一、问题描述

我们现在是在用Nacos替换Eureka,原来Eureka和Spring gateway运行正常,可以通过Spring gateway调用注册到Eureka中的服务。

当前Spring cloud的版本是Hoxton.SR8,Nacos discovery的版本为0.9.0.RELEASE,使用的Nacos版本为2.0.3。

Nacos替换Eureka改动的地方如下:

1、去掉POM中Eureka的引入;

2、去掉主类中引入的@EnableEurekaClient注解;

3、引入Nacos依赖:

<dependency>
  <groupId>org.springframework.cloud</groupId>
  <artifactId>spring-cloud-starter-alibaba-nacos-discovery</artifactId>
  <version>0.9.0.RELEASE</version>
  <exclusions>
    <exclusion>
        <groupId>com.alibaba.nacos</groupId>
        <artifactId>nacos-client</artifactId>
    </exclusion>
    <exclusion>
     <groupId>org.springframework.cloud</groupId>
     <artifactId>spring-cloud-starter-netflix-ribbon</artifactId>
    </exclusion>
  </exclusions>
</dependency>
<dependency>
  <groupId>com.alibaba.nacos</groupId>
  <artifactId>nacos-client</artifactId>
  <version>2.0.3</version>
</dependency>

4、引入spring-cloud-loadbalancer:

<dependency>
	<groupId>org.springframework.cloud</groupId>
	<artifactId>spring-cloud-loadbalancer</artifactId>
</dependency>

5、spring-cloud-starter-openfeign中exclude掉ribbon的依赖:

<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-cloud-starter-openfeign</artifactId>
    <exclusions>
        <exclusion>
            <groupId>org.springframework.cloud</groupId>
            <artifactId>spring-cloud-netflix-ribbon</artifactId>
        </exclusion>
    </exclusions>
</dependency>

6、application.yml中禁用nacos的ribbon:

ribbon:
  nacos:
    enabled: false

现在的现象:

  • Spring gateway启动不报错;
  • gateway也可以正常注册到Nacos;
  • 通过打断点到NacosDiscoveryClientAutoConfiguration,NacosDiscoveryClient会被初使化,但是调用时不会调用NacosDiscoveryClient.getInstances(String)获取被调用应用的Provider,会报类似下面的接口404错误:

{"msg":"/user-service/user/getByToken","result":404,"data":null}

其中user-service为注册到Nacos上面的服务名称,但是就是不能够被调用到。

  • Spring gateway的注册中心切换为Eureka,再次调用服务时,Eureka的CompositeDiscoveryClient.getInstances(String)方法就会被调用。

二、问题分析

前面走了一些弯路,也花了一些时间,最后想到的是与Eureka注册中心进行对比调试分析,下面例出关键的分析步骤。

Debug断点打到DiscoveryClientRouteDefinitionLocator两个构造函数上,只有参数是ReactiveDiscoveryClient的构造函数被加载:

继续跟踪到ReactiveCompositeDiscoveryClient.getServices()方法:

发现数组变量discoveryClients包含了两个实现,如下图所示:

其中有一个是EurekaReactiveDiscoveryClient,那在使用Nacos时就应该包含一个针对Nacos实现的ReactiveDiscoveryClient,于是切换为Nacos分支调试。

但是调试发现数组变量discoveryClients并没有包含针对Nacos实现的ReactiveDiscoveryClient:

于是参考EurekaReactiveDiscoveryClient以及NacosDiscoveryClient,立即就写了NacosReactiveDiscoveryClient:

import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

import org.springframework.cloud.alibaba.nacos.NacosDiscoveryProperties;
import org.springframework.cloud.alibaba.nacos.NacosServiceInstance;
import org.springframework.cloud.client.ServiceInstance;
import org.springframework.cloud.client.discovery.ReactiveDiscoveryClient;

import com.alibaba.nacos.api.naming.pojo.Instance;
import com.alibaba.nacos.api.naming.pojo.ListView;

import lombok.extern.slf4j.Slf4j;
import reactor.core.publisher.Flux;
/**
 * 
 * @author fenglibin
 *
 */
@Slf4j
public class NacosReactiveDiscoveryClient implements ReactiveDiscoveryClient {

	public static final String DESCRIPTION = "Spring Cloud Nacos Reactive Discovery Client";

	private NacosDiscoveryProperties discoveryProperties;

	public NacosReactiveDiscoveryClient(NacosDiscoveryProperties discoveryProperties) {
		this.discoveryProperties = discoveryProperties;
	}

	@Override
	public String description() {
		return DESCRIPTION;
	}

	@Override
	public Flux<ServiceInstance> getInstances(String serviceId) {
		try {
			List<Instance> instances = discoveryProperties.namingServiceInstance().selectInstances(serviceId, true);

			return Flux.fromStream(hostToServiceInstanceList(instances, serviceId).stream());
		} catch (Exception e) {
			throw new RuntimeException("Can not get hosts from nacos server. serviceId: " + serviceId, e);
		}
	}

	@Override
	public Flux<String> getServices() {
		try {
			ListView<String> services = discoveryProperties.namingServiceInstance().getServicesOfServer(1,
					Integer.MAX_VALUE);

			return Flux.fromStream(services.getData().parallelStream());
		} catch (Exception e) {
			log.error("get service name from nacos server fail,", e);
			return Flux.fromStream(Collections.<String>emptyList().stream());
		}
	}

	private static ServiceInstance hostToServiceInstance(Instance instance, String serviceId) {
		NacosServiceInstance nacosServiceInstance = new NacosServiceInstance();
		nacosServiceInstance.setHost(instance.getIp());
		nacosServiceInstance.setPort(instance.getPort());
		nacosServiceInstance.setServiceId(serviceId);

		Map<String, String> metadata = new HashMap<>();
		metadata.put("nacos.instanceId", instance.getInstanceId());
		metadata.put("nacos.weight", instance.getWeight() + "");
		metadata.put("nacos.healthy", instance.isHealthy() + "");
		metadata.put("nacos.cluster", instance.getClusterName() + "");
		metadata.putAll(instance.getMetadata());
		nacosServiceInstance.setMetadata(metadata);

		if (metadata.containsKey("secure")) {
			boolean secure = Boolean.parseBoolean(metadata.get("secure"));
			nacosServiceInstance.setSecure(secure);
		}
		return nacosServiceInstance;
	}

	private static List<ServiceInstance> hostToServiceInstanceList(List<Instance> instances, String serviceId) {
		List<ServiceInstance> result = new ArrayList<>(instances.size());
		for (Instance instance : instances) {
			result.add(hostToServiceInstance(instance, serviceId));
		}
		return result;
	}

}

或者集成如下starter:

    <dependency>
        <groupId>com.eeeffff</groupId>
        <artifactId>nacos-reactive-discovery-client</artifactId>
        <version>1.0.0</version>
    </dependency>

并在启动时创建Bean:

再次启动调试就可以看到数组变量discoveryClients中有NacosReactiveDiscoveryClient了:

 再次通过网关请求服务,就可以正常调用成功了。

 感觉Nacos的怪问题都让我给碰上了!

Logo

为开发者提供学习成长、分享交流、生态实践、资源工具等服务,帮助开发者快速成长。

更多推荐