KUDU学习笔记-spark-kudu

启动kudu

cd /etc/init.d/

—启动master
[hadoop@hadoop000 init.d]$ sudo ./kudu-master start
Started Kudu Master Server (kudu-master): [ OK ]

—启动tserver
[hadoop@hadoop000 init.d]$ sudo ./kudu-tserver start
Started Kudu Tablet Server (kudu-tserver): [ OK ]

(
/// 会 由于时间同步原因，导致启动失败 ；可通过一下命令
sudo yum install ntp
service ntpd start

service ntpd status
)

— 前往浏览器查看进程
http://hadoop000:8050/

----在IDEA 中添加相关依赖

org.apache.kudu
kudu-client
1.7.0

KUDU-API 的开发

import java.util

import org.apache.kudu.{ColumnSchema, Schema, Type}
import org.apache.kudu.client.{CreateTableOptions, KuduClient}

/**
  * Created by laiji on 2021/12/26.
  */
object kudu_API {

  def createTable(client: KuduClient, tableName: String): Unit = {
    import scala.collection.JavaConverters._
    val columns=List(
      new ColumnSchema.ColumnSchemaBuilder("word", Type.STRING).key(true).build(),
      new ColumnSchema.ColumnSchemaBuilder("cnt", Type.INT32).build()
    ).asJava
    val schema = new Schema(columns)

    val options = new CreateTableOptions()
    options.setNumReplicas(1)
    val parcols = new util.LinkedList[String]()

    parcols.add("word")

    options.addHashPartitions(parcols,3)

    client.createTable(tableName,schema,options)
  }

  def insertRows(client: KuduClient, tableName: String): Unit = {

    val table = client.openTable(tableName)

    val session = client.newSession()

    for (i <- 1 to 30) {
      val insert = table.newInsert()
      val row = insert.getRow
      row.addString("word",s"lj+$i")
      row.addInt("cnt",i*i)
      session.apply(insert)
    }

  }

  def deltable(client: KuduClient, tableName: String): Unit = {
    client.deleteTable(tableName)
  }

  def querytable(client: KuduClient, tableName: String) = {
    val table = client.openTable(tableName)
    val scanner = client.newScannerBuilder(table).build()

    while (scanner.hasMoreRows){
      val iterator = scanner.nextRows()

      while (iterator.hasNext){
        val result = iterator.next()
        println(result.getString("word")+"==>"+result.getInt("cnt"))
      }

    }



  }

  def updaterow(client: KuduClient, tableName: String) ={
    val table = client.openTable(tableName)
    val session = client.newSession()
    
    val update = table.newUpdate()

    val row = update.getRow

    row.addString("word","lj+9")
    row.addInt("cnt",19191919)
    session.apply(update)
  }

  def main(args: Array[String]): Unit = {

    val kudu_master="hadoop000"

    val client = new KuduClient.KuduClientBuilder(kudu_master).build()

    //创建一张表
    val tableName="ljj-1226"
    // 创建表
//    createTable(client,tableName)

    // 插入数据表
//    insertRows(client,tableName)

    // 删除表
//    deltable(client,tableName)


    // 查询表
    querytable(client,tableName)

    // 修改数据
    updaterow(client,tableName)

    querytable(client,tableName)

    client.close()
  }
}

spark -读写 kudu

import Util_ip.{new_schema, sink2kudu}
import org.apache.kudu.Schema
import org.apache.spark.sql.{DataFrame, SparkSession}
import trai.data_etl_process

/**
  * Created by laiji on 2022/1/15.
  */
object LogETL_1 extends data_etl_process{
  override def process(session: SparkSession): Unit ={


    val kudu_master="hadoop000"
    import session.implicits._
    val tableName="ods_20181007"
    val df = session.read.format("org.apache.kudu.spark.kudu")
      .option("kudu.table", tableName)
      .option("kudu.master","hadoop000")
      .option("kudu.operation.timeout.ms", "100000").load




    // 需要写入的表
    val frame = df.sample(0.01)
    val frame1 = frame.select($"ip",$"adplatformkey",$"provincename")

    frame1.printSchema()


    println("===========写入 kudu 表中==============")
    val table_name_new="renew_20220115"
    val mainkey="ip"
    sink2kudu.write2kudu(kudu_master:String,table_name_new:String,new_schema.renew_20220115_schema: Schema,frame1:DataFrame,mainkey:String)




    print("===========验证该表是否有写入===========")

    //
    val df_check = session.read.format("org.apache.kudu.spark.kudu")
      .option("kudu.table", table_name_new)
      .option("kudu.master","hadoop000")
      .option("kudu.operation.timeout.ms", "100000").load

    df_check.show()
    println("棒极了")
    
  }
}

工具类-写入kudu 工具类 定义

package Util_ip

import java.util

import org.apache.kudu.{ColumnSchema, Schema, Type}
import org.apache.kudu.client.CreateTableOptions
import org.apache.kudu.client.KuduClient.KuduClientBuilder
import org.apache.spark.sql.DataFrame

/**
  * Created by laiji on 2022/1/15.
  */
object sink2kudu {
  def write2kudu(
                  kudu_master:String,table_name_new:String,schema: Schema,frame:DataFrame,mainkey:String
                ): Unit ={
    //     创建连接口
    val client = new KuduClientBuilder(kudu_master).build()
    if (client.tableExists(table_name_new))
      {client.deleteTable(table_name_new)
      println("已把旧表删除。。。")}
    else
      {println(table_name_new + "该表不存在,开始创建")}
    import scala.collection.JavaConverters._
    val options = new CreateTableOptions()
    options.setNumReplicas(1)
    val parcols = new util.LinkedList[String]()
    parcols.add(mainkey)
    options.addHashPartitions(parcols,3)
    client.createTable(table_name_new,schema,options)
    println(table_name_new + "已创建成功")



    frame.write
      .options(Map("kudu.master" -> kudu_master , "kudu.table" -> table_name_new))
      .mode("append")
      .format("org.apache.kudu.spark.kudu").save()
    println("已成功写入表")


  }

}

schema 工具类

package Util_ip
import org.apache.kudu.{ColumnSchema, Schema, Type}
import org.apache.kudu.client.CreateTableOptions
import org.apache.kudu.client.KuduClient.KuduClientBuilder
import org.apache.spark.sql.DataFrame
/**
  * Created by laiji on 2022/1/15.
  */
object new_schema {
  def renew_20220115_schema:Schema={
//    val frame1 = frame.select($"ip",$"adplatformkey",$"provincename")

    import scala.collection.JavaConverters._
    val columns=List(
      new ColumnSchema.ColumnSchemaBuilder("ip", Type.STRING).key(true).build(),
      new ColumnSchema.ColumnSchemaBuilder("adplatformkey", Type.STRING).build(),
      new ColumnSchema.ColumnSchemaBuilder("provincename", Type.STRING).build()
    ).asJava
    val schema = new Schema(columns)
    schema
  }


  def ctye_age_schema:Schema={
    //    val frame1 = frame.select($"ip",$"adplatformkey",$"provincename")

    import scala.collection.JavaConverters._
    val columns=List(
      new ColumnSchema.ColumnSchemaBuilder("cityname", Type.STRING).key(true).build(),
      new ColumnSchema.ColumnSchemaBuilder("provincename", Type.STRING).build(),
      new ColumnSchema.ColumnSchemaBuilder("appname", Type.STRING).build(),
    new ColumnSchema.ColumnSchemaBuilder("age_sum", Type.DOUBLE).build()
    ).asJava
    val schema = new Schema(columns)
    schema
  }



  def sdkversion_appname_schema:Schema={
    //    val frame1 = frame.select($"ip",$"adplatformkey",$"provincename")

    import scala.collection.JavaConverters._
    val columns=List(
      new ColumnSchema.ColumnSchemaBuilder("sdkversion", Type.STRING).key(true).build(),
      new ColumnSchema.ColumnSchemaBuilder("appname", Type.STRING).build(),
      new ColumnSchema.ColumnSchemaBuilder("month", Type.INT32).build(),
      new ColumnSchema.ColumnSchemaBuilder("age_sum", Type.DOUBLE).build()
    ).asJava
    val schema = new Schema(columns)
    schema
  }

}