request-rate-limiter")
public class RequestRateLimiterGatewayFilterFactory
extends AbstractGatewayFilterFactory<RequestRateLimiterGatewayFilterFactory.Config> {
private static final String EMPTY_KEY = "____EMPTY_KEY__";
private final RateLimiter<?> defaultRateLimiter;
private final KeyResolver defaultKeyResolver;
/**
* Switch to deny requests if the Key Resolver returns an empty key, defaults to true.
*/
@Setter
private boolean denyEmptyKey = true;
/** HttpStatus to return when denyEmptyKey is true, defaults to FORBIDDEN. */
@Setter
private String emptyKeyStatusCode = HttpStatus.FORBIDDEN.name();
public RequestRateLimiterGatewayFilterFactory(RateLimiter<?> defaultRateLimiter, KeyResolver defaultKeyResolver) {
super(Config.class);
this.defaultRateLimiter = defaultRateLimiter;
this.defaultKeyResolver = defaultKeyResolver;
}
@Override
public GatewayFilter apply(Config config) {
KeyResolver resolver = getOrDefault(config.keyResolver, defaultKeyResolver);
RateLimiter<?> limiter = getOrDefault(config.rateLimiter, defaultRateLimiter);
boolean denyEmpty = getOrDefault(config.denyEmptyKey, this.denyEmptyKey);
HttpStatusHolder emptyKeyStatus = HttpStatusHolder
.parse(getOrDefault(config.emptyKeyStatus, this.emptyKeyStatusCode));
return (exchange, chain) -> resolver.resolve(exchange).defaultIfEmpty(EMPTY_KEY).flatMap(key -> {
if (EMPTY_KEY.equals(key)) {
if (denyEmpty) {
setResponseStatus(exchange, emptyKeyStatus);
return exchange.getResponse().setComplete();
}
return chain.filter(exchange);
}
String routeId = config.getRouteId();
if (routeId == null) {
Route route = exchange.getAttribute(ServerWebExchangeUtils.GATEWAY_ROUTE_ATTR);
Assert.notNull(route, "Route is null");
routeId = route.getId();
}
return limiter.isAllowed(routeId, key).flatMap(response -> {
for (Map.Entry<String, String> header : response.getHeaders().entrySet()) {
exchange.getResponse().getHeaders().add(header.getKey(), header.getValue());
}
if (response.isAllowed()) {
return chain.filter(exchange);
}
// 主要修改这行
return responseOk(exchange, Result.fail("Too_Many_Requests", "请求太频繁"));
});
});
}
private Mono<Void> responseOk(ServerWebExchange exchange, Object data) {
return responseOk(exchange, JacksonUtils.toJsonStr(data), MediaType.APPLICATION_JSON);
}
private Mono<Void> responseOk(ServerWebExchange exchange, String str, MediaType contentType) {
DataBuffer buffer = exchange.getResponse().bufferFactory().wrap(str.getBytes(StandardCharsets.UTF_8));
ServerHttpResponse response = exchange.getResponse();
response.setStatusCode(HttpStatus.OK);
response.getHeaders().setContentType(contentType);
response.getHeaders().setContentLength(str.getBytes(StandardCharsets.UTF_8).length);
return response.writeWith(Flux.just(buffer));
}
private <T> T getOrDefault(T configValue, T defaultValue) {
return (configValue != null) ? configValue : defaultValue;
}
public static class Config implements HasRouteId {
@Getter
private KeyResolver keyResolver;
@Getter
private RateLimiter<?> rateLimiter;
@Getter
private HttpStatus statusCode = HttpStatus.TOO_MANY_REQUESTS;
@Getter
private Boolean denyEmptyKey;
@Getter
private String emptyKeyStatus;
private String routeId;
public Config setKeyResolver(KeyResolver keyResolver) {
this.keyResolver = keyResolver;
return this;
}
public Config setRateLimiter(RateLimiter<?> rateLimiter) {
this.rateLimiter = rateLimiter;
return this;
}
public Config setStatusCode(HttpStatus statusCode) {
this.statusCode = statusCode;
return this;
}
public Config setDenyEmptyKey(Boolean denyEmptyKey) {
this.denyEmptyKey = denyEmptyKey;
return this;
}
public Config setEmptyKeyStatus(String emptyKeyStatus) {
this.emptyKeyStatus = emptyKeyStatus;
return this;
}
@Override
public void setRouteId(String routeId) {
this.routeId = routeId;
}
@Override
public String getRouteId() {
return this.routeId;
}
}
}二、熔断降级
思考：为什么需要熔断降级？
111111
我是老寇，我们下次再见啦！