链路追踪系统Skywalking接入说明

• 应用修改
  • 引入依赖
  • 日志模板
  • 注意事项
• 部署修改
  • Dockerfile
    • 一般应用
    • spring-cloud-gateway
  • 部署脚本
    • docker
    • docker-compose

应用修改

引入依赖

目的是实现手动获取链路信息、链路信息日志记录等扩展功能。

   <properties>
     <skywalking.version>8.14.0</skywalking.version>
   </properties> 

  <dependencies>
    <!-- 省略其他依赖 -->
    <dependency>
      <groupId>org.apache.skywalking</groupId>
      <artifactId>apm-toolkit-logback-1.x</artifactId>
      <version>${skywalking.version}</version>
    </dependency>
    <dependency>
       <groupId>org.apache.skywalking</groupId>
       <artifactId>apm-toolkit-trace</artifactId>
       <version>${skywalking.version}</version>
    </dependency>
  </dependencies>

或者直接引入com.lms:common-log:

   <properties>
     <lms.version>main-4.0.0</lms.version>
   </properties> 
  <dependencies>
   <!-- 省略其他依赖 -->
   <dependency>
      <groupId>com.lms</groupId>
      <artifactId>common-log</artifactId>
      <version>${lms.version}</version>
   </dependency>
  </dependencies>

日志模板

修改logback配置文件，以实现将traceId和spanId(segmentId)输出到日志。

1. LOG_PATTERN

   <property name="LOG_PATTERN" value=
               "|%d{yyyy-MM-dd HH:mm:ss.SSS}|%-5level|%replace(%X{tid}){'TID:',''}|%replace(%X{sw_ctx}){'(([\\w\\[\\-\\.:@/]*,){3})|(,-?\\d+\\])',''}|${APP_NAME}|%t|%C|%M|%L|%m%n"/>
   

   log4j2与logback的replace函数有细微差别，如有需要可查阅log4j2文档。

2. appender.encoder

           <encoder class="ch.qos.logback.core.encoder.LayoutWrappingEncoder">
               <layout class="org.apache.skywalking.apm.toolkit.log.logback.v1.x.mdc.TraceIdMDCPatternLogbackLayout">
                   <Pattern>${LOG_PATTERN}</Pattern>
               </layout>
               <charset>UTF-8</charset>
           </encoder>
   

注意事项

1. 在程序中手动获取链路信息

   // traceId在整个链路中唯一，一个trace包含一个或多个segment
   String traceId = TraceContext.traceId();
   // segmentId在单个进程内唯一,一个segment包含一个或多个span
   String segmentId = TraceContext.segmentId();
   // span包含在segment内，通常表示一个线程或一个组件(如spring-mvc,mq,jdbc等)
   String spanId = TraceContext.spanId();
   

2. 链路信息跨线程解决方案
   默认情况下，链路信息只能在线程内传递，如果应用代码中有跨线程逻辑，需要手动处理:

   • usage 1

         @TraceCrossThread
         public static class MyCallable<String> implements Callable<String> {
             @Override
             public String call() throws Exception {
                 return null;
             }
         }
         ExecutorService executorService = Executors.newFixedThreadPool(1);
         executorService.submit(new MyCallable());
     

   • usage 2

         @TraceCrossThread
         public static class MyRunnable implements Runnable {
     
             @Override
             public void run() {
               // TODO
             }
         }
         ExecutorService executorService = Executors.newFixedThreadPool(1);
         executorService.submit(new MyRunnable());
     

   • usage 3

         ExecutorService executorService = Executors.newFixedThreadPool(1);
         executorService.submit(CallableWrapper.of(new Callable<String>() {
             @Override public String call() throws Exception {
                 return null;
             }
         }));
     

   • usage 4

         ExecutorService executorService = Executors.newFixedThreadPool(1);
         executorService.execute(RunnableWrapper.of(new Runnable() {
             @Override public void run() {
                 //TODO
             }
         }));
     

   • usage 5

         CompletableFuture.supplyAsync(SupplierWrapper.of(() -> {
             return "SupplierWrapper";
         })).thenApplyAsync(FunctionWrapper.of(f -> {
             return "FunctionWrapper";
         })).thenAcceptAsync(ConsumerWrapper.of(c -> {
             System.out.println("ConsumerWrapper");
         }));
     

部署修改

Dockerfile

一般应用

修改FROM基础镜像和WORKDIR，示例:

FROM apache/skywalking-java-agent:8.14.0-java11
MAINTAINER lms [email protected]

RUN /bin/cp /usr/share/zoneinfo/Asia/Shanghai /etc/localtime \&& echo 'Asia/Shanghai' >/etc/timezone

WORKDIR /
ADD target/analysis-boot.jar analysis-boot.jar

ENTRYPOINT ["java", "-Xmx98m", "-Djava.security.egd=file:/dev/./urandom", "-jar", "/analysis-boot.jar"]

EXPOSE 8806

spring-cloud-gateway

skywalking-java-agent以插件的形式提供对不同框架的支持，而spring-cloud-gateway和spring-webflux插件是可选插件，需要手动引入。示例:

FROM apache/skywalking-java-agent:8.14.0-java11
MAINTAINER lms [email protected]

RUN cp /usr/share/zoneinfo/Asia/Shanghai /etc/localtime \&& echo 'Asia/Shanghai' >/etc/timezone
# 添加网关所需插件
RUN cp /skywalking/agent/optional-plugins/apm-spring-cloud-gateway-3.x-plugin-8.14.0.jar /skywalking/agent/plugins/
RUN cp /skywalking/agent/optional-plugins/apm-spring-webflux-5.x-plugin-8.14.0.jar /skywalking/agent/plugins/

WORKDIR /
ADD target/lms-gateway.jar lms-gateway.jar

ENTRYPOINT ["java", "-Xmx98m", "-Djava.security.egd=file:/dev/./urandom", "-jar", "/lms-gateway.jar"]

EXPOSE 9999

部署脚本

docker

至少需要添加以下三个配置项，以环境变量形式注入:

更多配置项参考官方文档。

示例:

docker run -d -p 8806:8806 -e SW_AGENT_NAME=lms::analysis -e [email protected] -e SW_AGENT_COLLECTOR_BACKEND_SERVICES=192.168.3.39:11800 lms-anlysis:latest

docker-compose

示例:

version: '3.1'
services:
  analysis:
 build:
   context: ./
   dockerfile: Dockerfile
 image: analysis:latest
 container_name: lms-analysis
 ports:
   - 8806:8806
 volumes:
   - /app/logs/analysis:/logs
 environment:
   - SW_AGENT_NAME=lms::analysis
   - [email protected]
   - SW_AGENT_COLLECTOR_BACKEND_SERVICES=192.168.3.39:11800