Hadoop集群总体规划
Hadoop集群安装采用下面步骤:
(这里我已经用Xshell连接到Master结点了)
Hadoop安装包链接:https://pan.baidu.com/s/1teHwnBH2Qm6F7iWZ3q-hSQ
提取码:cgnb
tar -zxf hadoop-2.6.5.tar.gz -C /opt/
首先进入/opt/hadoop-2.6.5/etc/hadoop/
目录
cd /opt/hadoop-3.1.4/etc/hadoop/
配置hadoop-env.sh
vi hadoop-env.sh
首先进入/opt/hadoop-2.6.5/etc/hadoop/
目录
cd /opt/hadoop-3.1.4/etc/hadoop/
配置core-site.xml
vi core-site.xml
在
间添加如下
hadoop.tmp.dir
/data/hadoop/tmp
fs.defaultFS
hdfs://master:8020
vi hdfs-site.xml
在
间添加如下
dfs.namenode.http-address
master:50070
dfs.replication
3
dfs.permissions.enabled
false
dfs.blocksize
134217728
dfs.namenode.name.dir
/data/hadoop/namenode
dfs.datanode.name.dir
/data/hadoop/datanode
首先进入/opt/hadoop-2.6.5/etc/hadoop/
目录
cd /opt/hadoop-3.1.4/etc/hadoop/
配置mapred-site.xml
vi mapred-site.xml
在
间添加如下
mapreduce.framework.name
yarn
yarn.app.mapreduce.am.env
HADOOP_MAPRED_HOME=/opt/hadoop-3.1.4
mapreduce.application.classpath
/opt/hadoop-3.1.4/share/hadoop/mapreduce/*:/opt/hadoop-3.1.4/share/hadoop/mapreduce/lib/*
配置yarn-site.xml
vi yarn-site.xml
在
间添加如下
yarn.resourcemanager.hostname
master
yarn.nodemanager.aux-services
mapreduce_shuffle
yarn.nodemanager.vmem-check-enabled
false
配置workers
vi workers
把里面内容改为
node1
node2
node3
进入/opt
目录,通过指令(依次执行)拷贝到node1、node2、node3目录
(可以多建立几个会话,速度会快些)
scp -r hadoop-3.1.4/ node1:/opt/
scp -r hadoop-3.1.4/ node2:/opt/
scp -r hadoop-3.1.4/ node3:/opt/
(依次执行下面语句)
mkdir -p /data/hadoop/tmp
mkdir -p /data/hadoop/namenode
ssh node1 "mkdir -p /data/hadoop/tmp & mkdir -p /data/hadoop/datanode"
ssh node2 "mkdir -p /data/hadoop/tmp & mkdir -p /data/hadoop/datanode"
ssh node3 "mkdir -p /data/hadoop/tmp & mkdir -p /data/hadoop/datanode"
进入hadoop安装下的bin目录
cd /opt/hadoop-3.1.4/bin/
执行
./hdfs namenode -format demo