openfeign集成spring cloud loadbalancer实现负载均衡流程

当我们在项目中调用自己实现的Feignclient负载均衡中如何起作用？请看下图：

图一

在图一中我们可以发现，在ConsumerController中调用自定义的DemoFeignClient方法时，通过spring容器中对DemoFeignclient的代理类的调用最终通过feign.SynchronousMethodHandler.invoke()->openfeign.loadbalancer.execute()->org.springframework.cloud.loadbalancer.blocking.client.FeignBlockingLoadBalancerClient.choose()->org.springframework.cloud.loadbalancer.core.RoundRobinLoadBalancer.choose()这样的调用链，最终执行spring cloud loadbalaner中的轮询负载均横策略！

然而，上图中的这些关键类是如何组合起来发挥作用呢？请我们一起继续分析！spring cloud项目启动后，spring容器解析并加载LoadBalancerClientConfiguration.java配置文件（如下图所示）

图二

然后将"reactorServiceInstanceLoadBalancer"注册到beanDefinitionMap中。
然后会扫描我们声明的controller,因为conroller中注入了DemoFeignClient，因此spring容器会递归创建DemoFeignClient，创建DemoFiegnClient过程中会通过AbstractAutowireCapableBeanFactory.obtainFromSupplier()注入instanceSupplier实例,也就是FeignClientFactoryBean实例！

下面是org.springframework.cloud.openfeign.FeignClientsRegistrar.registerFeignClient()方法主要逻辑：

private void registerFeignClient(BeanDefinitionRegistry registry, AnnotationMetadata annotationMetadata,
            Map attributes) {
          //忽略非核心代码片段
            BeanDefinitionBuilder definition = BeanDefinitionBuilder.genericBeanDefinition(clazz, () -> {
            //设置url
            factoryBean.setUrl(getUrl(beanFactory, attributes));
            //设置路径
            factoryBean.setPath(getPath(beanFactory, attributes));
            //decode404 布尔值            factoryBean.setDecode404(Boolean.parseBoolean(String.valueOf(attributes.get("decode404"))));
            Object fallback = attributes.get("fallback");
            //设置fallback
            if (fallback != null) {
                factoryBean.setFallback(fallback instanceof Class ? (Class) fallback
                        : ClassUtils.resolveClassName(fallback.toString(), null));
            }
            //设置fallback工厂类
            Object fallbackFactory = attributes.get("fallbackFactory");
            if (fallbackFactory != null) {
                factoryBean.setFallbackFactory(fallbackFactory instanceof Class ? (Class) fallbackFactory
                        : ClassUtils.resolveClassName(fallbackFactory.toString(), null));
            }
            //实例化我们声明的DemoClient
            return factoryBean.getObject();
        });

}

这还没有完，org.springframework.cloud.openfeign.FeignClientFactoryBean.getObject()中最终执行loadBalance()；在loadBalance()中实例化FeignInvocationHandler;

在项目启动成功之后，我们调用DemoFeignClient中的方法时

图三

通过图三可以发现，描述DemoFeignClient的RootBeanDefinition的类中还有一个叫“instanceSupplier”的类型属性，它的值是“FeignClientsRegistrar$lambda”，那么这个类时什么时候被注入进来的呢？答案就是上面分析项目启动过程中spring容器根据RootBeanDefinition对DemoFeignClient的描述，通过对"FeignClientsRegistrar$lambda"的调用完成对DemoFeignClient实例的创建。

在完成创建DemoFeignClient实例后，我们调用org.springframework.cloud.openfeign.loadbalancer。FeignBlockingLoadBalancerClient.execute()执行负载均衡策略时，执行到如下代码块

Set supportedLifecycleProcessors = LoadBalancerLifecycleValidator
                .getSupportedLifecycleProcessors(
                        loadBalancerClientFactory.getInstances(serviceId, LoadBalancerLifecycle.class),
                        RequestDataContext.class, ResponseData.class, ServiceInstance.class);

"loadBalancerClientFactory.getInstances(serviceId, LoadBalancerLifecycle.class)"逻辑会获取LoadBalancer实例，最终调用LoadBalancerClientConfiguration.java中的reactorServiceInstanceLoadBalancer()实现创建负载均衡实例逻辑

@Bean
    @ConditionalOnMissingBean
    public ReactorLoadBalancer reactorServiceInstanceLoadBalancer(Environment environment,
            LoadBalancerClientFactory loadBalancerClientFactory) {
        String name = environment.getProperty(LoadBalancerClientFactory.PROPERTY_NAME);
        //创建负载均衡实例，而且我们发现默认的负载均衡实例是轮询负载均衡实例
        return new RoundRobinLoadBalancer(
                loadBalancerClientFactory.getLazyProvider(name, ServiceInstanceListSupplier.class), name);
    }

ok，截止到目前为止，我们的负载均衡实例已经创建完成！

这样一来，我们通过openfeign进行远程调用时，通过下图的调用链

图四

FeignInvocationHandler.invoke()->SynchronousMethodHandler.invoke()->org.springframework.cloud.openfeign.loadbalancer.FeignBlockingLoadBalancerClient.execute()->org.springframework.cloud.loadbalancer.blocking.client.BlockingLoadBalancerClient.choose()->org.springframework.cloud.loadbalancer.core.RoundRobinLoadBalancer.choose(),这样的调用链最终根据RoundRobinLoadBalancer.choose()获取服务实例。

public Mono> choose(Request request) {
        ServiceInstanceListSupplier supplier = serviceInstanceListSupplierProvider
                .getIfAvailable(NoopServiceInstanceListSupplier::new);
        //返回服务实例
        return supplier.get(request).next()
                .map(serviceInstances -> processInstanceResponse(supplier, serviceInstances));
    }

拿到服务实例后，FeignBlockingLoadBalancerClient.execute()中最终通过LoadBalancerUtils.executeWithLoadBalancerLifecycleProcessing()获取响应结果！
下面是具体代码逻辑：

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 {
        supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onStartRequest(lbRequest, lbResponse));
        try {
            //通过openfeign调用，获取响应结果
            Response response = feignClient.execute(feignRequest, options);
            if (loadBalanced) {
                supportedLifecycleProcessors.forEach(
                        lifecycle -> lifecycle.onComplete(new CompletionContext<>(CompletionContext.Status.SUCCESS,
                                lbRequest, lbResponse, buildResponseData(response, useRawStatusCodes))));
            }
            return response;
        }
        catch (Exception exception) {
            if (loadBalanced) {
                supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onComplete(
                        new CompletionContext<>(CompletionContext.Status.FAILED, exception, lbRequest, lbResponse)));
            }
            throw exception;
        }
    }