【phoenix-开发】phoenix-spark的简单使用
2019-02-21 本文已影响0人
粮忆雨
参考官网：http://phoenix.apache.org/phoenix_spark.html
测试环境：CDH5.14.2集群集成phoenix4.14.0-cdh5.14.2 和spark2.4.0-cdh5.13.2
一、简单认识Phoenix
1、架设在hbase上的newsql层，具有标准sql和jdbc功能,支持全ACID事务
2、适合简单条件查询的OLTP场景，快速响应。能够更简单的构建二级索引。
3、除了oltp的场景之外，我们可能还需要对数据做一些理想的处理或者做复杂的聚合分析查询（phoenix并不适用于这种场景），我们需要借助spark来做计算。
我们都知道spark可以借助jdbc的方式去连接phoenix，但它只能通过在数字列上进行分块来并行化查询。它还需要一个已知的下界、上界和分区计数，以便创建拆分查询。而phoenix-spark集成能够利用Phoenix提供的底层分割，以便在多个工作者之间检索和保存数据。所需的只是一个数据库URL和一个表名。可以给出可选的SELECT列，以及用于有效过滤的下推谓词。
二、 IDEA下开发
1、pom文件：
不使用对应cdh集群本版的依赖是有原因的，因为phoenix-spark4.14.0-cdh5.14.2的依赖导致各种问题更多。
<?xml version="1.0" encoding="UTF-8"?>
<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>4.0.0</modelVersion>
<groupId>com.hdcloud.data.spark</groupId>
<artifactId>trading-nodeprice-handle</artifactId>
<version>1.0-SNAPSHOT</version>
<properties>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<spark.version>2.4.0</spark.version>
<scala.version>2.11</scala.version>
<scala.compat.version>2.11.8</scala.compat.version>
<phoenix.version>4.14.1</phoenix.version>
</properties>
<dependencies>
<dependency>
<groupId>mysql</groupId>
<artifactId>mysql-connector-java</artifactId>
<version>8.0.19</version>
<exclusions>
<exclusion>
<groupId>com.google.protobuf</groupId>
<artifactId>protobuf-java</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>com.lmax</groupId>
<artifactId>disruptor</artifactId>
<version>3.3.11</version>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-core_${scala.version}</artifactId>
<version>${spark.version}</version>
<exclusions>
<exclusion>
<groupId>com.lmax</groupId>
<artifactId>disruptor</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>org.apache.spark</groupId>
<artifactId>spark-sql_${scala.version}</artifactId>
<version>${spark.version}</version>
<exclusions>
<exclusion>
<groupId>com.lmax</groupId>
<artifactId>disruptor</artifactId>
</exclusion>
</exclusions>
</dependency>
<dependency>
<groupId>org.apache.phoenix</groupId>
<artifactId>phoenix-spark</artifactId>
<version>4.14.1-HBase-1.2</version>
<exclusions>
<exclusion>
<groupId>com.lmax</groupId>
<artifactId>disruptor</artifactId>
</exclusion>
</exclusions>
<!--<scope>provided</scope>-->
</dependency>
<dependency>
<groupId>org.apache.phoenix</groupId>
<artifactId>phoenix-core</artifactId>
<version>4.14.1-HBase-1.2</version>
<exclusions>
<exclusion>
<groupId>com.lmax</groupId>
<artifactId>disruptor</artifactId>
</exclusion>
</exclusions>
<!--<scope>provided</scope>-->
</dependency>
</dependencies>
<repositories>
<repository>
<id>cloudera</id>
<url>https://repository.cloudera.com/artifactory/cloudera-repos/</url>
</repository>
</repositories>
<build>
<resources>
<resource>
<directory>src/main/resources</directory>
<filtering>false</filtering>
</resource>
</resources>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<version>3.1</version>
<configuration>
<source>1.8</source>
<target>1.8</target>
<encoding>${project.build.sourceEncoding}</encoding>
</configuration>
</plugin>
<plugin>
<groupId>org.scala-tools</groupId>
<artifactId>maven-scala-plugin</artifactId>
<version>2.15.2</version>
<executions>
<execution>
<goals>
<goal>compile</goal>
<goal>testCompile</goal>
</goals>
</execution>
</executions>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-shade-plugin</artifactId>
<version>2.4.3</version>
<executions>
<execution>
<phase>package</phase>
<goals>
<goal>shade</goal>
</goals>
<configuration>
<filters>
<filter>
<artifact>*:*</artifact>
<excludes>
<exclude>META-INF/*.SF</exclude>
<exclude>META-INF/*.DSA</exclude>
<exclude>META-INF/*.RSA</exclude>
</excludes>
</filter>
</filters>
</configuration>
</execution>
</executions>
</plugin>
</plugins>
</build>
</project>
以下测试功能主要是对一个phoenix的事实表进行维表管理的处理过程。需要注意的是如果开启可phoenix和hbase的表空间映射，需要在resource目录下的hbase-site.xml配置 phoenix.schema.isNamespaceMappingEnabled=true和phoenix.schema.mapSystemTablesToNamespace=true
package com.hdcloud.data.spark
import org.apache.hadoop.conf.Configuration
import org.apache.spark.sql.functions.to_timestamp
import org.apache.spark.sql.types.DataTypes
import org.apache.spark.sql.{SaveMode, SparkSession}
import org.apache.phoenix.spark._
/**
*
* @date 2020/5/11 17:39
* @author liangriyu
*/
object NodePriceHandle {
val MYSQL_TB_DIM_CITY = "dim_node_city"
val MYSQL_TB_DIM_REGION = "dim_node_region"
val PHOENIX_TB_NODE_PRICE_ORG = "\"data_trading_assistant\".\"bus_node_price\""
val PHOENIX_TB_NODE_PRICE_EX = "\"data_trading_assistant\".\"bus_node_price_ex\""
val PHOENIX_ZOOKEEPER = "bigdata-dev01,bigdata-dev02,bigdata-dev03:2181"
def main(args: Array[String]): Unit = {
val spark = SparkSession.builder()
// .master("local[1]")
.appName("spark-phoenix-nodeprice")
.config("spark.some.config.option", "some-value")
.getOrCreate()
//获取维表数据（城市、大区）
val mysqlReader = spark.read.format("jdbc")
mysqlReader.option("url", "jdbc:mysql://192.168.1.200:6033/data_trading_assistant?useUnicode=true&characterEncoding=utf8&serverTimezone=UTC")
mysqlReader.option("driver", "com.mysql.cj.jdbc.Driver")
mysqlReader.option("user", "root")
mysqlReader.option("password", "root")
mysqlReader.option("dbtable", MYSQL_TB_DIM_CITY)
val cityDF = mysqlReader.load()
// cityDF.createOrReplaceTempView("dim_city")
mysqlReader.option("dbtable", MYSQL_TB_DIM_REGION)
val regionDF = mysqlReader.load()
// regionDF.createOrReplaceTempView("dim_region")
mysqlReader.option("dbtable", "(select id,node_name,city_name,area_name from bus_node_info) as bus_node_info")
val nodeInfoDF = mysqlReader.load()
// nodeInfoDF.show(10)
//广播维表
val broadcastCityDF = spark.sparkContext.broadcast(cityDF)
val broadcastRegionDF = spark.sparkContext.broadcast(regionDF)
//读取原始节点电价事实表 jdbc/phoenix-spark/phoenixTableAsDataFrame
val configuration = new Configuration()
// configuration.set("hbase.zookeeper.quorum", PHOENIX_ZOOKEEPER)
configuration.addResource("hbase-site.xml")
val nodePriceDF = spark.sqlContext
.phoenixTableAsDataFrame(PHOENIX_TB_NODE_PRICE_ORG
, Array("out_type", "node_name", "price_date","city_name","area_name","price","prov_name","node_type"
,"voltage","create_time"), conf = configuration)
// val nodePriceDF = spark.read.format("org.apache.phoenix.spark")
// .option("zkUrl", PHOENIX_ZOOKEEPER)
// .option("table", PHOENIX_TB_NODE_PRICE_ORG)
// .load()
//扩展节点信息
val joinedCityDF = nodeInfoDF.join(broadcastCityDF.value, nodeInfoDF("city_name") === cityDF("city_name"), "left_outer")
.select(nodeInfoDF("id").alias("node_id"), nodeInfoDF("node_name"), cityDF("id").alias("city_id"), cityDF("prov_id"), nodeInfoDF("area_name"))
val joinedRegionDF = joinedCityDF.join(broadcastRegionDF.value, joinedCityDF("area_name") === regionDF("region_name"), "left_outer")
.select(joinedCityDF("node_id"), joinedCityDF("node_name"), joinedCityDF("city_id"), joinedCityDF("prov_id"), regionDF("id").as("area_id"))
//扩展节点价格
val broadcastNodeDF = spark.sparkContext.broadcast(joinedRegionDF)
val joinedNodeInfoDF = nodePriceDF.join(broadcastNodeDF.value, nodePriceDF("node_name") === joinedRegionDF("node_name"), "left_outer")
.select(nodePriceDF("out_type").alias("\"out_type\""), nodePriceDF("price_date").alias("\"price_date\""),
nodePriceDF("node_name").alias("\"node_name\""), nodePriceDF("city_name").alias("\"city_name\""),
nodePriceDF("area_name").alias("\"area_name\""), nodePriceDF("price").alias("\"price\""),
nodePriceDF("prov_name").alias("\"prov_name\""), nodePriceDF("node_type").alias("\"node_type\""),
nodePriceDF("voltage").alias("\"voltage\""), nodePriceDF("create_time").alias("\"create_time\""),
joinedRegionDF("city_id").alias("\"city_id\""), joinedRegionDF("prov_id").alias("\"prov_id\""),
joinedRegionDF("area_id").alias("\"area_id\""),joinedRegionDF("node_id").alias("\"node_id\""))
//添加列和修改列类型
val addedColumnDF=joinedNodeInfoDF
.withColumn("\"price_timestamp\"", to_timestamp(joinedNodeInfoDF("\"price_date\"")).cast(DataTypes.LongType))
.withColumn("\"out_type\"",joinedNodeInfoDF("\"out_type\"").cast(DataTypes.IntegerType))
.withColumn("\"voltage\"",joinedNodeInfoDF("\"voltage\"").cast(DataTypes.IntegerType))
.withColumn("\"node_type\"",joinedNodeInfoDF("\"node_type\"").cast(DataTypes.IntegerType))
// addedColumnDF.show(10)
// addedColumnDF.write.format("org.apache.phoenix.spark")
// .option("zkUrl", PHOENIX_ZOOKEEPER)
// .option("table", PHOENIX_TB_NODE_PRICE_EX)
// .mode(SaveMode.Overwrite)
// .save()
//
addedColumnDF.show(10)
// addedColumnDF.saveToPhoenix(Map("table" -> PHOENIX_TB_NODE_PRICE_EX, "zkUrl" -> PHOENIX_ZOOKEEPER))
addedColumnDF.saveToPhoenix(PHOENIX_TB_NODE_PRICE_EX,conf = configuration)
spark.stop()
}
}
另外，在服务器上提交job运行时，发现saveToPhoenix方法的configuration参数并没有加载到hbase-site.xml的phoenix.schema.isNamespaceMappingEnabled=true和phoenix.schema.mapSystemTablesToNamespace=true等参数
使用--files /etc/hbase/conf/hbase-site.xml（推荐）或者手动重写org.apache.phoenix.spark.DataFrameFunctions类，手动设置了上述参数（临时方案）
/*
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
package org.apache.phoenix.spark
import org.apache.hadoop.conf.Configuration
import org.apache.hadoop.io.NullWritable
import org.apache.phoenix.mapreduce.PhoenixOutputFormat
import org.apache.phoenix.mapreduce.util.PhoenixConfigurationUtil
import org.apache.phoenix.util.SchemaUtil
import org.apache.spark.sql.DataFrame
import scala.collection.JavaConversions._
class DataFrameFunctions(data: DataFrame) extends Serializable {
def saveToPhoenix(parameters: Map[String, String]): Unit = {
saveToPhoenix(parameters("table"), zkUrl = parameters.get("zkUrl"), tenantId = parameters.get("TenantId"),
skipNormalizingIdentifier=parameters.contains("skipNormalizingIdentifier"))
}
def saveToPhoenix(tableName: String, conf: Configuration = new Configuration,
zkUrl: Option[String] = None, tenantId: Option[String] = None, skipNormalizingIdentifier: Boolean = false): Unit = {
// Retrieve the schema field names and normalize to Phoenix, need to do this outside of mapPartitions
val fieldArray = getFieldArray(skipNormalizingIdentifier, data)
// Create a configuration object to use for saving
conf.set("phoenix.schema.isNamespaceMappingEnabled", "true")
conf.set("phoenix.schema.mapSystemTablesToNamespace", "true")
@transient val outConfig = ConfigurationUtil.getOutputConfiguration(tableName, fieldArray, zkUrl, tenantId, Some(conf))
// Retrieve the zookeeper URL
val zkUrlFinal = ConfigurationUtil.getZookeeperURL(outConfig)
// Map the row objects into PhoenixRecordWritable
val phxRDD = data.rdd.mapPartitions{ rows =>
// Create a within-partition config to retrieve the ColumnInfo list
@transient val partitionConfig = ConfigurationUtil.getOutputConfiguration(tableName, fieldArray, zkUrlFinal, tenantId)
@transient val columns = PhoenixConfigurationUtil.getUpsertColumnMetadataList(partitionConfig).toList
rows.map { row =>
val rec = new PhoenixRecordWritable(columns)
row.toSeq.foreach { e => rec.add(e) }
(null, rec)
}
}
// Save it
phxRDD.saveAsNewAPIHadoopFile(
Option(
conf.get("mapreduce.output.fileoutputformat.outputdir")
).getOrElse(
Option(conf.get("mapred.output.dir")).getOrElse("")
),
classOf[NullWritable],
classOf[PhoenixRecordWritable],
classOf[PhoenixOutputFormat[PhoenixRecordWritable]],
outConfig
)
}
def getFieldArray(skipNormalizingIdentifier: Boolean = false, data: DataFrame) = {
if (skipNormalizingIdentifier) {
data.schema.fieldNames.map(x => x)
} else {
data.schema.fieldNames.map(x => SchemaUtil.normalizeIdentifier(x))
}
}
}
三、IDEA下打jar包的几种方式
方式一：
灵活，可以手动过滤一些依赖，瘦身jar
step1:
step2:
step3:
删除依赖jar包
方式二：
直接使用maven插件，方便快捷
四、 提交服务器运行的一些问题
为了避免job启动时和habse集群的jar有版本冲突和集群jar版本低问题（如：NoSuchMethodError: org.apache.hadoop.hbase.client.HBaseAdmin.<init>(Lorg/apache/hadoop/hbase/client/HConnection;)V等），
在spark-evn.sh中配置预先加载以下jar依赖。因为jvm中存在对应的jar就不会覆盖加载。
export SPARK_DIST_CLASSPATH=/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/*:$SPARK_DIST_CLASSPATH
export SPARK_DIST_CLASSPATH=$SPARK_DIST_CLASSPATH:/opt/cloudera/parcels/CDH/lib/hbase/lib/*
CDH中在
spark2-conf/spark-env.sh 的 Spark 2 服务高级配置代码段（安全阀）中配置
如：
或者在spark job指定--conf spark.executor.extraClassPath=/opt/cloudera/parcels/CDH/lib/hive/lib/hive-hbase-handler-1.1.0-cdh5.14.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-spark-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-core-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-queryserver-client-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-api-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-core-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-hbase-compat-1.1-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-api-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-common-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-core-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-discovery-api-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-discovery-core-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-zookeeper-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/disruptor-3.3.11.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-annotations-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-client-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-common-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-hadoop2-compat-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-hadoop-compat-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-prefix-tree-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-procedure-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-protocol-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-server-1.2.5.jar --conf spark.driver.extraClassPath=/opt/cloudera/parcels/CDH/lib/hive/lib/hive-hbase-handler-1.1.0-cdh5.14.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-spark-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-core-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/phoenix-queryserver-client-4.14.1-HBase-1.2.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-api-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-core-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/tephra-hbase-compat-1.1-0.14.0-incubating.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-api-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-common-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-core-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-discovery-api-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-discovery-core-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/twill-zookeeper-0.8.0.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/disruptor-3.3.11.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-annotations-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-client-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-common-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-hadoop2-compat-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-hadoop-compat-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-prefix-tree-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-procedure-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-protocol-1.2.5.jar:/opt/cloudera/parcels/SPARK2/lib/spark2/ex-jars/hbase/hbase-server-1.2.5.jar --conf spark.dynamicAllocation.enabled=true --conf spark.shuffle.service.enabled=true --conf spark.dynamicAllocation.initialExecutors=5 --conf spark.dynamicAllocation.minExecutors=2 --conf spark.dynamicAllocation.maxExecutors=12 --conf spark.dynamicAllocation.executorIdleTimeout=120s --conf spark.dynamicAllocation.schedulerBacklogTimeout=10s --files /etc/hbase/conf/hbase-site.xml
job参数
spark2-submit --master yarn \
--name trading-nodeprice-handle \
--deploy-mode cluster \
--driver-memory 4G \
--num-executors 5 \
--executor-cores 5 \
--executor-memory 4G \
--class com.hdcloud.data.spark.NodePriceHandle \
hdfs:///workspace/spark/trading/trading-nodeprice-handle-1.0-SNAPSHOT.jar \
--conf spark.executor.userClassPathFirst=true \
--conf spark.driver.userClassPathFirst=true \
--files /opt/cloudera/parcels/CDH/lib/hbase/conf/hbase-site.xml
其他问题记录（非本次测试问题）
1、在spark-default.conf配置spark.executor.extraClassPath和spark.driver.extraClassPath指向包含phoenix-<version>-client.jar的目录。否则缺少jar报如下错误：
Caused by: java.lang.ClassNotFoundException: Class org.apache.phoenix.spark.PhoenixRecordWritable not found
spark.executor.extraClassPath /opt/apps/spark/external_jars/*
spark.driver.extraClassPath /opt/apps/spark/external_jars/*
2、在hadoop安装hadoop-env.sh添加export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:/opt/apps/hbase-1.1.1/lib/*，避免缺少hbase jar包如下错误：
Caused by: java.lang.ClassNotFoundException: org.apache.hadoop.hbase.HBaseConfiguration
