Spring Cloud Edgware.SR1下如何将Sleuth Trace ID传递至EurekaHttpClient？

我之前在维护同版本的微服务架构时，也碰到过一模一样的问题——Edgware.SR1版本的Spring Cloud Sleuth默认不会对Eureka的HTTP请求做追踪包装，导致服务注册、心跳、拉取注册表这些Eureka相关调用，完全不会出现在Zipkin的链路视图里。下面是我实测可行的解决思路和代码实现：

1. 自定义EurekaHttpClient拦截器注入Trace信息 #

我们需要写一个拦截器，把当前线程的Sleuth Trace ID、Span ID等信息，放到Eureka请求的Header里，让Eureka Server能接收到这些追踪标识，进而被Zipkin捕获链路。

import brave.Span;
import brave.Tracer;
import com.netflix.discovery.EurekaHttpClient;
import com.netflix.discovery.EurekaHttpRequest;
import com.netflix.discovery.EurekaHttpResponse;
import com.netflix.discovery.interceptor.EurekaHttpClientInterceptor;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.stereotype.Component;

@Component
public class TraceEurekaHttpClientInterceptor implements EurekaHttpClientInterceptor {

    private final Tracer tracer;

    @Autowired
    public TraceEurekaHttpClientInterceptor(Tracer tracer) {
        this.tracer = tracer;
    }

    @Override
    public <R> EurekaHttpResponse<R> intercept(EurekaHttpRequest request, EurekaHttpClient delegate) {
        Span currentSpan = tracer.currentSpan();
        if (currentSpan != null) {
            // 把Sleuth标准的Trace头添加到Eureka请求里
            request.getHeaders().put("X-B3-TraceId", currentSpan.context().traceIdString());
            request.getHeaders().put("X-B3-SpanId", currentSpan.context().spanIdString());
            request.getHeaders().put("X-B3-Sampled", currentSpan.context().sampled() != null ? currentSpan.context().sampled().toString() : "1");
        }
        return delegate.execute(request);
    }
}

2. 配置Eureka客户端加载自定义拦截器 #

接下来要让Eureka客户端用上我们的拦截器，通过自定义EurekaHttpClientFactory来替换默认的客户端工厂：

import com.netflix.discovery.EurekaHttpClient;
import com.netflix.discovery.EurekaHttpClientFactory;
import com.netflix.discovery.shared.transport.EurekaTransportConfig;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.cloud.netflix.eureka.EurekaClientConfigurer;
import org.springframework.cloud.netflix.eureka.EurekaClientFactoryBean;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

import java.util.Collections;

@Configuration
public class EurekaTraceConfig implements EurekaClientConfigurer {

    private final TraceEurekaHttpClientInterceptor traceInterceptor;

    @Autowired
    public EurekaTraceConfig(TraceEurekaHttpClientInterceptor traceInterceptor) {
        this.traceInterceptor = traceInterceptor;
    }

    @Override
    public void configureEurekaHttpClientFactory(EurekaClientFactoryBean factory) {
        factory.setEurekaHttpClientFactory(new EurekaHttpClientFactory() {
            @Override
            public EurekaHttpClient newClient(EurekaTransportConfig transportConfig) {
                // 获取默认Eureka客户端，再添加我们的追踪拦截器
                EurekaHttpClient defaultClient = EurekaHttpClientFactory.DEFAULT.newClient(transportConfig);
                return defaultClient.interceptWith(Collections.singletonList(traceInterceptor));
            }
        });
    }
}

3. 验证配置效果 #

做完上面两步后重启服务，你可以：

• 查看Eureka Server的访问日志，确认请求头里已经带上了X-B3-TraceId、X-B3-SpanId这些标识；
• 刷新Zipkin UI，就能看到从业务服务/网关到Eureka Server的完整链路节点了。

额外注意点 #

• 确保项目里已经正确引入spring-cloud-starter-sleuth和spring-cloud-starter-zipkin依赖，版本要和Edgware.SR1匹配（对应Sleuth 1.3.x版本）；
• 如果是多节点Eureka集群，不需要给Server端做额外配置，只要客户端传递了Trace头，Zipkin就能自动捕获链路。

内容的提问来源于stack exchange，提问作者Matvey