Apache Dolphin Scheduler

Apache DolphinScheduler 是一个分布式去中心化，易扩展的可视化 DAG 工作流任务调度系统。简称 DS，包括 Web 及若干服务，它依赖 PostgreSQL 和 Zookeeper，自身的服务模块包括：api, alert, master, worker（有一个 logger 服务，运行在 worker 中）等。详细部署可以参考：Docker 部署 Dolphin Scheduler

官方提供了 docker-compose.yml，位于项目的 docker/docker-swarm/ 目录下，本文以 v1.3.8 版本为例，讲解 docker-compose.yml 内的具体内容，该版本的 Compose 基于 apache/dolphinscheduler:1.3.8 的 Docker 镜像，DS Docker 构建可以参考之前写的这篇博客：Apache Dolphin Scheduler - Dockerfile 详解，主要的配置修改、流程启动都封装在 Dockerfile 中

Docker Compose

version: "3.1"

services:
  
  # PostgreSQL
  dolphinscheduler-postgresql:
    image: postgres:11.12
    environment:
      # 设置时区
      TZ: Asia/Shanghai
      # PostgreSQL 相关的配置
      POSTGRES_USER: root
      POSTGRES_PASSWORD: root
      POSTGRES_DB: dolphinscheduler
    # 数据卷
    volumes:
    - dolphinscheduler-postgresql:/var/lib/postgresql/data
    # 重启策：在容器退出时总是重启容器
    restart: unless-stopped
    # 配置网络
    networks:
    - dolphinscheduler

  # Zookeeper
  dolphinscheduler-zookeeper:
    image: zookeeper:3.6.3
    environment:
      TZ: Asia/Shanghai
      # Zookeeper 相关配置
      ZOO_DATA_LOG_DIR: /data
      ZOO_4LW_COMMANDS_WHITELIST: srvr,ruok,wchs,cons
    volumes:
    - dolphinscheduler-zookeeper:/data
    restart: unless-stopped
    networks:
    - dolphinscheduler

  # DS 服务模块
  dolphinscheduler-api:
    image: apache/dolphinscheduler:1.3.8
    command: api-server
    ports:
    - 12345:12345
    environment:
      TZ: Asia/Shanghai
    # 引入外部环境变量
    env_file: config.env.sh
    # 健康检查
    healthcheck:
      test: ["CMD", "/root/checkpoint.sh", "ApiApplicationServer"]
      interval: 30s
      timeout: 5s
      retries: 3
    # 依赖 PostgreSQL 和 Zookeeper
    depends_on:
    - dolphinscheduler-postgresql
    - dolphinscheduler-zookeeper
    volumes:
    - dolphinscheduler-logs:/opt/dolphinscheduler/logs
    - dolphinscheduler-shared-local:/opt/soft
    - dolphinscheduler-resource-local:/dolphinscheduler
    restart: unless-stopped
    networks:
    - dolphinscheduler

  dolphinscheduler-alert:
    image: apache/dolphinscheduler:1.3.8
    command: alert-server
    environment:
      TZ: Asia/Shanghai
    env_file: config.env.sh
    healthcheck:
      test: ["CMD", "/root/checkpoint.sh", "AlertServer"]
      interval: 30s
      timeout: 5s
      retries: 3
    depends_on:
    - dolphinscheduler-postgresql
    volumes:
    - dolphinscheduler-logs:/opt/dolphinscheduler/logs
    restart: unless-stopped
    networks:
    - dolphinscheduler

  dolphinscheduler-master:
    image: apache/dolphinscheduler:1.3.8
    command: master-server
    environment:
      TZ: Asia/Shanghai
    env_file: config.env.sh
    healthcheck:
      test: ["CMD", "/root/checkpoint.sh", "MasterServer"]
      interval: 30s
      timeout: 5s
      retries: 3
    depends_on:
    - dolphinscheduler-postgresql
    - dolphinscheduler-zookeeper
    volumes:
    - dolphinscheduler-logs:/opt/dolphinscheduler/logs
    - dolphinscheduler-shared-local:/opt/soft
    restart: unless-stopped
    networks:
    - dolphinscheduler

  dolphinscheduler-worker:
    image: apache/dolphinscheduler:1.3.8
    command: worker-server
    environment:
      TZ: Asia/Shanghai
    env_file: config.env.sh
    healthcheck:
      test: ["CMD", "/root/checkpoint.sh", "WorkerServer"]
      interval: 30s
      timeout: 5s
      retries: 3
    depends_on:
    - dolphinscheduler-postgresql
    - dolphinscheduler-zookeeper
    volumes:
    - dolphinscheduler-worker-data:/tmp/dolphinscheduler
    - dolphinscheduler-logs:/opt/dolphinscheduler/logs
    - dolphinscheduler-shared-local:/opt/soft
    - dolphinscheduler-resource-local:/dolphinscheduler
    restart: unless-stopped
    networks:
    - dolphinscheduler
# 声明使用到的网络
networks:
  dolphinscheduler:
    driver: bridge

# 声明使用到的数据卷
volumes:
  dolphinscheduler-postgresql:
  dolphinscheduler-zookeeper:
  dolphinscheduler-worker-data:
  dolphinscheduler-logs:
  dolphinscheduler-shared-local:
  dolphinscheduler-resource-local:

每一个 service 都定义了 TZ 的环境变量，设置容器的时区为亚洲上海，restart 重启策略都设置为：unless-stopped，即：在容器退出时总是重启容器

在 yml 的最后定义了 Compose 使用到的 networks 和 volumes

所有 service 使用同一个网络：dolphinscheduler，driver 定义为：bridge，默认就是 bridge，bridge 用于应用部署在不同容器，它们之间需要通信的情况

DS 的每个服务模块都通过 env_file 导入独立的环境变量文件 config.env.sh

healthcheck 是健康检查，调用容器内的 checkpoint.sh，并传入服务名称，检查该 Java 进程是否存在。两次健康检查的间隔 30s，超时时间为 5s，如果超过这个时间，本次健康检查就被视为失败，retries 重试次数设置为 3，当连续失败指定次数后，则将容器状态视为 unhealthy

PostgreSQL：POSTGRES_USER, POSTGRES_PASSWORD, POSTGRES_DB 分别定义了 PostgreSQL 的用户名、密码、一个名为：dolphinscheduler 的数据库

Zookeeper：添加环境变量 ZOO_4LW_COMMANDS_WHITELIST: srvr,ruok,wchs,cons，把这四个命令加入白名单，避免在使用这四个四字命令时提示：stat is not executed because it is not in the whitelist

conf 配置信息

#============================================================================
# Database
#============================================================================
# postgresql
DATABASE_TYPE=postgresql
DATABASE_DRIVER=org.postgresql.Driver
DATABASE_HOST=dolphinscheduler-postgresql
DATABASE_PORT=5432
DATABASE_USERNAME=root
DATABASE_PASSWORD=root
DATABASE_DATABASE=dolphinscheduler
DATABASE_PARAMS=characterEncoding=utf8
# mysql
# DATABASE_TYPE=mysql
# DATABASE_DRIVER=com.mysql.jdbc.Driver
# DATABASE_HOST=dolphinscheduler-mysql
# DATABASE_PORT=3306
# DATABASE_USERNAME=root
# DATABASE_PASSWORD=root
# DATABASE_DATABASE=dolphinscheduler
# DATABASE_PARAMS=useUnicode=true&characterEncoding=UTF-8

#============================================================================
# ZooKeeper
#============================================================================
ZOOKEEPER_QUORUM=dolphinscheduler-zookeeper:2181
ZOOKEEPER_ROOT=/dolphinscheduler

#============================================================================
# Common
#============================================================================
# common opts
DOLPHINSCHEDULER_OPTS=
# common env
DATA_BASEDIR_PATH=/tmp/dolphinscheduler
RESOURCE_STORAGE_TYPE=HDFS
RESOURCE_UPLOAD_PATH=/dolphinscheduler
FS_DEFAULT_FS=file:///
FS_S3A_ENDPOINT=s3.xxx.amazonaws.com
FS_S3A_ACCESS_KEY=xxxxxxx
FS_S3A_SECRET_KEY=xxxxxxx
HADOOP_SECURITY_AUTHENTICATION_STARTUP_STATE=false
JAVA_SECURITY_KRB5_CONF_PATH=/opt/krb5.conf
LOGIN_USER_KEYTAB_USERNAME=hdfs@HADOOP.COM
LOGIN_USER_KEYTAB_PATH=/opt/hdfs.keytab
KERBEROS_EXPIRE_TIME=2
HDFS_ROOT_USER=hdfs
RESOURCE_MANAGER_HTTPADDRESS_PORT=8088
YARN_RESOURCEMANAGER_HA_RM_IDS=
YARN_APPLICATION_STATUS_ADDRESS=http://ds1:8088/ws/v1/cluster/apps/%s
# skywalking
SKYWALKING_ENABLE=false
SW_AGENT_COLLECTOR_BACKEND_SERVICES=127.0.0.1:11800
SW_GRPC_LOG_SERVER_HOST=127.0.0.1
SW_GRPC_LOG_SERVER_PORT=11800
# dolphinscheduler env
HADOOP_HOME=/opt/soft/hadoop
HADOOP_CONF_DIR=/opt/soft/hadoop/etc/hadoop
SPARK_HOME1=/opt/soft/spark1
SPARK_HOME2=/opt/soft/spark2
PYTHON_HOME=/usr/bin/python
JAVA_HOME=/usr/local/openjdk-8
HIVE_HOME=/opt/soft/hive
FLINK_HOME=/opt/soft/flink
DATAX_HOME=/opt/soft/datax

#============================================================================
# Master Server
#============================================================================
MASTER_SERVER_OPTS=-Xms1g -Xmx1g -Xmn512m
MASTER_EXEC_THREADS=100
MASTER_EXEC_TASK_NUM=20
MASTER_DISPATCH_TASK_NUM=3
MASTER_HOST_SELECTOR=LowerWeight
MASTER_HEARTBEAT_INTERVAL=10
MASTER_TASK_COMMIT_RETRYTIMES=5
MASTER_TASK_COMMIT_INTERVAL=1000
MASTER_MAX_CPULOAD_AVG=-1
MASTER_RESERVED_MEMORY=0.3

#============================================================================
# Worker Server
#============================================================================
WORKER_SERVER_OPTS=-Xms1g -Xmx1g -Xmn512m
WORKER_EXEC_THREADS=100
WORKER_HEARTBEAT_INTERVAL=10
WORKER_MAX_CPULOAD_AVG=-1
WORKER_RESERVED_MEMORY=0.3
WORKER_GROUPS=default

#============================================================================
# Alert Server
#============================================================================
ALERT_SERVER_OPTS=-Xms512m -Xmx512m -Xmn256m
# xls file
XLS_FILE_PATH=/tmp/xls
# mail
MAIL_SERVER_HOST=
MAIL_SERVER_PORT=
MAIL_SENDER=
MAIL_USER=
MAIL_PASSWD=
MAIL_SMTP_STARTTLS_ENABLE=true
MAIL_SMTP_SSL_ENABLE=false
MAIL_SMTP_SSL_TRUST=
# wechat
ENTERPRISE_WECHAT_ENABLE=false
ENTERPRISE_WECHAT_CORP_ID=
ENTERPRISE_WECHAT_SECRET=
ENTERPRISE_WECHAT_AGENT_ID=
ENTERPRISE_WECHAT_USERS=

#============================================================================
# Api Server
#============================================================================
API_SERVER_OPTS=-Xms512m -Xmx512m -Xmn256m

#============================================================================
# Logger Server
#============================================================================
LOGGER_SERVER_OPTS=-Xms512m -Xmx512m -Xmn256m

config.env.sh 定义了用到的配置，通过 env_file 的方式传入容器，它会覆盖容器内的默认配置

参考资料

Networking overview
Zookeeper 四字命令
zookeeper四字命令提示命令不在白名单中
The “env_file” configuration option