resilience4j 是一款 java 平台轻量级容错库，支持熔断、限流、重试等功能。由于Netflix Hystrix 闭源，我们急需一款功能强大的容错工具库，来保护我们的环境。resilience4j 提供了spring boot 的starter，所以集成resilience4j很简单，但是也有一些坑。因此记录一下。

github：https://github.com/resilience4j/resilience4j

官方文档：https://resilience4j.readme.io/docs/circuitbreaker

我使用的是resilience4j 最新版本 0.16.0，spring boot 2.1.6。

相关依赖

<dependencies>

    <dependency>

        <groupId>org.springframework.boot</groupId>

        <artifactId>spring-boot-starter-actuator</artifactId>

    </dependency>

    <dependency>

        <groupId>org.springframework.boot</groupId>

        <artifactId>spring-boot-starter-web</artifactId>

    </dependency>

  <dependency>

        <groupId>org.springframework.boot</groupId>

        <artifactId>spring-boot-starter-aop</artifactId>

    </dependency>

    <dependency>

        <groupId>io.github.resilience4j</groupId>

        <artifactId>resilience4j-spring-boot2</artifactId>

        <version>0.16.0</version>

    </dependency>

    <dependency>

        <groupId>org.springframework.boot</groupId>

        <artifactId>spring-boot-starter-test</artifactId>

        <scope>test</scope>

    </dependency>

</dependencies>

配置resilience4j，在spring 中的application.yml 中增加如下配置：

resilience4j.circuitbreaker:

instances:

backendA:

  registerHealthIndicator: true

  ringBufferSizeInClosedState: 10

  ringBufferSizeInHalfOpenState: 3

  waitDurationInOpenState: 10s

  failureRateThreshold: 50

  eventConsumerBufferSize: 2

backendB:

  registerHealthIndicator: true

  ringBufferSizeInClosedState: 10

  ringBufferSizeInHalfOpenState: 3

  waitDurationInOpenState: 10s

  failureRateThreshold: 50

  eventConsumerBufferSize: 2

recordFailurePredicate: io.github.robwin.exception.RecordFailurePredicate

resilience4j.retry:

instances:

backendA:

  maxRetryAttempts: 3

  waitDuration: 10s

  enableExponentialBackoff: true

  exponentialBackoffMultiplier: 2

  retryExceptions:

    - org.springframework.web.client.HttpServerErrorException

  ignoreExceptions:

    - com.learn.demo.resilience.ResilienceException

    - java.io.IOException

backendB:

  maxRetryAttempts: 3

  waitDuration: 10s

  retryExceptions:

    - com.learn.demo.resilience.ResilienceException

  ignoreExceptions:

    - java.io.IOException

resilience4j.bulkhead:

instances:

backendA:

  maxConcurrentCall: 10

  maxWaitDuration: 20ms

backendB:

  maxWaitDuration: 10ms

  maxConcurrentCall: 20

resilience4j.thread-pool-bulkhead:

instances:

backendC:

  threadPoolProperties:

    maxThreadPoolSize: 1

    coreThreadPoolSize: 1

    queueCapacity: 1

resilience4j.ratelimiter:

instances:

backendA:

  limitForPeriod: 10

  limitRefreshPeriod: 1s

  timeoutDuration: 0

  registerHealthIndicator: true

  eventConsumerBufferSize: 100

backendB:

  limitForPeriod: 6

  limitRefreshPeriod: 500ms

  timeoutDuration: 3s

server:

port: 9090

注意

resilience4j 的配置在0.16.0 发生了一下变化。网上一些文档都是基于0.13.0版本。还有0.16.0版本配置不能被IDEA 识别，也不能自动联想，所以使用0.16.0版本尽量参考官网的配置。

熔断

配置介绍

注解

@CircuitBreaker(name = "backendA")

code

@Retry(name = "backendA")

@CircuitBreaker(name = "backendA")

@RateLimiter(name = "backendA")

@Service

public class ServiceA {

@Bulkhead(name = "backendA")

public String sucess(){

    System.out.println("sucess a");

    return "sucess a";

}

@Bulkhead(name = "backendA")

public String fail() throws ResilienceException {

    throw new ResilienceException("fail a");

}

}

多次失败达到阈值之后，再调用会抛出异常

io.github.resilience4j.circuitbreaker.CallNotPermittedException: CircuitBreaker 'backendA' is OPEN and does not permit further calls

rest 调用返回结果：

{

"timestamp": "2019-07-10T09:45:21.065+0000",

"status": 500,

"error": "Internal Server Error",

"message": "CircuitBreaker 'backendA' is OPEN and does not permit further calls",

"path": "/serviceA/failed"

}

这种情况可以配合spring aop，拦截异常，返回特性http 状态码或者信息，让前端捕获特定http 状态码，以此来呈现熔断之后的ui。

限流

配置

注解

@RateLimiter(name = "backendA")

code

同上

达到限制时抛出异常

io.github.resilience4j.ratelimiter.RequestNotPermitted: RateLimiter 'backendA' does not permit further calls

注意

经过试验，当@RateLimiter和@CircuitBreaker一起使用时,需要熔断器增加忽略RequestNotPermitted异常。否则会触发熔断。

ignoreExceptions:

• io.github.resilience4j.ratelimiter.RequestNotPermitted

隔离（bulkhead）

其主要作用是限制并发次数

配置

注解

@Bulkhead(name = "backendA")

重试

配置

注解

@Retry(name = "backendA")

resilience4j 为我们提供了很棒的容错库，简单易用。