如何解决elastic-apm-node的APM Server传输错误(ENOTFOUND)

问题详情 #

• 部署架构：APM Server（apm-server）和Node.js微服务（test-service）均运行在Docker容器中，已通过elastic-apm-node配置APM Agent
• 异常现象：apm.start()初始化成功后，很快抛出getaddrinfo ENOTFOUND错误，无法连接到APM Server
• 错误日志：

2024-07-18 15:13:20 {"log.level":"error","@timestamp":"2024-07-18T19:13:20.963Z","log.logger":"elastic-apm-node","ecs.version":"8.10.0","message":"APM Server transport error (ENOTFOUND): getaddrinfo ENOTFOUND apm-server"}

• Agent代码：

const apm = require('elastic-apm-node').start({
  serviceName: 'test-service',
  serverUrl: "http://apm-server:8200",
  secretToken: process.env.ELASTIC_APM_SECRET_TOKEN
});

const app = require('express')()

app.get('/', function (req, res) {
  res.send('Hello World!')
})

app.listen(5001)

• docker-compose.yml关键片段：

apm-server:
  image: docker.elastic.co/apm/apm-server:8.14.2
  container_name: apm-server
  restart: always
  command: >
    apm-server -e
      -E apm-server.rum.enabled=true
      -E setup.kibana.host=kibana:5601
      -E setup.template.settings.index.number_of_replicas=0
      -E apm-server.kibana.enabled=true
      -E apm-server.kibana.host=kibana:5601
      -E apm-server.kibana.username:kibana_system
      -E apm-server.kibana.password:xxxxxx
      -E apm-server.kibana.protocol=http
      -E apm-server.secret_token=
      -E strict.perms=false
      -E apm-server.auth.anonymous.enabled=true
  ports:
    - "8200:8200"
  volumes:
    - ./apm-server.yml:/usr/share/apm-server/apm-server.yml:ro
  networks:
    - elastic
  depends_on:
    - elasticsearch

test-service:
  build:
    context: ../test-service
  container_name: test-service
  environment:
    - RABBITMQ_URL=amqp://rabbitmq
    - ELASTICSEARCH_NODE=http://elasticsearch:9200
    - ELASTIC_APM_SERVER_URL=http://apm-server:8200
    - ELASTIC_APM_SECRET_TOKEN=secrettoken
    - PORT=5001
  ports:
    - "5001:5001"
  depends_on:
    - elasticsearch
    - rabbitmq
    - apm-server

• 已尝试操作：切换ESM/CommonJS方式加载Agent，问题未解决

解决方案 #

错误原因 #

test-service容器和apm-server容器不在同一个Docker网络中。Docker自定义网络（如配置中的elastic）支持容器名自动DNS解析，但test-service未加入该网络，导致无法通过apm-server这个容器名找到对应的服务。

修复步骤 #

1. 修改docker-compose.yml，在test-service配置块中添加networks配置，加入elastic网络：

test-service:
  build:
    context: ../test-service
  container_name: test-service
  environment:
    - RABBITMQ_URL=amqp://rabbitmq
    - ELASTICSEARCH_NODE=http://elasticsearch:9200
    - ELASTIC_APM_SERVER_URL=http://apm-server:8200
    - ELASTIC_APM_SECRET_TOKEN=secrettoken
    - PORT=5001
  ports:
    - "5001:5001"
  depends_on:
    - elasticsearch
    - rabbitmq
    - apm-server
  # 添加网络配置
  networks:
    - elastic

2. 重新构建并启动test-service容器：

docker compose up -d --build test-service

补充说明 #

• depends_on仅控制容器的启动顺序，不保证网络连通性，即使test-service在apm-server之后启动，不在同一网络依然无法通信
• 默认的Docker bridge网络不支持容器名的自动DNS解析，必须使用自定义网络才能实现容器间通过名称访问

内容的提问来源于stack exchange，提问作者matt606