1、 通过访问hive metastore的方式,这种方式通过访问hive的metastore元数据的方式获取表结构信息和该表数据所存放的HDFS路径,这种方式的特点是效率高、数据吞吐量大、使用spark操作起来更加友好。
2、 通过spark jdbc的方式访问,就是通过链接hiveserver2的方式获取数据,这种方式底层上跟spark链接其他rdbms上一样,可以采用sql的方式先在其数据库中查询出来结果再获取其结果数据,这样大部分数据计算的压力就放在了数据库上。
两种方式的具体实现示例
首先创建Spark Session对象:
val spark = SparkSession.builder()
.appName("test")
.enableHiveSupport()
.getOrCreate()
方式一(推荐) 直接采用Spark on Hive的方式读取数据,这样SparkSession在使用sql的时候会去找集群hive中的库表,加载其hdfs数据与其元数据组成DataFrame
val df = spark.sql("select * from test.user_info")
方式二 采用spark jdbc的方式,如果有特别的使用场景的话也可以通过这种方法来实现。
import org.apache.spark.sql.SparkSession
import org.apache.spark.sql.jdbc.{JdbcDialect, JdbcDialects}
object test{
def main(args: Array[String]): Unit = {
val spark = SparkSession.builder()
.master("local[2]")
.appName("test")
.getOrCreate()
register() //如果不手动注册,只能获取到数据库中的表结构,而不能获取到数据
val df = spark.read
.format("jdbc")
.option("driver","org.apache.hive.jdbc.HiveDriver")
.option("url","jdbc:hive2://xxx:10000/")
.option("user","hive")
.option("password",xxx)
.option("fetchsize", "2000")
.option("dbtable","test.user_info")
.load()
df.show(10)
}
def register(): Unit = {
JdbcDialects.registerDialect(HiveSqlDialect)
}
case object HiveSqlDialect extends JdbcDialect {
override def canHandle(url: String): Boolean = url.startsWith("jdbc:hive2")
override def quoteIdentifier(colName: String): String = {
colName.split('.').map(part => s"`$part`").mkString(".")
}
}
}
DataFrame是Spark SQL提供的一个编程抽象,与RDD类似,也是一个分布式的数据集合。但与RDD不同的是,DataFrame的数据都被组织到有名字的列中,就像关系型数据库中的表一样。此外,多种数据都可以转化为DataFrame,例如Spark计算过程中生成的RDD、结构化数据文件、Hive中的表、外部数据库等。
在Spark中,一个DataFrame所代表的是一个元素类型为Row的Dataset,即DataFrame只是Dataset[Row]的一个类型别名。相对于RDD,Dataset提供了强类型支持,在RDD的每行数据加了类型约束。而且使用DatasetAPI同样会经过Spark SQL优化器的优化,从而提高程序执行效率。
DataFrame和R的数据结构以及python pandas DataFrame的数据结构和操作基本一致。
object MovieLenDataSet {
case class User(UserID:String, Gender:String, Age:String, Occupation:String, Zip_Code:String)
def main(args: Array[String]): Unit = {
Logger.getLogger("org").setLevel(Level.ERROR)
val spark = SparkSession.builder()
.appName("MovieLenDataSet")
.master("local[*]")
.getOrCreate()
import spark.implicits._
val dataPath = "/home/ffzs/data/ml-1m"
val schema4users = StructType(
"UserID::Gender::Age::Occupation::Zip_code"
.split("::")
.map(it => StructField(it, StringType, nullable = true))
)
val usersRdd = spark.sparkContext.textFile(f"$dataPath/users.dat")
val usersRows = usersRdd.map(_.split("::"))
.map(it => {
it.map(_.trim)
})
.map(it => Row(it(0), it(1), it(2), it(3), it(4)))
val usersDF: DataFrame = spark.createDataFrame(usersRows, schema4users)
val usersDataSet = usersDF.as[User]
usersDataSet.show(5)
}
}
spark的dataframe存储中都会调用write的mode方法:
data.write.mode(“append”).saveAsTable(s"u s e r i d . {userid}.userid.{datasetid}")
data.write.mode(SaveMode.Overwrite).parquet(hdfspath)
但不同时候的参数是不同的。
先看一下源码:
spark-v2.3.0:
def mode(saveMode: SaveMode): DataFrameWriter[T] = {
this.mode = saveMode
this
}
/**
* Specifies the behavior when data or table already exists. Options include:
* - `overwrite`: overwrite the existing data.
* - `append`: append the data.
* - `ignore`: ignore the operation (i.e. no-op).
* - `error` or `errorifexists`: default option, throw an exception at runtime.
*
* @since 1.4.0
*/
def mode(saveMode: String): DataFrameWriter[T] = {
this.mode = saveMode.toLowerCase(Locale.ROOT) match {
case "overwrite" => SaveMode.Overwrite
case "append" => SaveMode.Append
case "ignore" => SaveMode.Ignore
case "error" | "errorifexists" | "default" => SaveMode.ErrorIfExists
case _ => throw new IllegalArgumentException(s"Unknown save mode: $saveMode. " +
"Accepted save modes are 'overwrite', 'append', 'ignore', 'error', 'errorifexists'.")
}
this
}
SaveMode.Overwrite(对应着字符串"overwrite"):表示如果目标文件目录中数据已经存在了,则用需要保存的数据覆盖掉已经存在的数据
SaveMode.Append(对应着字符串"append"):表示如果目标文件目录中数据已经存在了,则将数据追加到目标文件中
数据追加方式是:先将表中的所有索引删除,再追加数据
SaveMode.Ignore(对应着字符串为:“ignore”):表示如果目标文件目录中数据已经存在了,则不做任何操作
SaveMode.ErrorIfExists(对应着字符串"error"):表示如果目标文件目录中数据已经存在了,则抛异常(这个是默认的配置)
以前spark.write时总要先把原来的删了,但其实是可以设置写入模式的。
val df = spark.read.parquet(input)
df.write.mode("overwrite").parquet(output)
dataframe写入的模式一共有4种:
def mode(saveMode: String): DataFrameWriter = {
this.mode = saveMode.toLowerCase match {
case "overwrite" => SaveMode.Overwrite
case "append" => SaveMode.Append
case "ignore" => SaveMode.Ignore
case "error" | "default" => SaveMode.ErrorIfExists
case _ => throw new IllegalArgumentException(s"Unknown save mode: $saveMode. " +
"Accepted modes are 'overwrite', 'append', 'ignore', 'error'.")
}
this
}
1、SaveMode.Append
task失败重试,并不会删除上一次失败前写入的数据(文件根据分区号命名),重新执行时会继续追加数据。所以会出现数据重复。
2、SaveMode.Overwrite
task失败重试,会删除该分区上次失败所写入的数据文件,然后创建一个新的数据文件写入数据。所以不会出现数据重复。
当启动一个spark任务的时候,就会占用一个端口,默认为4040,从日志可以看到当端口被占用时,它会默认依次增加16次到4056,如果还是失败的话,就会报错退出。
解决方法:
以下代码仅供学习参考
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0modelVersion>
<groupId>org.examplegroupId>
<artifactId>SparkReadHql_TestartifactId>
<version>1.0-SNAPSHOTversion>
<properties>
<project.build.sourceEncoding>UTF-8project.build.sourceEncoding>
<project.reporting.outputEncoding>UTF-8project.reporting.outputEncoding>
<maven.compiler.encoding>UTF-8maven.compiler.encoding>
<encoding>UTF-8encoding>
<hadoop.version>3.1.3hadoop.version>
<hive.version>3.1.2hive.version>
<scala.version>2.12.11scala.version>
<spark.version>3.0.0spark.version>
properties>
<dependencies>
<dependency>
<groupId>junitgroupId>
<artifactId>junitartifactId>
<version>4.11version>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-clientartifactId>
<version>${hadoop.version}version>
<exclusions>
<exclusion>
<groupId>io.nettygroupId>
<artifactId>nettyartifactId>
exclusion>
exclusions>
dependency>
<dependency>
<groupId>io.nettygroupId>
<artifactId>netty-allartifactId>
<version>4.1.18.Finalversion>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-commonartifactId>
<version>${hadoop.version}version>
dependency>
<dependency>
<groupId>org.scala-langgroupId>
<artifactId>scala-libraryartifactId>
<version>${scala.version}version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-sql_2.12artifactId>
<version>${spark.version}version>
<exclusions>
<exclusion>
<groupId>com.google.guavagroupId>
<artifactId>guavaartifactId>
exclusion>
exclusions>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-hive_2.12artifactId>
<version>${spark.version}version>
dependency>
<dependency>
<groupId>log4jgroupId>
<artifactId>log4jartifactId>
<version>1.2.15version>
<exclusions>
<exclusion>
<groupId>javax.jmsgroupId>
<artifactId>jmsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jdmkgroupId>
<artifactId>jmxtoolsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jmxgroupId>
<artifactId>jmxriartifactId>
exclusion>
exclusions>
dependency>
dependencies>
<build>
<sourceDirectory>src/main/javasourceDirectory>
<plugins>
<plugin>
<groupId>net.alchim31.mavengroupId>
<artifactId>scala-maven-pluginartifactId>
<version>3.2.2version>
<executions>
<execution>
<goals>
<goal>compilegoal>
<goal>testCompilegoal>
goals>
<configuration>
<args>
<arg>-dependencyfilearg>
<arg>${project.build.directory}/.scala_dependenciesarg>
args>
configuration>
execution>
executions>
plugin>
plugins>
build>
project>
package com.xxxx
import org.apache.spark.sql.SparkSession
import java.io.File
import java.io.FileInputStream
import scala.io.{BufferedSource, Source}
object SparkReadHqlTest {
def main(args: Array[String]): Unit = {
val filePath: String = args(0)
val input_date: String = args(1)
val session: SparkSession = SparkSession.builder()//.master("local[2]")
.appName("SparkSeesionApp")
.enableHiveSupport() //支持hive
.getOrCreate()
// session.sparkContext.setLogLevel("WARN")
val sql: String = doFile(filePath)
val strings: Array[String] = sql.split(";")
var i = 0;
strings.foreach(sql=>{
val startTime: Long = System.currentTimeMillis()
println("==============第 "+(i+1)+" 次===sql开始=================")
println(sql)
//替换参数
// session.sql(sql.replace("'${hivevar:input_date}'", input_date)).show()
session.sql(sql).show()
val stopTime: Long = System.currentTimeMillis()
val processTime: Long = (startTime - stopTime) / 1000
println("===============第 "+(i+1)+" 次==sql结束====耗时=="+processTime+" 秒==========")
i = i+1
})
//关闭SparkSession
session.stop()
}
//读取外部sql文件文件
def doFile(fileName: String): String = {
val file: File = new File(fileName)
val stream: FileInputStream = new FileInputStream(file)
val buff: BufferedSource = Source.fromInputStream(stream,"UTF-8")
//读取拼装SQL
val sql: String = buff.getLines().mkString("\n")
sql
}
}
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0modelVersion>
<groupId>org.examplegroupId>
<artifactId>SparkReadHqlFileartifactId>
<version>1.0-SNAPSHOTversion>
<dependencies>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-clientartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-hdfsartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-commonartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-core_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-sql_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-hive_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>log4jgroupId>
<artifactId>log4jartifactId>
<version>1.2.15version>
<exclusions>
<exclusion>
<groupId>javax.jmsgroupId>
<artifactId>jmsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jdmkgroupId>
<artifactId>jmxtoolsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jmxgroupId>
<artifactId>jmxriartifactId>
exclusion>
exclusions>
dependency>
dependencies>
project>
package org.example;
import org.apache.commons.lang3.StringUtils;
import org.slf4j.Logger;
import org.apache.spark.sql.SparkSession;
import org.slf4j.LoggerFactory;
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
public class SparkReadFile {
private static final Logger logger= LoggerFactory.getLogger(SparkReadFile.class);
public static void main(String[] args) throws IOException {
// 传入参数非空判断
validateArgs(args);
// 1.创建sparkSession
SparkSession spark = SparkSession.builder().config("hive.metastore.uris", args[1]) //hive的metastore地址
.config("Spark.serializer", "org.apache.spark.serializer.KryoSerializer")
.config("hive.exec.dynamic.partition.mode", "nonstrict").enableHiveSupport().getOrCreate();
// 2.解析sql文件
BufferedReader bufferedReader =null;
String tmpStr;
String execStatus ="";
try {
bufferedReader =new BufferedReader(new FileReader(args[0])); //sql文件名
StringBuilder tempSqlContent =new StringBuilder();
while((tmpStr =bufferedReader.readLine()) !=null){
tempSqlContent.append(tmpStr+"\n");
}
// 替代sql语句中的变量${batchDate}为对应的分区信息
String[] sqlList = tempSqlContent.toString().replaceAll("\\$\\{batchDate\\}", args[2]).split(";");
for (int i = 0; i <sqlList.length-1; i++) {
logger.info("sql语句:{}",sqlList[i]);
// 3.执行SQL语句
spark.sql(sqlList[i]).show(false);
}
} catch (Exception e) {
logger.error("\n作业执行失败,{}\n"+e.getMessage(),e);
execStatus="1";
} finally {
// 4.关闭流
if (null !=bufferedReader){
bufferedReader.close();
}
if (null !=spark){
spark.close();
}
if ("1".equals(execStatus)){
System.exit(-1);
}
}
}
//参数非空判断
public static void validateArgs(String[] agrs){
if (null == agrs || args.length !=3 || StringUtils.isAnyEmpty(args)){
System.exit(-1);
}
}
}
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0modelVersion>
<groupId>org.examplegroupId>
<artifactId>MasterClusterToZHartifactId>
<version>1.0-SNAPSHOTversion>
<dependencies>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-clientartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-hdfsartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.hadoopgroupId>
<artifactId>hadoop-commonartifactId>
<version>3.1.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-core_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-sql_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>org.apache.sparkgroupId>
<artifactId>spark-hive_2.12artifactId>
<version>3.2.1version>
dependency>
<dependency>
<groupId>log4jgroupId>
<artifactId>log4jartifactId>
<version>1.2.15version>
<exclusions>
<exclusion>
<groupId>javax.jmsgroupId>
<artifactId>jmsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jdmkgroupId>
<artifactId>jmxtoolsartifactId>
exclusion>
<exclusion>
<groupId>com.sun.jmxgroupId>
<artifactId>jmxriartifactId>
exclusion>
exclusions>
dependency>
dependencies>
project>
package org.example;
import org.apache.commons.lang3.StringUtils;
import org.apache.spark.sql.Dataset;
import org.apache.spark.sql.Row;
import org.apache.spark.sql.SaveMode;
import org.apache.spark.sql.SparkSession;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
public class ReadHiveToTenant {
private static final Logger logger= LoggerFactory.getLogger(ReadHiveToTenant.class);
public static void main(String[] args) {
// 传入参数非空判断
validateArgs(args);
// 1.初始化SparkSession对象,主集群Metastore
SparkSession spark = createSparkSession(args[0]);
// 2.读取主集群hive表分区数据
Dataset<Row> partitonDF = spark.read().table(args[1]).where(args[2]);
long count =partitonDF.count();
logger.info(args[1]+"表的数据量:-----:"+count);
//3.写入租户hive表的HDFS路径
partitonDF.write().mode(SaveMode.Overwrite).save(args[3]);
//4.关闭资源
if (null !=spark){
spark.close();
}
//5.修复hive表
SparkSession tenantSparkSession = createSparkSession(args[4]) ;
tenantSparkSession.sql("MSCK REPAIR TABLE"+agrs[5]);
if (null !=tenantSparkSession){
tenantSparkSession.close();
}
}
// 获取一个SparkSession对象
public static SparkSession createSparkSession(String hiveMetastore){
SparkSession sparkSession = SparkSession.builder().config("hive.metastore.uris", hiveMetastore)
.config("hive.exec.dynamic.partition", true) //开启动态分区
.config("hive.exec.dynamic.partition.mode", "nonstrict").enableHiveSupport().getOrCreate();
return sparkSession;
}
//参数非空判断
public static void validateArgs(String[] agrs){
if (null == agrs || args.length !=6 || StringUtils.isAnyEmpty(args)){
System.exit(-1);
}
}
}
以上代码示例仅供学习参考,方便收藏,对代码进行整理汇总
参考博客: