深入理解@LoadBalanced注解的实现原理与客户端负载均衡

前提

在阅读这篇博客之前,希望你对SpringCloud套件熟悉和理解,更希望关注下微服务开发平台

概述

在使用springcloud ribbon客户端负载均衡的时候，可以给RestTemplate bean 加一个@LoadBalanced注解，就能让这个RestTemplate在请求时拥有客户端负载均衡的能力,先前有细嚼过但是没有做过笔记,刚好处理此类问题记录下

@LoadBalanced

/**
 * 注释将RestTemplate bean标记为配置为使用LoadBalancerClient。
 */
@Target({ ElementType.FIELD, ElementType.PARAMETER, ElementType.METHOD })
@Retention(RetentionPolicy.RUNTIME)
@Documented
@Inherited
@Qualifier
public @interface LoadBalanced {
}

通过源码可以发现这是一个LoadBalanced标记注解并且标记了@Qualifier(基于Spring Boot的自动配置机制),我们可以溯源到LoadBalancerAutoConfiguration

LoadBalancerAutoConfiguration

/**
 * 功能区的自动配置（客户端负载平衡）
 */
@Configuration
@ConditionalOnClass(RestTemplate.class)
@ConditionalOnBean(LoadBalancerClient.class)
@EnableConfigurationProperties(LoadBalancerRetryProperties.class)
public class LoadBalancerAutoConfiguration {

	@LoadBalanced
	@Autowired(required = false)
	private List<RestTemplate> restTemplates = Collections.emptyList();   //这里持有@LoadBalanced标记的RestTemplate实例

	@Autowired(required = false)
	private List<LoadBalancerRequestTransformer> transformers = Collections.emptyList();

	@Bean
	public SmartInitializingSingleton loadBalancedRestTemplateInitializerDeprecated(
			final ObjectProvider<List<RestTemplateCustomizer>> restTemplateCustomizers) {
		return () -> restTemplateCustomizers.ifAvailable(customizers -> {
			for (RestTemplate restTemplate : LoadBalancerAutoConfiguration.this.restTemplates) {
				for (RestTemplateCustomizer customizer : customizers) {
          //为restTemplate添加定制
					customizer.customize(restTemplate);
				}
			}
		});
	}

   // ... 

	/**
	 * 以下针对classpath存在RetryTemplate.class的情况配置，先忽略
	 */
	@Configuration
	@ConditionalOnClass(RetryTemplate.class)
	public static class RetryAutoConfiguration {

		@Bean
		@ConditionalOnMissingBean
		public LoadBalancedRetryFactory loadBalancedRetryFactory() {
			return new LoadBalancedRetryFactory() {
			};
		}
	}

  // ...
}

@LoadBalanced和@Autowried结合使用,意思就是这里注入的RestTempate Bean是所有加有@LoadBalanced注解标记的(持有@LoadBalanced标记的RestTemplate实例)

这段自动装配的代码的含义不难理解，就是利用了RestTempllate的拦截器，使用RestTemplateCustomizer对所有标注了@LoadBalanced的RestTemplate Bean添加了一个LoadBalancerInterceptor拦截器，而这个拦截器的作用就是对请求的URI进行转换获取到具体应该请求哪个服务实例ServiceInstance。

关键问下自己：为什么？

• RestTemplate实例是怎么被收集的？
• 怎样通过负载均衡规则获取具体的具体的server？

继续扒看源码>

上面可以看出，会LoadBalancerAutoConfiguration类对我们加上@LoadBalanced注解的bean 添加loadBalancerInterceptor拦截器

LoadBalancerInterceptor

/**
* 功能区的自动配置（客户端负载平衡）。
*/
public class LoadBalancerInterceptor implements ClientHttpRequestInterceptor {

	private LoadBalancerClient loadBalancer;

	private LoadBalancerRequestFactory requestFactory;

	public LoadBalancerInterceptor(LoadBalancerClient loadBalancer,
			LoadBalancerRequestFactory requestFactory) {
		this.loadBalancer = loadBalancer;
		this.requestFactory = requestFactory;
	}

	public LoadBalancerInterceptor(LoadBalancerClient loadBalancer) {
		// for backwards compatibility
		this(loadBalancer, new LoadBalancerRequestFactory(loadBalancer));
	}

	@Override
	public ClientHttpResponse intercept(final HttpRequest request, final byte[] body,
			final ClientHttpRequestExecution execution) throws IOException {
		final URI originalUri = request.getURI();
		String serviceName = originalUri.getHost();
		Assert.state(serviceName != null,
				"Request URI does not contain a valid hostname: " + originalUri);
		return this.loadBalancer.execute(serviceName,
				this.requestFactory.createRequest(request, body, execution));
	}

}

重点看intercept方法 当我们restTemplate执行请求操作时，就会被拦截器拦截进入intercept方法,而loadBalancer是LoadBalancerClient的具体实现

RibbonLoadBalancerClient

	public <T> T execute(String serviceId, LoadBalancerRequest<T> request, Object hint)
			throws IOException {
		ILoadBalancer loadBalancer = getLoadBalancer(serviceId);
		Server server = getServer(loadBalancer, hint);
		if (server == null) {
			throw new IllegalStateException("No instances available for " + serviceId);
		}
		RibbonServer ribbonServer = new RibbonServer(serviceId, server,
				isSecure(server, serviceId),
				serverIntrospector(serviceId).getMetadata(server));

		return execute(serviceId, ribbonServer, request);
	}

看到这里相信都遇到过类似的错误，恍然大悟

No instances available for  xxxxx

总结

• 1.根据serviceId 获取对应的loadBalancer
• 2.根据loadBalancer获取具体的server（这里根据负载均衡规则，获取到具体的服务实例）
• 3.创建RibbonServer
• 4.执行具体请求

这里

注意: @LoadBalanced 标记注解获取到最后通过负载均衡规则获取具体的具体的server来发起请求

案例

/**
 * 服务注册中心配置
 *
 * @author <a href="mailto:shangzhi.ibyte@gmail.com">iByte</a>
 * @since 1.0.1
 */
@Configuration
@EnableConfigurationProperties(ModuleMappingHelper.class)
public class DiscoveryConfig {
    @Autowired
    Environment environment;

    /**
     * DiscoveryHeaderHelper默认bean
     * @return
     */
    @Bean
    public DiscoveryHeaderHelper discoveryHeaderHelper() {
        DiscoveryHeaderHelper discoveryHeaderHelper = new DiscoveryHeaderHelper(environment);
        DiscoveryHeaderHelper.INSTANCE = discoveryHeaderHelper;
        return discoveryHeaderHelper;
    }

    /**
     * resttemplate构建
     */
    @Resource
    private RestTemplateBuilder restTemplateBuilder;

    /**
     * resttemplate请求bean,更改系统本身的builder
     * @return
     */
    @Bean
    @LoadBalanced
    public RestTemplate restTemplate() {
        RestTemplate restTemplate = restTemplateBuilder.configure(new RestTemplate());
        //RestTemplate interceptors 远程调用请求增加头部信息处理
        restTemplate.getInterceptors().add(new RestApiHeaderInterceptor());
        //RestTemplate Set the error handler 错误处理
        restTemplate.setErrorHandler(new RestResponseErrorHandler());
        return  restTemplate;
    }

    @Bean
    public DiscoveryClient.DiscoveryClientOptionalArgs discoveryClientOptionalArgs() {
        DiscoveryClient.DiscoveryClientOptionalArgs discoveryClientOptionalArgs = new DiscoveryClient.DiscoveryClientOptionalArgs();
        discoveryClientOptionalArgs.setAdditionalFilters(Collections.singletonList(new DiscoveryHeaderClientFilter()));
        discoveryClientOptionalArgs.setEventListeners(Collections.singleton(new EurekaClientEventListener()));
        return discoveryClientOptionalArgs;
    }
}

源码地址 > DiscoveryConfig