piflow/README_CN.md

11 KiB
Raw Blame History

PiFlow是一个简单易用功能强大的大数据流水线系统。

目录

特性

  • 简单易用

    • 可视化配置流水线
    • 监控流水线
    • 查看流水线日志
    • 检查点功能
  • 扩展性强:

    • 支持自定义开发数据处理组件
  • 性能优越:

    • 基于分布式计算引擎Spark开发
  • 功能强大:

    • 提供100+的数据处理组件
    • 包括Hadoop 、Spark、MLlib、Hive、Solr、Redis、MemCache、ElasticSearch、JDBC、MongoDB、HTTP、FTP、XML、CSV、JSON等
    • 集成了微生物领域的相关算法

架构

要求

  • JDK 1.8 及以上版本
  • Apache Maven 3.1.0 及以上版本
  • Git Client
  • Spark-2.1.0 及以上版本
  • Hadoop-2.6.0 及以上版本

开始

如何Build: mvn clean package -Dmaven.test.skip=true

      [INFO] Replacing original artifact with shaded artifact.
      [INFO] Replacing /opt/project/piflow/piflow-server/target/piflow-server-0.9.jar with /opt/project/piflow/piflow-server/target/piflow-server-0.9-shaded.jar
      [INFO] ------------------------------------------------------------------------
      [INFO] Reactor Summary:
      [INFO] 
      [INFO] piflow-project ..................................... SUCCESS [  4.602 s]
      [INFO] piflow-core ........................................ SUCCESS [ 56.533 s]
      [INFO] piflow-bundle ...................................... SUCCESS [02:15 min]
      [INFO] piflow-server ...................................... SUCCESS [03:01 min]
      [INFO] ------------------------------------------------------------------------
      [INFO] BUILD SUCCESS
      [INFO] ------------------------------------------------------------------------
      [INFO] Total time: 06:18 min
      [INFO] Finished at: 2018-12-24T16:54:16+08:00
      [INFO] Final Memory: 41M/812M
      [INFO] ------------------------------------------------------------------------

如何运行Piflow Server

  • 使用Intellij Idea:

    • 编辑config.properties文件
    • build piflow工程生成piflow-server.jar
    • 运行cn.piflow.api.Main
    • 切记设置SPARK_HOME
  • 直接运行release版本:

    • 下载release版本地址https://github.com/cas-bigdatalab/piflow_release
    • 将build好的piflow-server.jar拷贝到piflow_release文件夹由于git不能上传超过1G大文件故需自行build piflow-server.jar
    • 编辑config.properties文件
    • 运行start.sh 或者后台运行 nohup ./start.sh > piflow.log 2>&1 &
  • 如何配置config.properties

    #server ip and port
    server.ip=10.0.86.191
    server.port=8002
    h2.port=50002
    
    #spark and yarn config
    spark.master=yarn
    spark.deploy.mode=cluster
    yarn.resourcemanager.hostname=10.0.86.191
    yarn.resourcemanager.address=10.0.86.191:8032
    yarn.access.namenode=hdfs://10.0.86.191:9000
    yarn.stagingDir=hdfs://10.0.86.191:9000/tmp/
    yarn.jars=hdfs://10.0.86.191:9000/user/spark/share/lib/*.jar
    yarn.url=http://10.0.86.191:8088/ws/v1/cluster/apps/
    
    #hive config
    hive.metastore.uris=thrift://10.0.86.191:9083
    
    #piflow-server.jar path
    piflow.bundle=/opt/piflowServer/piflow-server-0.9.jar
    
    #checkpoint hdfs path
    checkpoint.path=hdfs://10.0.86.89:9000/piflow/checkpoints/
    
    #debug path
    debug.path=hdfs://10.0.88.191:9000/piflow/debug/
    
    #yarn url
    yarn.url=http://10.0.86.191:8088/ws/v1/cluster/apps/
    
    #the count of data shown in log
    data.show=10
    
    #h2 db port
    h2.port=50002
    

如何运行Piflow Web

如何使用:

  • 命令行方式
    • 流水线样例配置

      {
        "flow":{
        "name":"test",
        "uuid":"1234",
        "checkpoint":"Merge",
        "stops":[
        {
          "uuid":"1111",
          "name":"XmlParser",
          "bundle":"cn.piflow.bundle.xml.XmlParser",
          "properties":{
              "xmlpath":"hdfs://10.0.86.89:9000/xjzhu/dblp.mini.xml",
              "rowTag":"phdthesis"
          }
        },
        {
          "uuid":"2222",
          "name":"SelectField",
          "bundle":"cn.piflow.bundle.common.SelectField",
          "properties":{
              "schema":"title,author,pages"
          }
      
        },
        {
          "uuid":"3333",
          "name":"PutHiveStreaming",
          "bundle":"cn.piflow.bundle.hive.PutHiveStreaming",
          "properties":{
              "database":"sparktest",
              "table":"dblp_phdthesis"
          }
        },
        {
          "uuid":"4444",
          "name":"CsvParser",
          "bundle":"cn.piflow.bundle.csv.CsvParser",
          "properties":{
              "csvPath":"hdfs://10.0.86.89:9000/xjzhu/phdthesis.csv",
              "header":"false",
              "delimiter":",",
              "schema":"title,author,pages"
          }
        },
        {
          "uuid":"555",
          "name":"Merge",
          "bundle":"cn.piflow.bundle.common.Merge",
          "properties":{
            "inports":"data1,data2"
          }
        },
        {
          "uuid":"666",
          "name":"Fork",
          "bundle":"cn.piflow.bundle.common.Fork",
          "properties":{
            "outports":"out1,out2,out3"
          }
        },
        {
          "uuid":"777",
          "name":"JsonSave",
          "bundle":"cn.piflow.bundle.json.JsonSave",
          "properties":{
            "jsonSavePath":"hdfs://10.0.86.89:9000/xjzhu/phdthesis.json"
          }
        },
        {
          "uuid":"888",
          "name":"CsvSave",
          "bundle":"cn.piflow.bundle.csv.CsvSave",
          "properties":{
            "csvSavePath":"hdfs://10.0.86.89:9000/xjzhu/phdthesis_result.csv",
            "header":"true",
            "delimiter":","
          }
        }
      ],
      "paths":[
        {
          "from":"XmlParser",
          "outport":"",
          "inport":"",
          "to":"SelectField"
        },
        {
          "from":"SelectField",
          "outport":"",
          "inport":"data1",
          "to":"Merge"
        },
        {
          "from":"CsvParser",
          "outport":"",
          "inport":"data2",
          "to":"Merge"
        },
        {
          "from":"Merge",
          "outport":"",
          "inport":"",
          "to":"Fork"
        },
        {
          "from":"Fork",
          "outport":"out1",
          "inport":"",
          "to":"PutHiveStreaming"
        },
        {
          "from":"Fork",
          "outport":"out2",
          "inport":"",
          "to":"JsonSave"
        },
        {
          "from":"Fork",
          "outport":"out3",
          "inport":"",
          "to":"CsvSave"
        }
      ]
      

      } }

    • 运行命令

      • curl -0 -X POST http://serverIP:serverPort/flow/start -H "Content-type: application/json" -d '你的流水线json配置文件'
  • 访问piflow web: 试运行地址 "http://piflow.ml/piflow-web", user/password: admin/admin
    • 登录
    • 流水线列表
    • 流水线配置
    • 流水线资源配置
    • 运行流水线
    • 删除流水线
    • 流水线保存模板
    • 创建流水线:用户点击创建按钮,需要输入流水线名称及描述信息,同时可设置流水线需要的资源.
    • 配置流水线:用户可通过拖拽方式进行流水线的配置方式类似visio
    • 搜索流水线组件:画布左边栏显示组件组和组件,可按关键字搜索,户选择好组件后可拖至画布中央
    • 流水线基本信息:画布右侧显示流水线基本信息,包括流水线名称及描述
    • 流水线配置:画布中央选择任一数据处理组件,右侧显示该数据处理组件的基本信息,包括名称描述,作者等信息.选择AttributeInfo tab,显示该数据处理组件的属性信息,用户可根据实际需求进行配置
    • 运行流水线:用户配置好流水线后,可点击运行按钮运行流水线
    • 流水线监控:进入流水线监控页面。监控页面会显示整条流水线的执行状况,包括运行状态、执行进度、执行时间等,击具体数据处理组件,显示该数据处理组件的运行状况,包括运行状态、执行时间。
    • 查看流水线日志
    • 运行中流水线列表: 已运行流水线会显示在Process List中包括开始时间、结束时间、进度、状态等。同时可对已运行流水线进行查看在运行停止和删除操作
    • 运行流水线检查点
    • 创建保存模板
    • 模板列表
    • 下载模板模板会保存成xml文件存放到本地
    • 上传模板
    • 加载模板