spring cloud 熔断、限流、服务降级、负载均衡总结

springcloud 微服务网关

zuul.strip-prefix=true
zuul.routes.postgres-client.path =/ps01/**
zuul.routes.postgres-client.service-id=postgres-client
zuul.routes.postgres-client02.path=/ps02/**
zuul.routes.postgres-client02.service-id=postgres-client02
zuul.semaphore.max-semaphores=100
zuul.ribbon-isolation-strategy=semaphore

@Component
public class MyRateLimiter extends ZuulFilter {//  令牌桶static RateLimiter rateLimiter= RateLimiter.create(8, 2, TimeUnit.SECONDS);@Overridepublic String filterType() {return PRE_TYPE;}@Overridepublic int filterOrder() {return SERVLET_DETECTION_FILTER_ORDER - 1;}@Overridepublic boolean shouldFilter() {return true;}@Overridepublic Object run() throws ZuulException {RequestContext context = RequestContext.getCurrentContext();boolean allowed = rateLimiter.tryAcquire();if (!allowed) {context.setSendZuulResponse(false);context.setResponseStatusCode(501);}return null;}
}

ribbon.eureka.enabled=true
ribbon.ReadTimeout = 10000
ribbon.ConnectTimeout= 10000
ribbon.MaxAutoRetries= 0
ribbon.MaxAutoRetriesNextServer=1
ribbon.OkToRetryOnAllOperations=false
spring.cloud.sentinel.transport.port=8719
spring.cloud.sentinel.transport.dashboard=192.168.1.4:8080
nacos-gateway.ribbon.NFLoadBalancerRuleClassName=com.netflix.loadbalancer.RandomRule
spring.cloud.gateway.discovery.locator.enabled=true
spring.cloud.gateway.routes[0].id=order-servicespring.cloud.gateway.routes[0].predicates[0]=Path=/order/**
spring.cloud.gateway.routes[0].uri=lb://order-service
spring.cloud.gateway.redis-rate-limiter.include-headers=true
spring.cloud.gateway.routes[0].filters[0]=RewritePath=/order(?<segment>/?.*), $\{segment}spring.cloud.gateway.routes[0].filters[0].name=RequestRateLimiter
# 如果返回的key是空的话，则不进行限流
spring.cloud.gateway.routes[0].filters[0].args[0]=deny-empty-key=false
# 每秒产生多少个令牌
spring.cloud.gateway.routes[0].filters[0].args[1]=redis-rate-limiter.replenishRate=0# 1秒内最大的令牌，即在1s内可以允许的突发流程，设置为0，表示阻止所有的请求
spring.cloud.gateway.routes[0].filters[0].args[2]=redis-rate-limiter.burstCapacity=0# 每次请求申请几个令牌spring.cloud.gateway.routes[0].filters[0].args[3]=redis-rate-limiter.requestedTokens=0
# 自定义限流规则
spring.cloud.gateway.routes[0].filters[0].args[0]=rate-limiter="#{@defaultGatewayRateLimiter}"
# 返回限流的key
spring.cloud.gateway.routes[0].filters[0].args[1]=key-resolver="#{@defaultGatewayKeyResolver}"
spring.cloud.gateway.routes[0].filters[0].args[2]=deny-empty-key=false
# 限流后向客户端返回的响应码429,请求太多
spring.cloud.gateway.routes[0].filters[0].args[3]=status-code=TOO_MANY_REQUESTS
# 每次请求申请几个令牌  default-gateway-rate-limiter 的值是在 defaultGatewayRateLimiter 中定义的
spring.cloud.gateway.routes[0].filters[0].args[4]=default-gateway-rate-limiter.requestedTokens=1

1.使用hystrix

@HystrixCommand(commandKey = "saveStudent",commandProperties = {@HystrixProperty(name="execution.isolation.strategy",value = "THREAD")},threadPoolKey = "saveStudent",threadPoolProperties = {@HystrixProperty(name= CORE_SIZE,value = "3"),@HystrixProperty(name= MAX_QUEUE_SIZE,value = "5"),@HystrixProperty(name= QUEUE_SIZE_REJECTION_THRESHOLD,value = "7"),@HystrixProperty(name = CIRCUIT_BREAKER_ENABLED, value = "true"),// 是否开启断路器@HystrixProperty(name = CIRCUIT_BREAKER_REQUEST_VOLUME_THRESHOLD, value = "10"),// 请求次数@HystrixProperty(name = CIRCUIT_BREAKER_SLEEP_WINDOW_IN_MILLISECONDS, value = "10000"), // 时间窗口期@HystrixProperty(name = CIRCUIT_BREAKER_ERROR_THRESHOLD_PERCENTAGE, value = "60"),@HystrixProperty(name = EXECUTION_ISOLATION_THREAD_TIMEOUT_IN_MILLISECONDS, value = "4000"),//统计滚动窗口的持续时间，以毫秒为单位。// 该时间用于断路器判断健康度时需要收集信息的持续时间，断路器在收集指标信息的时候会根据设置的时间窗长度拆分成多个“桶”来累计各个度量值，每个“桶”记录一段时间内的采集指标。）@HystrixProperty(name = METRICS_ROLLING_STATS_TIME_IN_MILLISECONDS, value = "10000")},fallbackMethod = "saveStudent")

feign.hystrix.enabled=true
hystrix.threadpool.default.coreSize=100 #并发执行的最大线程数，默认10
hystrix.threadpool.default.maxQueueSize=5  #BlockingQueue的最大队列数，当设为－1，会使用 SynchronousQueue，值为正时使用LinkedBlcokingQueue。该设置只会在初始化时有效，之后不能修改
hystrix.threadpool.default.queueSizeRejectionThreshold=1 #即使maxQueueSize没有达到，达到queueSizeRejectionThreshold该值后，请求也会被拒绝。因为maxQueueSize不能被动态修改，这个参数将允许我们动态设置该值。if maxQueueSize == ?1，该字段将不起作用
hystrix.command.default.execution.isolation.strategy=SEMAPHORE #隔离策略，默认是Thread, 可选Thread｜Semaphor
hystrix.command.default.execution.isolation.thread.timeoutInMilliseconds=1000  #命令执行超时时间，默认1000ms
hystrix.command.default.execution.timeout.enabled=true #执行是否启用超时，默认启用true
hystrix.command.default.execution.isolation.strategy.semaphore.max-semaphores=100

service-B.ribbon.NFLoadBalancerRuleClassName=com.netflix.loadbalancer.RandomRule
ribbon.eureka.enabled=true
ribbon.ReadTimeout = 10000
ribbon.ConnectTimeout= 10000
ribbon.MaxAutoRetries= 0
ribbon.MaxAutoRetriesNextServer=1
ribbon.OkToRetryOnAllOperations=false

sentinel 限流熔断

sentinel 配置

        <dependency><groupId>org.springframework.boot</groupId><artifactId>spring-boot-starter-actuator</artifactId></dependency><dependency><groupId>com.alibaba.cloud</groupId><artifactId>spring-cloud-starter-alibaba-nacos-discovery</artifactId></dependency><dependency><groupId>com.alibaba.cloud</groupId><artifactId>spring-cloud-starter-alibaba-sentinel</artifactId></dependency><dependency><groupId>com.alibaba.csp</groupId><artifactId>sentinel-datasource-apollo</artifactId><version>1.8.0</version></dependency>

spring.cloud.nacos.discovery.server-addr=127.0.0.1:8848
management.endpoints.web.exposure.include=*
spring.cloud.sentinel.transport.port=8719
spring.cloud.sentinel.transport.dashboard=localhost:8480
feign.sentinel.enabled=true
#spring.cloud.sentinel.datasource.ds.file.file=classpath:flowrule.json
#spring.cloud.sentinel.datasource.ds.file.data-type=json
#spring.cloud.sentinel.datasource.ds2.file.rule-type=flow
spring.cloud.sentinel.datasource.ds2.apollo.namespace-name=application
spring.cloud.sentinel.datasource.ds2.apollo.flow-rules-key=flowRules
spring.cloud.sentinel.datasource.ds2.apollo.default-flow-rule-value=[]
spring.cloud.sentinel.datasource.ds2.apollo.rule-type=flow

@SentinelResource(value = "sayHello",blockHandler ="handleException",blockHandlerClass = ExceptionUtil.class)

Field   说明 默认值
resource   资源名，资源名是限流规则的作用对象
count  限流阈值
grade  限流阈值类型，QPS 或线程数模式  QPS 模式
limitApp   流控针对的调用来源  default，代表不区分调用来源
strategy   判断的根据是资源自身，还是根据其它关联资源 (refResource)，还是根据链路入口   根据资源本身
controlBehavior    流控效果（直接拒绝 / 排队等待 / 慢启动模式）  直接拒绝在Spring Cloud Sentinel中，降级（degrade）是指当资源被大量调用，且每个调用都执行了较长时间，为了保护系统不被打垮，Sentinel会对资源进行降级处理，即停止该资源的访问。降级规则可以通过以下参数进行配置：
Field  说明 默认值
grade：降级规则的级别，分为0（允许），1（警告），2（错误）。
count：触发降级的最小请求数。
timeWindow：时间窗口，单位为毫秒，请求次数在这个时间窗口内达到阈值则触发降级。
minRequestAmount：降级后每分钟允许的最小请求数，用于自动恢复服务。
slowRatioThreshold：慢请求比例阈值，当慢请求比例超过这个值就会进行降级。
statIntervalMs：统计时间窗口，用于计算慢请求比例，单位为毫秒。

flow 规则

[{"resource": "sayHello","controlBehavior": 0,"count": 1,"grade": 1,"limitApp": "default","strategy": 0}
]

degrade.json

[{"resource": "sayHello","count": 100.0,"timeWindow": 10,"grade": 1,"countType": 1
}]

zuul 源码解析

hystrix

gateway 源码解析

sentinel 源码解析

遇到问题解决方式

springcloud 2020.0.3重大改变

当前组件替代品

Hystrix Resilience4j

Hystrix Dashboard / Turbine Micrometer + Monitoring System

Ribbon Spring Cloud Loadbalancer

Zuul 1 Spring Cloud Gateway

Archaius 1 Spring Boot external config + Spring Cloud Config

以下模块已从spring-cloud-netflix中删除：

spring-cloud-netflix-archaius
spring-cloud-netflix-concurrency-limits
spring-cloud-netflix-core
spring-cloud-netflix-dependencies
spring-cloud-netflix-hystrix
spring-cloud-netflix-hystrix-contract
spring-cloud-netflix-hystrix-dashboard
spring-cloud-netflix-hystrix-stream
spring-cloud-netflix-ribbon
spring-cloud-netflix-sidecar
spring-cloud-netflix-turbine
spring-cloud-netflix-turbine-stream
spring-cloud-netflix-zuul
spring-cloud-starter-netflix-archaius
spring-cloud-starter-netflix-hystrix
spring-cloud-starter-netflix-hystrix-dashboard
spring-cloud-starter-netflix-ribbon
spring-cloud-starter-netflix-turbine
spring-cloud-starter-netflix-turbine-stream
spring-cloud-starter-netflix-zuul
Support for ribbon, hystrix and zuul was removed across the release train projects.