spark 2.2.0 高可用搭建

xinhu1300

　　一、概述
　　1.实验环境基于以前搭建的haoop HA；
　　2.spark HA所需要的zookeeper环境前文已经配置过，此处不再重复。
　　3.所需软件包为：scala-2.12.3.tgz、spark-2.2.0-bin-hadoop2.7.tar
　　4.主机规划

bd1 　　bd2 　　bd3	Worker
bd4 　　bd5	Master、Worker

　　二、配置Scala
　　1.解压并拷贝
[root@bd1 ~]# tar -zxf scala-2.12.3.tgz
[root@bd1 ~]# cp -r scala-2.12.3 /usr/local/　　2.配置环境变量
[root@bd1 ~]# vim /etc/profile
export SCALA_HOME=/usr/local/scala
export PATH=:$SCALA_HOME/bin:$PATH
[root@bd1 ~]# source /etc/profile　　3.验证
[root@bd1 ~]# scala -version
Scala code runner version 2.12.3 -- Copyright 2002-2017, LAMP/EPFL and Lightbend, Inc.　　三、配置Spark
　　1.解压并拷贝
[root@bd1 ~]# tar -zxf spark-2.2.0-bin-hadoop2.7.tgz
[root@bd1 ~]# cp spark-2.2.0-bin-hadoop2.7 /usr/local/spark2.配置环境变量
[root@bd1 ~]# vim /etc/profile
export SCALA_HOME=/usr/local/scala
export PATH=:$SCALA_HOME/bin:$PATH
[root@bd1 ~]# source /etc/profile　　3.修改spark-env.sh    #文件不存在需要拷贝模板
[root@bd1 conf]# vim spark-env.sh
export JAVA_HOME=/usr/local/jdk
export HADOOP_HOME=/usr/local/hadoop
export HADOOP_CONF_DIR=/usr/local/hadoop/etc/hadoop
export SCALA_HOME=/usr/local/scala
export SPARK_DAEMON_JAVA_OPTS="-Dspark.deploy.recoveryMode=ZOOKEEPER -Dspark.deploy.zookeeper.url=bd4:2181,bd5:2181 -Dspark.deploy.zookeeper.dir=/spark"
export SPARK_WORKER_MEMORY=1g
export SPARK_WORKER_CORES=2
export SPARK_WORKER_INSTANCES=1　　4.修改spark-defaults.conf    #文件不存在需要拷贝模板
[root@bd1 conf]# vim spark-defaults.conf
spark.master                     spark://master:7077
spark.eventLog.enabled           true
spark.eventLog.dir               hdfs://master:/user/spark/history
spark.serializer                 org.apache.spark.serializer.KryoSerializer　　5.在HDFS文件系统中新建日志文件目录

hdfs dfs -mkdir -p /user/spark/history
hdfs dfs -chmod 777 /user/spark/history　　6.修改slaves
[root@bd1 conf]# vim slaves
bd1
bd2
bd3
bd4
bd5　　四、同步到其他主机

　　1.使用scp同步Scala到bd2-bd5
scp -r /usr/local/scala root@bd2:/usr/local/
scp -r /usr/local/scala root@bd3:/usr/local/
scp -r /usr/local/scala root@bd4:/usr/local/
scp -r /usr/local/scala root@bd5:/usr/local/　　2.同步Spark到bd2-bd5
scp -r /usr/local/spark root@bd2:/usr/local/
scp -r /usr/local/spark root@bd3:/usr/local/
scp -r /usr/local/spark root@bd4:/usr/local/
scp -r /usr/local/spark root@bd5:/usr/local/　　五、启动集群并测试HA

　　1.启动顺序为：zookeeper-->hadoop-->spark
　　2.启动spark
　　bd4：
[root@bd4 sbin]# cd /usr/local/spark/sbin/
[root@bd4 sbin]# ./start-all.sh
starting org.apache.spark.deploy.master.Master, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.master.Master-1-bd4.out
bd4: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-bd4.out
bd2: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-bd2.out
bd3: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-bd3.out
bd5: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-bd5.out
bd1: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-bd1.out
[root@bd4 sbin]# jps
3153 DataNode
7235 Jps
3046 JournalNode
7017 Master
3290 NodeManager
7116 Worker
2958 QuorumPeerMain　　bd5：

[root@bd5 sbin]# ./start-master.sh
starting org.apache.spark.deploy.master.Master, logging to /usr/local/spark/logs/spark-root-org.apache.spark.deploy.master.Master-1-bd5.out
[root@bd5 sbin]# jps
3584 NodeManager
5602 RunJar
3251 QuorumPeerMain
8564 Master
3447 DataNode
8649 Jps
8474 Worker
3340 JournalNode

　　3.停掉bd4的Master进程
[root@bd4 sbin]# kill -9 7017
[root@bd4 sbin]# jps
3153 DataNode
7282 Jps
3046 JournalNode
3290 NodeManager
7116 Worker
2958 QuorumPeerMain

　　五、总结
　　一开始时想把Master放到bd1和bd2上，但是启动Spark后发现两个节点上都是Standby。然后修改配置文件转移到bd4和bd5上，才顺利运行。换言之Spark HA的Master必须位于Zookeeper集群上才能正常运行，即该节点上要有JournalNode这个进程。