大数据平台搭建-spark集群安装

本系列文章主要阐述大数据计算平台相关框架的搭建，包括如下内容：

• 基础环境安装
• zookeeper集群的搭建
• kafka集群的搭建
• hadoop/hbase集群的搭建
• spark集群的搭建
• flink集群的搭建
• elasticsearch集群的搭建
• alluxio集群的搭建

版本要求

java

版本：1.8.*(1.8.0_60)
下载地址：http://www.oracle.com/technetwork/java/javase/downloads/jdk8-downloads-2133151.html

scala

版本：2.11.*(2.11.8)
下载地址：http://www.scala-lang.org/download/2.11.8.html

zookeeper

版本:3.4.*(zookeeper-3.4.8)
下载地址：http://mirrors.hust.edu.cn/apache/zookeeper/zookeeper-3.4.8/

spark

版本:2.0.*(spark-2.1.0-bin-hadoop2.7)
下载地址：http://spark.apache.org/downloads.html

spark安装

前置条件

java安装

见链接http://www.cnblogs.com/aidodoo/p/7007917.html

scala安装

见链接http://www.cnblogs.com/aidodoo/p/7007917.html

zookeeper安装

见链接http://www.cnblogs.com/aidodoo/p/7048867.html

压缩解压

tar -zxvf spark-2.1.0-bin-hadoop2.7.tgz
mv spark-2.1.0-bin-hadoop2.7 spark

环境变量

vi ~/.bash_profile   
export JAVA_HOME=/wls/oracle/jdk      
export SCALA_HOME=/wls/oracle/scala   
export ZOOKEEPER_HOME=/wls/oracle/zookeeper   
export HADOOP_HOME=/wls/oracle/hadoop   
export HBASE_HOME=/wls/oracle/hbase   
export SPARK_HOME=/wls/oracle/spark   

export PATH=$PATH:$HOME/bin:$JAVA_HOME/bin:$SCALA_HOME/bin:$SPARK_HOME/bin:$ZOOKEEPER_HOME/bin:$HADOOP_HOME/bin:$HBASE_HOME/bin   
export CLASSPATH=.:$JAVA_HOME/jre/lib/rt.jar:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tool.jar   

export JAVA_HOME  SCALA_HOME  ZOOKEEPER_HOME HADOOP_HOME SPARK_HOME   

配置更改

spark集群主要涉及到如下配置文件的变更

spark-default.conf
spark-env.sh
slaves

spark-default.conf

spark.master            spark://SZB-L0045546:7077
spark.executor.memory   4g
spark.eventLog.enabled  true
spark.serializer        org.apache.spark.serializer.KryoSerializer
spark.cores.max			10
spark.scheduler.mode	FIFO
spark.shuffle.compress	true
spark.ui.port			4040
spark.eventLog.enabled	true
spark.eventLog.dir 		/wls/oracle/bigdata/spark/sparkeventlog
spark.kryoserializer.buffer 512m
spark.rpc.numRetries	5
spark.port.maxRetries	16
spark.rpc.askTimeout    120s
spark.network.timeout	120s
spark.rpc.lookupTimeout  120s

spark.executor.extraJavaOptions -XX:+UseG1GC -XX:InitiatingHeapOccupancyPercent=40 -XX:G1ReservePercent=10 -XX:G1HeapRegionSize=8M -XX:MaxPermSize=300M -XX:+PrintReferenceGC -XX:+PrintAdaptiveSizePolicy -XX:+UnlockDiagnosticVMOptions -XX:+G1SummarizeConcMark -verbose:gc -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintGCDateStamps

spark-env.sh

export JAVA_HOME=/wls/oracle/jdk
export SCALA_HOME=/wls/oracle/scala
export SPARK_HOME=/wls/oracle/spark
export HADOOP_HOME=/wls/oracle/hadoop
export PATH=$PATH:$JAVA_HOME/bin:$SCALA_HOME/bin:$SPARK_HOME/bin
export SPARK_WORKER_MEMORY=28g
export SPARK_WORKER_INSTANCES=1
export SPARK_WORKER_CORES=8
export SPARK_PID_DIR=/wls/oracle/bigdata/spark/sparkpids
export SPARK_LOCAL_DIRS=/wls/oracle/bigdata/spark/sparkdata
export SPARK_WORKER_DIR=/wls/oracle/bigdata/spark/sparkwork
export SPARK_WORKER_OPTS="-Dspark.worker.cleanup.enabled=true -Dspark.worker.cleanup.interval=300"
export SPARK_DAEMON_JAVA_OPTS="-Dspark.deploy.recoveryMode=ZOOKEEPER -Dspark.deploy.zookeeper.url=SZB-L0045546,SZB-L0045551,SZB-L0045552 -Dspark.deploy.zookeeper.dir=/spark"
export SPARK_LIBARY_PATH=.:$JAVA_HOME/lib:$JAVA_HOME/jre/lib:$HADOOP_HOME/lib/native
export SPARK_CLASSPATH="$SPARK_CLASSPATH:$SPARK_HOME/lib/*:$SPARK_CLASSPATH
:SCALA_HOME/lib/*:$SPARK_CLASSPATH:$SPARK_HOME/jars/*"

其中$SPARK_HOME/jars/为spark系统自带的jar包,而$SPARK_HOME/lib/为我们自定义的外部jar包，如kafka,mongo等应用相关jar包

slaves

列出所有worker节点的主机名

SZB-L0045551
SZB-L0045552
SZB-L0047815
SZB-L0047816

运维命令

启动集群
/wls/oracle/spark/sbin/start_all.sh
停止集群
/wls/oracle/spark/sbin/start_all.sh

---

关于作者
爱编程、爱钻研、爱分享、爱生活
关注分布式、高并发、数据挖掘
如需捐赠，请扫码