分布式链路追踪实战：OpenTelemetry与Jaeger全链路追踪部署方案

微服务架构中一次请求跨越多个服务节点，故障定位和性能分析需要全链路追踪支持。OpenTelemetry是CNCF的观测性标准，统一了Traces、Metrics和Logs三种信号的采集协议。Jaeger是Uber开源的分布式追踪后端，支持Span存储、查询和可视化。分布式链路追踪通过Trace ID关联各服务节点的Span，还原完整调用链路。

OpenTelemetry SDK集成与Span传播机制

OpenTelemetry通过Trace Context和Baggage两种传播协议在服务间传递追踪上下文。W3C Trace Context使用traceparent HTTP头传递Trace ID、Span ID和采样标志。

Java应用集成OpenTelemetry SDK：

// pom.xml依赖
<dependency>
    <groupId>io.opentelemetry.instrumentation</groupId>
    <artifactId>opentelemetry-spring-boot-starter</artifactId>
    <version>2.6.0</version>
</dependency>
<dependency>
    <groupId>io.opentelemetry</groupId>
    <artifactId>opentelemetry-exporter-otlp</artifactId>
    <version>1.39.0</version>
</dependency>

application.yml配置OTLP导出器：

otel:
  service.name: order-service
  exporter.otlp.endpoint: http://otel-collector:4317
  exporter.otlp.protocol: grpc
  traces.exporter: otlp
  sampler.parentbased: true
  sampler.traceidratiobased: 0.1

traces.exporter设为otlp通过gRPC发送Span数据到Collector。sampler.traceidratiobased设为0.1表示10%采样率，生产环境中高流量服务可降低采样率以减少存储开销。

手动创建Span记录业务逻辑：

import io.opentelemetry.api.trace.Tracer;
import io.opentelemetry.api.GlobalOpenTelemetry;

public class OrderService {
    private static final Tracer tracer = GlobalOpenTelemetry.getTracer("order-service");
    
    public Order createOrder(OrderRequest request) {
        Span span = tracer.spanBuilder("createOrder")
            .setAttribute("order.userId", request.getUserId())
            .setAttribute("order.amount", request.getAmount())
            .startSpan();
        
        try (Scope scope = span.makeCurrent()) {
            // 验证库存
            Span inventorySpan = tracer.spanBuilder("checkInventory").startSpan();
            boolean available = inventoryService.check(request.getProductId());
            inventorySpan.setAttribute("inventory.available", available);
            inventorySpan.end();
            
            if (!available) {
                span.recordException(new RuntimeException("库存不足"));
                span.setStatus(StatusCode.ERROR);
                throw new BusinessException("库存不足");
            }
            
            // 创建订单
            Order order = orderRepository.save(request);
            span.setAttribute("order.id", order.getId());
            return order;
        } finally {
            span.end();
        }
    }
}

setAttribute添加业务标签，便于在Jaeger UI中按标签查询。recordException记录异常堆栈，setStatus标记Span状态。子Span（checkInventory）自动继承父Span的Trace ID。

OpenTelemetry Collector管道配置与数据转发

Collector是OpenTelemetry的核心组件，负责接收、处理和导出遥测数据。支持多接收器（OTLP gRPC/HTTP、Zipkin、Jaeger）和多导出器（Jaeger、Prometheus、Elasticsearch）。

# otel-collector-config.yaml
receivers:
  otlp:
    protocols:
      grpc:
        endpoint: 0.0.0.0:4317
      http:
        endpoint: 0.0.0.0:4318

processors:
  batch:
    timeout: 5s
    send_batch_size: 1024
  memory_limiter:
    check_interval: 2s
    limit_mib: 512
  filter:
    error_mode: ignore
    traces:
      span:
        - 'attributes["http.status_code"] == 404'

exporters:
  otlp/jaeger:
    endpoint: jaeger:4317
    tls:
      insecure: true
  logging:
    loglevel: warn

service:
  pipelines:
    traces:
      receivers: [otlp]
      processors: [memory_limiter, filter, batch]
      exporters: [otlp/jaeger]

batch处理器批量发送Span减少网络请求。memory_limiter防止Collector内存溢出。filter处理器过滤404状态的Span，减少无效数据存储。

Jaeger后端部署与查询分析

Jaeger支持内存存储（测试用）和Elasticsearch/Cassandra存储（生产用）。Docker Compose部署Jaeger + Elasticsearch：

version: "3.8"
services:
  jaeger:
    image: jaegertracing/all-in-one:1.57
    environment:
      SPAN_STORAGE_TYPE: elasticsearch
      ES_SERVER_URLS: http://elasticsearch:9200
      ES_TAGS_AS_FIELDS_ALL: "true"
    ports:
      - "16686:16686"
    depends_on:
      - elasticsearch
  
  elasticsearch:
    image: docker.elastic.co/elasticsearch/elasticsearch:8.13.0
    environment:
      discovery.type: single-node
      xpack.security.enabled: "false"
      ES_JAVA_OPTS: "-Xms2g -Xmx2g"

Jaeger UI（端口16686）支持按服务名、Trace ID、标签和持续时间查询。Find Traces页面中设置minDuration可筛选慢请求，Tags搜索框输入order.userId=12345可按业务字段精确查询。

Trace详情页展示调用链路的瀑布图，每个Span显示服务名、操作名、起止时间和耗时占比。错误Span以红色标记，展开可查看异常堆栈和标签。通过Trace Flame Graph视图直观展示并发Span和串行调用关系。