CentOS6.5安装Spark1.6环境（Master-Slave模式）

最新推荐文章于 2021-06-11 19:53:32 发布

nooneyanghuijava

最新推荐文章于 2021-06-11 19:53:32 发布

阅读量2.6k

点赞数

分类专栏： spark

本文链接：https://blog.csdn.net/nooneyanghuijava/article/details/51952828

版权

spark 专栏收录该内容

1 篇文章 0 订阅

订阅专栏

1、查看系统环境

cat /etc/redhat-release
uname -r
uname -m

关闭所有服务器的防火墙

/etc/init.d/iptables stop
chkconfig iptables off
chkconfig --list iptables

2、Spark集群机器规划

一共准备了四台机器

Host	IP	Hadoop	Spark
Node1	192.168.2.128	Master	Master
Node2	192.168.2.130	Slave	Slave
Node3	192.168.2.131	Slave	Slave
Node4	192.168.2.1132	Slave	Slave

3、下载Hadoop2.6.0版本和对应的Spark1.6.0版本、下载jdk-7u65-linux-x64.rpm

在node1上创建一个的soft目录将对应的软件下载到soft目录

mkdir /soft/
cd /soft/
wget http://archive.apache.org/dist/hadoop/core/hadoop-2.6.0/hadoop-2.6.0.tar.gz
wget http://archive.apache.org/dist/spark/spark-1.6.0/spark-1.6.0-bin-hadoop2.6.tgz

4、部署spark分布式集群

4.1、对应IP和系统主机名

4.1.1将系统中的IP和主机名进行一个映射配置/etc/hosts中，方便后期直接操作直接使用主机名称

vim /etc/hosts
#编辑内容为
192.168.2.128 node1
192.168.2.130 node2
192.168.2.131 node3
192.168.2.132 node4

4.1.2、将node1上的/etc/hosts配置文件发送到其它三台主机上

scp /etc/hosts root@192.168.2.130:/etc/hosts
scp /etc/hosts root@192.168.2.131:/etc/hosts
scp /etc/hosts root@192.168.2.132:/etc/hosts

4.2、在四台服务器上创建spark用户

#node1 创建用户
useradd spark
#node2 创建用户
useradd spark
#node3 创建用户
useradd spark
#node4 创建用户
useradd spark

4.3、配置主机直接免密码登录

配置四台主机通过ssh服务进行系统的免密码登录，后期hadoop、spark向slave发送命令时就无需输入对应的口令了。

4.3.1、配置SSH免登录

#node1
ssh-keygen -t rsa -P ''
#node2
ssh-keygen -t rsa -P ''
#node3
ssh-keygen -t rsa -P ''
#node4
ssh-keygen -t rsa -P ''

4.3.2、将node2、node3、node4三台服务器上的/home/spark/.ssh/id_rsa.pub文件拷贝到node1上并进行重命名

#node2
scp /home/spark/.ssh/id_rsa.pub spark@node1:/home/spark/.ssh/node2.id_rsa.pub
#node3
scp /home/spark/.ssh/id_rsa.pub spark@node1:/home/spark/.ssh/node3.id_rsa.pub
#node4
scp /home/spark/.ssh/id_rsa.pub spark@node1:/home/spark/.ssh/node4.id_rsa.pub
#node1
cat /home/spark/.ssh/id_rsa.pub >> /home/spark/.ssh/authorized_keys
cat /home/spark/.ssh/node2.id_rsa.pub >> /home/spark/.ssh/authorized_keys
cat /home/spark/.ssh/node3.id_rsa.pub >> /home/spark/.ssh/authorized_keys
cat /home/spark/.ssh/node4.id_rsa.pub >> /home/spark/.ssh/authorized_keys
chmod 600 /home/spark/.ssh/authorized_keys
scp /home/spark/.ssh/authorized_keys spark@node2:/home/spark/.ssh/authorized_keys
scp /home/spark/.ssh/authorized_keys spark@node3:/home/spark/.ssh/authorized_keys
scp /home/spark/.ssh/authorized_keys spark@node4:/home/spark/.ssh/authorized_keys

4.4、安装JDK

将/soft/jdk-7u65-linux-x64.rpm软件通过scp命令拷贝到其它节点上然后在各个节点上安装jdk

#node1
rpm -ivh /soft/jdk-7u65-linux-x64.rpm

配置jdk的环境变量

vim /etc/profile

配置内容为

export JAVA_HOME=/usr/java/latest
export JRE_HOME=$JAVA_HOME/jre
export PATH=$JAVA_HOME/bin:$PATH
export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

配置完成后通过使其生效

source /etc/profile

其它服务器安装node1上进行配置即可

4.5、安装hadoop

在node1上解压hadoop软件到/usr/local/

tar xf /soft/hadoop-2.6.0.tar.gz -C /usr/local/
ln -sv /usr/local/hadoop-2.6.0/ /usr/local/hadoop

1、将hadoop配置到环境变量中/etc/profile内容：

export HADOOP_HOME=/usr/local/hadoop-2.6.0
export HADOOP_CONF_DIR=$HADOOP_HOME/etc/hadoop
export PATH=$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$PATH

修改hadoop的配置文件

1、文件 slaves，将作为 DataNode 的主机名写入该文件，每行一个，默认为 localhost，所以在伪分布式配置时，节点即作为 NameNode 也作为 DataNode。分布式配置可以保留 localhost，也可以删掉，让 Master 节点仅作为 NameNode 使用。

vim /usr/local/hadoop-2.6.0/etc/hadoop/slaves

文件内容为：

node2
node3
node4

2、修改/usr/local/hadoop/etc/hadoop/core-site.xml文件

<configuration>
        <property>
                <name>fs.defaultFS</name>
                <value>hdfs://node1:9000</value>
        </property>
        <property>
                <name>hadoop.tmp.dir</name>
                <value>/hadoopdata/tmp</value>
                <description>Abase for other temporary directories.</description>
        </property>
</configuration>

3、文件 hdfs-site.xml，dfs.replication 一般设为 3

<configuration>
        <property>
                <name>dfs.namenode.secondary.http-address</name>
                <value>node1:50090</value>
        </property>
        <property>
                <name>dfs.replication</name>
                <value>3</value>
        </property>
        <property>
                <name>dfs.namenode.name.dir</name>
                <value>/hadoopdata/tmp/dfs/name</value>
        </property>
        <property>
                <name>dfs.datanode.data.dir</name>
                <value>/hadoopdata/tmp/dfs/data</value>
        </property>
</configuration>

4、文件 mapred-site.xml （可能需要先重命名，默认文件名为 mapred-site.xml.template），然后配置修改如下

<configuration>
        <property>
                <name>mapreduce.framework.name</name>
                <value>yarn</value>
        </property>
        <property>
                <name>mapreduce.jobhistory.address</name>
                <value>node1:10020</value>
        </property>
        <property>
                <name>mapreduce.jobhistory.webapp.address</name>
                <value>node1:19888</value>
        </property>
</configuration>

5、修改配置文件 yarn-site.xml

<configuration>
        <property>
                <name>yarn.resourcemanager.hostname</name>
                <value>node1</value>
        </property>
        <property>
                <name>yarn.nodemanager.aux-services</name>
                <value>mapreduce_shuffle</value>
        </property>
</configuration>

6、创建hadoop存储数据临时目录

mkdir /hadoopdata/tmp -pv
mkdir -p /hadoopdata/tmp/dfs/{name,data}
chown -R spark:spark /hadoopdata/

7、将node1节点上的hadoop分发到其它的节点上（首先删除/usr/local/hadoop/share/doc/这个文件都是文档）

scp -r /usr/local/hadoop-2.6.0/ root@node2:/usr/local/hadoop-2.6.0/

8、将node1上/etc/profile拷贝到其它服务器上

scp /etc/profile root@node2:/etc/profile
source /etc/profile

9、修改hadoop安装目录用户所组所主

chown -R spark:spark /usr/local/hadoop-2.6.0/

10、格式化hdfs

hdfs namenode -format

11、启动hdfs、yarn

start-dfs.sh
start-yarn.sh

12、查看hadoop集群是否成功

 hdfs dfsadmin -report

http://node1:50070/dfshealth.html#tab-datanode

4.6、安装scala

spark依赖的scala的版本为2.1.x版本所以选择的scala版本为scala-2.10.6.tgz

1、解压scala到/usr/local/中

tar xf /soft/scala-2.10.6.tgz -C /usr/local/

2、配置环境变量

vim /etc/profile
#内容为：
export SCALA_HOME=/usr/local/scala-2.10.6
export PATH=$SCALA_HOME/bin:$PATH
source /etc/profile

其它服务器按照node1服务器上进行安装即可

4.6、安装spark

1、在node1上解压spark软件到/usr/local/

tar xf /soft/spark-1.6.0-bin-hadoop2.6.tgz -C /usr/local/

2、修改spark的配置文件spark-env.sh

cp /usr/local/spark-1.6.0-bin-hadoop2.6/conf/spark-env.sh.template /usr/local/spark-1.6.0-bin-hadoop2.6/conf/spark-env.sh

配置内容如下

export JAVA_HOME=/usr/java/latest
export SCALA_HOME=/usr/local/scala-2.10.6
export HADOOP_HOME=/usr/local/hadoop-2.6.0
export HADOOP_CONF_DIR=/usr/local/hadoop-2.6.0/etc/hadoop
export SPARK_MASTER_IP=node1
export SPARK_WORKER_MEMORY=1g
export SPARK_EXECUTOR_MEMORY=1g
export SPARK_DRIVER_MEMORY=1g
export SPARK_WORKER_CORES=1

3、修改/usr/local/spark-1.6.0-bin-hadoop2.6/conf/slaves

配置内容如下

node2
node3
node4

4、修改/usr/local/spark-1.6.0-bin-hadoop2.6/conf/spark-defaults.conf

cp /usr/local/spark-1.6.0-bin-hadoop2.6/conf/spark-defaults.conf.template /usr/local/spark-1.6.0-bin-hadoop2.6/conf/spark-defaults.conf
#配置内容
spark.executor.extraJavaOptions        -XX:+PrintGCDetails -Dkey=value -Dnumbers="one two three"
spark.eventLog.enabled            true
spark.eventLog.dir            hdfs://node1:9000/historyserverforSpark
spark.yarn.historyServer.address    node1:18080
spark.history.fs.logDirectory        hdfs://node1:9000/historyserverforSpark

注意需要说明的是：配置了spark历史记录信息情况必须要在hdfs文件系统中创建historyserverforSpark此目录，如果没有创建spark是无法启动记录历史信息的进程

hadoop dfs -mkdir /historyserverforSpark

5、配置系统环境变量

vim /etc/profile
#内容为：
export SPARK_HOME=/usr/local/spark-1.6.0-bin-hadoop2.6
export PATH=$SPARK_HOME/bin:$SPARK_HOME/sbin:$PATH
source /etc/profile

6、通过scp将spark分发到其它节点

scp -r /usr/local/spark-1.6.0-bin-hadoop2.6/ root@node2:/usr/local/spark-1.6.0-bin-hadoop2.6/
scp /etc/profile root@node2:/etc/profile

7、修改spark目录的所主所组

chown -R spark:spark /usr/local/spark-1.6.0-bin-hadoop2.6/

8、启动集群spark

/usr/local/spark-1.6.0-bin-hadoop2.6/sbin/start-all.sh 
#启动一个记录历史程序的进程
/usr/local/spark-1.6.0-bin-hadoop2.6/sbin/start-history-server.sh

查看是否正常启动

9、启动HistoryServer记录spark程序历史记录

/usr/local/spark-1.6.0-bin-hadoop2.6/sbin/start-history-server.sh

5、测试运行一个spark程序

/usr/local/spark-1.6.0-bin-hadoop2.6/bin/spark-submit --class org.apache.spark.examples.SparkPi --master spark://node1:7077 /usr/local/spark-1.6.0-bin-hadoop2.6/lib/spark-examples-1.6.0-hadoop2.6.0.jar 1000

spark提交任务

./bin/spark-submit \
--class <main-class> \
--master <master-url> \
--deploy-mode <deploy-mode> \
--conf <key>=<value> \
... # other options
<application-jar> \
[application-arguments]

1、--class: The entry point for your application (e.g. org.apache.spark.examples.SparkPi)#程序运行的主路径包含（包名+main类名）
2、--master: The master URL for the cluster (e.g. spark://23.195.26.187:7077)#集群中master的访问入口
3、--deploy-mode: Whether to deploy your driver on the worker nodes (cluster) or locally as an external client (client) (default: client)#spark作业提交模式
4、--conf: Arbitrary Spark configuration property in key=value format. For values that contain spaces wrap “key=value” in quotes (as shown).#spark提交作业的参数设置，参数设置为（key=value）这种模式
5、application-jar: Path to a bundled jar including your application and all dependencies. The URL must be globally visible inside of your cluster, for instance, an hdfs:// path or a file:// path that is present on all nodes.#spark提交作业的jar路径
6、application-arguments: Arguments passed to the main method of your main class, if any#此参数可以通过main中的参数输入到程序中