[TOC]
1 说明
1.1 效果图
实现1个Master两个Worker的伪集群环境搭建。
Cluster1.2 实验环境
item | desc |
---|---|
OS | CentOS release 6.7 (Final) |
java | jdk-8u101-linux-x64.tar.gz |
scala | scala-2.11.8.tgz |
hadoop | hadoop-2.6.5.tar.gz |
spark | spark-2.0.1-bin-hadoop2.6.tgz |
确保三台主机上的域名解析配置如下:
[root@h1 /]# cat /etc/hosts
127.0.0.1 localhost.localdomain localhost
192.168.161.128 h1
192.168.161.129 h2
192.168.161.130 h3
::1 localhost6.localdomain6 localhost6
1.3 配置ssh免登陆
配置好h1到h2,h3的ssh免登陆。
至于如何配置ssh免登陆可以参考本人的另一篇文章:ssh/OpenSSH
2 集群搭建
2.1 配置环境变量
这个没啥好说的了。以下是本人的配置:
# java
export JAVA_HOME=/soft/jdk1.8.0_101
export CLASSPATH=.:$JAVA_HOME/lib/tools.jar:$JAVA_HOME/lib/dt.jar
export PATH=$PATH:$JAVA_HOME/bin
# scala
export SCALA_HOME=/soft/scala-2.11.8
export PATH=$PATH:$SCALA_HOME/bin
2.2 配置hadoop
# 下载
wget http://archive.apache.org/dist/hadoop/common/hadoop-2.6.5/hadoop-2.6.5.tar.gz
# 解压
tar -zxvf hadoop-2.6.5.tar.gz
# 可选(本人喜欢将软件统一安装的根目录的soft目录下)
mv hadoop-2.6.5 /soft/
2.2.1 core-site.xml
vim /soft/hadoop-2.6.5/etc/hadoop/core-site.xml
内容如下:
<configuration>
<property>
<name>fs.defaultFS</name>
<value>hdfs://h1:9000</value>
</property>
<property>
<name>hadoop.tmp.dir</name>
<value>/soft/hadoop-2.6.5/tmp</value>
</property>
<property>
<name>hadoop.native.lib</name>
<value>true</value>
</property>
</configuration>
2.2.2 hdfs-site.xm
vim /soft/hadoop-2.6.5/etc/hadoop/hdfs-site.xml
内容如下:
<configuration>
<property>
<name>sdf.replication</name>
<value>2</value>
</property>
<property>
<name>dfs.namenode.secondary.http-address</name>
<value>h1:50090</value>
</property>
<property>
<name>dfs.namenode.name.dir</name>
<value>/soft/hadoop-2.6.5/dfs/data</value>
</property>
<property>
<name>dfs.namenode.checkpoint.dir</name>
<value>file:///soft/hadoop-2.6.5/dfs/namesecondary</value>
</property>
</configuration>
2.2.3 mapred-site.xml
cp mapred-site.xml.template mapred-site.xml
vim mapred-site.xml
内容如下:
<configuration>
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
</configuration>
2.2.4 yarn-site.xml
vim /soft/hadoop-2.6.5/etc/hadoop/yarn-site.xml
内容如下:
<configuration>
<property>
<name>yarn.resourcemanager.hostname</name>
<value>h1</value>
</property>
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
</configuration>
2.2.5 hadoop环境变量
export HADOOP_HOME=/soft/hadoop-2.6.5
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
export HADOOP_COMMON_LIB_NATIVE_DIR=$HADOOP_HOME/lib/native
export HADOOP_OPTS="-Djava.library.path=$HADOOP_HOME/lib"
export PATH=$PATH:$HADOOP_HOME/bin
2.2.6 slaves
vim /soft/hadoop-2.6.5/etc/hadoop/slaves
内容如下:
h1
h2
h3
2.2.7 复制到其他节点
# 先格式化HDFS
/soft/hadoop-2.6.5/bin/hdfs namenode -format
# 复制到主机 h2
scp -r /soft/hadoop-2.6.5/ root@h2:/soft
# 复制到主机 h3
scp -r /soft/hadoop-2.6.5/ root@h3:/soft
# 同时别忘了其他机器上的环境变量
2.2.8 验证
# 启动hdfs----在h1上操作即可
/soft/hadoop-2.6.5/sbin/start-all.sh
# 看到正常启动之后分别在h1,h2,h3上查看java进程快照
# h1
[root@h1 sbin]# jps
4225 SecondaryNameNode
4053 DataNode
3957 NameNode
4539 NodeManager
4574 Jps
4447 ResourceManager
# h2
[root@h2 soft]# jps
3499 Jps
3390 NodeManager
3231 DataNode
# h3
[root@h3 soft]# jps
3194 DataNode
3452 Jps
3342 NodeManager
通过浏览器查看HDFS状态:http://h1:50070/dfshealth.html#tab-overview
tab-datanode标签可以看到如下内容:
datanode infomation2.3 配置spark
# 下载
wget http://d3kbcqa49mib13.cloudfront.net/spark-2.0.1-bin-hadoop2.6.tgz
# 解压
tar -zxvf spark-2.0.1-bin-hadoop2.6.tgz
# 可选(本人喜欢将软件统一安装的根目录的soft目录下)
mv spark-2.0.1-bin-hadoop2.6 /soft/
2.3.1 spark环境变量
export SPARK_HOME=/soft/spark-2.0.1-bin-hadoop2.6/
export PATH=$PATH:$SPARK_HOME/bin
2.3.2 spark-env.sh
cd /soft/spark-2.0.1-bin-hadoop2.6/conf
cp spark-env.sh.template spark-env.sh
vim spark-env.sh
在最后加入如下内容:
export JAVA_HOME=/soft/jdk1.8.0_101
export SCALA_HOME=/soft/scala-2.11.8
export HADOOP_HOME=/soft/hadoop-2.6.5
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
export SPARK_MASTER_IP=h1
export SPARK_WORKER_MEMORY=512M
export SPARK_EXECUTOR_MEMORY=512M
export SPARK_DRIVER_MEMORY=512M
export SPARK_WORKER_CORES=1
2.3.3 slaves
cd /soft/spark-2.0.1-bin-hadoop2.6/conf
cp slaves.template slaves
vim slaves
在最后加入如下内容:
h2
h3
2.3.4 spark-defaults.conf
cd /soft/spark-2.0.1-bin-hadoop2.6/conf
cp spark-defaults.conf.template spark-defaults.conf
vim spark-defaults.conf
在最后加入如下内容:
spark.executor.extraJavaOptions -XX:+PrintGCDetails -DKey=value -Dnumbers="one two three"
spark.eventLog.enabled true
spark.eventLog.dir hdfs://h1:9000/historyserverforSpark
spark.yarn.historyServer.address h1:18080
spark.history.fs.logDirectory hdfs://h1:9000/historyserverforSpark
2.3.5 复制到其他节点
# 复制到h2
scp -r /soft/spark-2.0.1-bin-hadoop2.6/ root@h2:/soft
# 复制到h3
scp -r /soft/spark-2.0.1-bin-hadoop2.6/ root@h3:/soft
# 同时别忘了其他节点的环境变量
验证
# 启动 - 在h1上操作即可
/soft/spark-2.0.1-bin-hadoop2.6/start-all.sh
# h1
[root@h1 sbin]# jps
4435 SecondaryNameNode
4164 NameNode
4583 ResourceManager
4263 DataNode
4680 NodeManager
5020 Master
5084 Jps
# h2
[root@h2 soft]# jps
3284 DataNode
3543 Worker
3591 Jps
3384 NodeManager
# h3
[root@h3 soft]# jps
3280 DataNode
3538 Worker
3380 NodeManager
3588 Jps
通过浏览器查看HDFS状态:http://h1:8080,可以看到如下内容:
spark-works3 执行HelloWorld
此处的HelloWorld指的是Spark内置的例子中的一个用来计算PI的程序。
注意:
在执行该jar包前请先确保hadoop和Spark都已经启动。
另外,如果配置了日志服务,请先为其创建对应的日志目录再执行后续步骤:
hdfs dfs -mkdir /historyserverforSpark
执行
./bin/spark-submit \
--class org.apache.spark.examples.SparkPi \
--master spark://h1:7077 \
--executor-memory 512m \
--total-executor-cores 10 \
/soft/spark-2.0.1-bin-hadoop2.6/examples/jars/spark-examples_2.11-2.0.1.jar \
10
PI结果
下一篇文章将分享在Eclipse中编写并执行SparkHello:spark筑基篇-01-Eclipse开发Spark HelloWorld
网友评论