上一遍文章中其实最主要的就是JAVA环境变量以及hadoop环境变量的设置，这两个设置好了的话，运行hadoop基本上不会出问题。
在hadoop的基础上安装spark好简单。
安装Spark之前需要先安装Hadoop集群，因为之前已经安装了hadoop，所以我直接在之前的hadoop集群上安装spark。
硬件环境：
hddcluster1 10.0.0.197 redhat7
hddcluster2 10.0.0.228 centos7 这台作为master
hddcluster3 10.0.0.202 redhat7
hddcluster4 10.0.0.181 centos7
软件环境：
scala-2.11.7
spark-2.0.2-bin-hadoop2.7.tgz
#所有操作用hadoop
基本流程：
1、master解压scala-2.11.7和spark-2.0.2-bin-hadoop2.7.tgz到相应的目录
2、配置scala和spark环境变量
3、修改配置文件
4、拷贝scala和spark到各个节点，授权
5、启动spark集群
|
1
2
3
4
5
6
7
8
9
10
11
|
#hadoop用户下操作，下载scala,安装wget http://downloads.lightbend.com/scala/2.11.7/scala-2.11.7.tgz
tar -zxvf scala-2.11.7.tgz
mv scala-2.11.7 /usr/local/scala
sudo mv scala-2.11.7 /usr/local/scala
vim .bashrc #添加
export SCALA_HOME=/usr/local/scala
export PATH=$PATH:$HADOOP_HOME/sbin:$HADOOP_HOME/bin:$SCALA_HOME/bin
source .bashrc
[hadoop@hddcluster2 ~]$ scala -versionScala code runner version 2.11.7 -- Copyright 2002-2013, LAMP/EPFL
|
|
1
2
3
4
5
6
7
|
#在官网下载spark-2.0.2-bin-hadoop2.7.tgztar -zxvf spark-2.0.2-bin-hadoop2.7.tgz
mv spark-2.0.2-bin-hadoop2.7 spark
sudo mv spark /usr/local/
vim .bashrc #添加
export SPARK_HOME=/usr/local/spark
export PATH=$PATH:$HADOOP_HOME/sbin:$HADOOP_HOME/bin:$SCALA_HOME/bin:$SPARK_HOME
|
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
|
#修改spark配置文件cd /usr/local/spark/conf
cp spark-env.sh.template spark-env.sh
vi spark-env.sh #添加下面内容
###jdk direxport JAVA_HOME=/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.111-2.b15.el7_3.x86_64
###scala direxport SCALA_HOME=/usr/local/scala
###the ip of master node of sparkexport SPARK_MASTER_IP=10.0.0.228
###the max memory size of workerexport SPARK_WORKER_MEMORY=8G
###hadoop configuration file direxport HADOOP_CONF_DIR=/usr/local/hadoop/etc/hadoop/
#修改slavecp slaves.template slaves
vi slaves #把localhost改为下面内容
hddcluster1hddcluster2hddcluster3hddcluster4 |
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
|
#把/usr/local/spark 和 /usr/local/scala打包，然后复制到slave节点、cd /usr/local
tar -zcf ~/master.spark.tar.gz ./spark
tar -zcf ~/master.scala.tar.gz ./scala
scp master.spark.tar.gz hddcluster1:~
scp master.scala.tar.gz hddcluster1:~
#登录各个节点进行解压到/usr/localtar -zxf master.spark.tar.gz -C /usr/local/
tar -zxf master.scala.tar.gz -C /usr/local/
chown -R hadoop:hadoop /usr/local/spark
chown -R hadoop:hadoop /usr/local/scala
再配置.bashrc环境变量和master的一样。加上hadoop上一篇的.bashrc内容是这样子：#scalaexport SCALA_HOME=/usr/local/scala
#sparkexport SPARK_HOME=/usr/local/spark
#java and hadoopexport JAVA_HOME=/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.111-2.b15.el7_3.x86_64
export HADOOP_HOME=/usr/local/hadoop
export HADOOP_INSTALL=$HADOOP_HOME
export HADOOP_MAPRED_HOME=$HADOOP_HOME
export HADOOP_COMMON_HOME=$HADOOP_HOME
export HADOOP_HDFS_HOME=$HADOOP_HOME
export YARN_HOME=$HADOOP_HOME
export HADOOP_COMMON_LIB_NATIVE_DIR=$HADOOP_HOME/lib/native
export PATH=$PATH:$HADOOP_HOME/sbin:$HADOOP_HOME/bin:$SCALA_HOME/bin:$SPARK_HOME
export HADOOP_PREFIX=$HADOOP_HOME
export HADOOP_OPTS="-Djava.library.path=$HADOOP_PREFIX/lib:$HADOOP_PREFIX/lib/native"
到此Spark集群搭建完毕 |
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
|
.启动Spark集群：启动Spark之前需要先将hadoop的dfs以及yarn启动。/usr/local/spark/sbin/start-all.sh
启动所有服务之后，在命令行输入jps:
[hadoop@hddcluster2 ~]$ jps29601 ResourceManager32098 SparkSubmit29188 DataNode29364 SecondaryNameNode29062 NameNode29915 NodeManager30251 Master30380 Worker30062 JobHistoryServer18767 Jps 比hadoop集群启动时多了Master和worker
/usr/local/spark/bin/spark-shell.sh
出现scala>时说明成功。在浏览器中输入10.0.0.228:8080时，会看到如下图，有4个Worker |
本文转自 yanconggod 51CTO博客，原文链接:http://blog.51cto.com/yanconggod/1885082