Spring Cloud 远程接口调用OpenFeign负载均衡实现原理详解

环境：Spring Cloud 2021.0.7 + Spring Boot 2.7.12

---

配置依赖

maven依赖

  org.springframework.cloud
  spring-cloud-starter-openfeign


  org.springframework.cloud
  spring-cloud-loadbalancer


  
    
      org.springframework.cloud
      spring-cloud-dependencies
      ${spring-cloud.version}
      pom
      import
    
  

开启注解功能

@SpringBootApplication
// 开启Feign功能，在该注解中你还可以配置，如下3个重要的信息：
// 1. 为所有的FeignClient提供统一默认的配置
// 2. 指定扫描那些包写的类
// 3. 指定有哪些@FeignClient类
@EnableFeignClients
public class AppApplication {

  public static void main(String[] args) {
    SpringApplication.run(AppApplication.class, args);
  }

}

FeignClient生成Bean原理

容器在启动过程中会找到所有@FeignClient的接口类，然后将这些类注册为容器Bean，而每一个Feign客户端对应的是FactoryBean对象FeignClientFactoryBean。

具体如何找这些带有@FeignClient注解的接口类可以查看FeignClientsRegistrar该类就在@EnableFeignClients中被导入。

FeignClientFactoryBean

public class FeignClientFactoryBean implements FactoryBean {
  public Object getObject() {
    return getTarget();
  }
   T getTarget() {
    FeignContext context = beanFactory != null ? beanFactory.getBean(FeignContext.class) : applicationContext.getBean(FeignContext.class);
    Feign.Builder builder = feign(context);
    if (!StringUtils.hasText(url)) {
      if (!name.startsWith("http")) {
        url = "http://" + name;
      }
      else {
        url = name;
      }
      url += cleanPath();
      // 负载均衡处理
      return (T) loadBalance(builder, context, new HardCodedTarget<>(type, name, url));
    }
    // ...
  }
  protected  T loadBalance(Feign.Builder builder, FeignContext context, HardCodedTarget target) {
    // 在OpenFeign中核心实现负载均衡的类就是具体的Client类
    // Feign负载均衡能力实现通过具体Client实现，每一个FeignClient客户端都会对应一个子容器AnnotationConfigApplicationContext
    // 根据@FeignClient配置的服务名name或value为key，从一个LoadBalancerClientFactory（父类）中的Map中查找该name对应的容器
    // 如果不存在则创建一个AnnotationConfigApplicationContext。每个子容器都设置了父容器，如果通过子容器查找不到Client的实现，那么会从父容器中查找
    Client client = getOptional(context, Client.class);
  }
}

Client实现

Client的具体实现可以有如下：

1. apache httpclient

2. okhttp

3. default（jdk）

具体使用哪个是根据你环境引入了哪个依赖（httpclient，okhttp）

  io.github.openfeign
  feign-httpclient
  ${version}



  io.github.openfeign
  feign-okhttp
  ${version}

具体选择通过如下配置

@Import({ 
  HttpClientFeignLoadBalancerConfiguration.class, 
  OkHttpFeignLoadBalancerConfiguration.class, 
  HttpClient5FeignLoadBalancerConfiguration.class, 
  DefaultFeignLoadBalancerConfiguration.class })
public class FeignLoadBalancerAutoConfiguration {
}

如果你的环境有多个实现，那么这里会根据这里的导入顺序加载。这里以最后一个
DefaultFeignLoadBalancerConfiguration为例。

class DefaultFeignLoadBalancerConfiguration {

  @Bean
  @ConditionalOnMissingBean
  // 没有启用spring-retry重试功能
  @Conditional(OnRetryNotEnabledCondition.class)
  public Client feignClient(LoadBalancerClient loadBalancerClient, LoadBalancerClientFactory loadBalancerClientFactory) {
    // 这里构造函数第一个参数将会成为最终执行远程接口调用的实现
    return new FeignBlockingLoadBalancerClient(new Client.Default(null, null), loadBalancerClient, loadBalancerClientFactory);
  }
}

在没有导入httpclient或者okhttp情况下，使用的Client实现是
FeignBlockingLoadBalancerClient。

负载均衡实现

构造
FeignBlockingLoadBalancerClient传入了负载均衡客户端LoadBalancerClient及负载均衡客户端工厂LoadBalancerClientFactory该工厂是用来创建每一个Feign客户端对应的子容器
AnnotationConfigApplicationContext及从对应子容器获取相应的Bean实例对象，如：Client，Request.Options，Logger.Level等。

public class FeignBlockingLoadBalancerClient implements Client {
  // 此Client代理对象是上面的new Client.Default(null, null)
  private final Client delegate;
  private final LoadBalancerClient loadBalancerClient;
  private final LoadBalancerClientFactory loadBalancerClientFactory;
  public FeignBlockingLoadBalancerClient(Client delegate, LoadBalancerClient loadBalancerClient, LoadBalancerClientFactory loadBalancerClientFactory) {
    this.delegate = delegate;
    this.loadBalancerClient = loadBalancerClient;
    this.loadBalancerClientFactory = loadBalancerClientFactory;
  }
  @Override
  public Response execute(Request request, Request.Options options) throws IOException {
    final URI originalUri = URI.create(request.url());
    // 获取服务名serviceId
    String serviceId = originalUri.getHost();
    String hint = getHint(serviceId);
    DefaultRequest lbRequest = new DefaultRequest<>(new RequestDataContext(buildRequestData(request), hint));
    // ...
    // 通过负载均衡客户端获取指定serviceId的服务实例
    ServiceInstance instance = loadBalancerClient.choose(serviceId, lbRequest);
    // ...
    // 通过获取到的ServiceInstance实例，重新构造请求地址
    String reconstructedUrl = loadBalancerClient.reconstructURI(instance, originalUri).toString();
    // 重新构建一个新的请求
    Request newRequest = buildRequest(request, reconstructedUrl);
    LoadBalancerProperties loadBalancerProperties = loadBalancerClientFactory.getProperties(serviceId);
    return executeWithLoadBalancerLifecycleProcessing(delegate, options, newRequest, lbRequest, lbResponse, supportedLifecycleProcessors, loadBalancerProperties.isUseRawStatusCodeInResponseData());
  }

  protected Request buildRequest(Request request, String reconstructedUrl) {
    return Request.create(request.httpMethod(), reconstructedUrl, request.headers(), request.body(),
      request.charset(), request.requestTemplate());
  }
}

LoadBalancerClient具体实现：​​​​​​​BlockingLoadBalancerClient

public class BlockingLoadBalancerClient implements LoadBalancerClient {

  private final ReactiveLoadBalancer.Factory loadBalancerClientFactory;

  public BlockingLoadBalancerClient(ReactiveLoadBalancer.Factory loadBalancerClientFactory) {
    this.loadBalancerClientFactory = loadBalancerClientFactory;
  }

  public  ServiceInstance choose(String serviceId, Request request) {
    // 获取一个负载均衡器，默认是轮询策略
    ReactiveLoadBalancer loadBalancer = loadBalancerClientFactory.getInstance(serviceId);
    if (loadBalancer == null) {
      return null;
    }
    Response loadBalancerResponse = Mono.from(loadBalancer.choose(request)).block();
    if (loadBalancerResponse == null) {
      return null;
    }
    return loadBalancerResponse.getServer();
  }
  // 重新构造请求的uri
  public URI reconstructURI(ServiceInstance serviceInstance, URI original) {
    return LoadBalancerUriTools.reconstructURI(serviceInstance, original);
  }
}
public final class LoadBalancerUriTools {
  public static URI reconstructURI(ServiceInstance serviceInstance, URI original) {
    // ...
    return doReconstructURI(serviceInstance, original);
  }
  private static URI doReconstructURI(ServiceInstance serviceInstance, URI original) {
    String host = serviceInstance.getHost();
    String scheme = Optional.ofNullable(serviceInstance.getScheme()).orElse(computeScheme(original, serviceInstance));
    int port = computePort(serviceInstance.getPort(), scheme);

    if (Objects.equals(host, original.getHost()) && port == original.getPort() && Objects.equals(scheme, original.getScheme())) {
      return original;
    }
    boolean encoded = containsEncodedParts(original);
    return UriComponentsBuilder.fromUri(original).scheme(scheme).host(host).port(port).build(encoded).toUri();
  }
}

轮询算法​​​​​​​

public class RoundRobinLoadBalancer implements ReactorServiceInstanceLoadBalancer {
  public Mono> choose(Request request) {
    ServiceInstanceListSupplier supplier = serviceInstanceListSupplierProvider
      .getIfAvailable(NoopServiceInstanceListSupplier::new);
    return supplier.get(request).next()
      .map(serviceInstances -> processInstanceResponse(supplier, serviceInstances));
  }

  private Response processInstanceResponse(ServiceInstanceListSupplier supplier, List serviceInstances) {
    Response serviceInstanceResponse = getInstanceResponse(serviceInstances);
    if (supplier instanceof SelectedInstanceCallback && serviceInstanceResponse.hasServer()) {
      ((SelectedInstanceCallback) supplier).selectedServiceInstance(serviceInstanceResponse.getServer());
    }
    return serviceInstanceResponse;
  }

  private Response getInstanceResponse(List instances) {
    // ...
    // 如果只有一个实例
    if (instances.size() == 1) {
      return new DefaultResponse(instances.get(0));
    }

    // Ignore the sign bit, this allows pos to loop sequentially from 0 to
    // Integer.MAX_VALUE
    int pos = this.position.incrementAndGet() & Integer.MAX_VALUE;

    ServiceInstance instance = instances.get(pos % instances.size());

    return new DefaultResponse(instance);
  }
}

执行远程调用

接着上面
FeignBlockingLoadBalancerClient#execute方法最终的返回方法执行​​​​​​​

final class LoadBalancerUtils {
  static Response executeWithLoadBalancerLifecycleProcessing(Client feignClient, Request.Options options,
    Request feignRequest, org.springframework.cloud.client.loadbalancer.Request lbRequest,
    org.springframework.cloud.client.loadbalancer.Response lbResponse,
    Set supportedLifecycleProcessors, boolean useRawStatusCodes) throws IOException {
    return executeWithLoadBalancerLifecycleProcessing(feignClient, options, feignRequest, lbRequest, lbResponse, supportedLifecycleProcessors, true, useRawStatusCodes);
  }
  static Response executeWithLoadBalancerLifecycleProcessing(Client feignClient, Request.Options options,
    Request feignRequest, org.springframework.cloud.client.loadbalancer.Request lbRequest,
    org.springframework.cloud.client.loadbalancer.Response lbResponse,
    Set supportedLifecycleProcessors, boolean loadBalanced, boolean useRawStatusCodes) throws IOException {
    // 这里执行生命周期实际调用前动作
    try {
      // 执行时间的调用，而这里的feignClient就是在FeignBlockingLoadBalancerClient传递过来的，new Client.Default(null, null)
      Response response = feignClient.execute(feignRequest, options);
      // 这里执行生命周期回调，省略
      return response;
    }
    // ...
  }
}

Client.Default

public interface Client {
  public Response execute(Request request, Options options) throws IOException {
    // 通过JDK自带的网络连接进行处理
    HttpURLConnection connection = convertAndSend(request, options);
    return convertResponse(connection, request);
  }
}

以上就是一个Feign客户端请求的负载均衡实现原理

完毕~！~！

求助三连~~~