系统拓扑
角色 | ip地址 | hdfs | yarn |
Master | 10.0.0.201 | NameNode | ResourceManager |
slave | 10.0.0.202 | DataNode | NodeManager |
slave | 10.0.0.203 | DataNode | NodeManager |
前提条件
安装jdk
关闭selinux
关闭防火墙或者配置端口
设置ssh免密登录
在所有节点生RSA密钥对, 不要设置私钥密码。
ssh-keygen
把所有节点的公钥加入authorized_keys文件。
cat id_rsa.pub >> authorized_keys
cat id_rsa.pub.202 >> authorized_keys
cat id_rsa.pub.203 >> authorized_keys
scp authorized_keys root@10.0.0.202:/root/.ssh/authorized_keys
scp authorized_keys root@10.0.0.203:/root/.ssh/authorized_keys
ssh node1
ssh node2
ssh node3
设置/etc/hosts
scp /etc/hosts root@10.0.0.202:/etc/hosts
scp /etc/hosts root@10.0.0.203:/etc/hosts
设置环境变量
/etc/profile
配置Hadoop
解压hadoop
并创建目录
mkdir –p dfs/name
mkdir –p dfs/data
vim /etc/profile
exportHADOOP_HOME=/usr/local/src/Hadoop-2.7.0
vim ./etc/Hadoop/hadoop-env.sh
export JAVA_HOME=/usr/local/src/jdk1.7.0_60
core-site.xml
hdfs-site.xml
mapred-site.xml
yarn-site.xml
yarn.nodemanager.resource.memory-mb必须大于等于2048
图中有误,改为yarn.nodemanager.aux-services.mapreduce_shuffle.class
修改slaves
分发hadoop
scp -r ./hadoop-2.7.0 root@10.0.0.202:/usr/local/src
scp -r ./hadoop-2.7.0 root@10.0.0.203:/usr/local/src
运行hadoop
运行HDFS
bin/hdfs namenode –format (运行一次)
sbin/start-dfs.sh
运行yarn
sbin/yarn-daemon.sh --config ./etc/hadoop start proxyserver
sbin/start-yarn.sh
运行mapreduce JobHistory Server
sbin/mr-jobhistory-daemon.sh –config ./etc/hadoopstart historyserver
查看进程
201
202
203
测试hadoop
hdfs namenode
secondary namenode
yarn ResourceManager
MapReduce JobHistory Server
运行mapreduce worldcount
bin/hadoop jar./share/hadoop/mapreduce/hadoop-mapred uce-examples-2.7.0.jarwordcount /example/input /example/output
停止hadoop
停止hdfs
sbin/stop-dfs.sh
停止yarn
sbin/stop-yarn.sh
停止WebAppProxy
sbin/yarn-daemon.sh –config ./etc/hadoopstop proxyserver
停止MapReduce JobHistory Server
sbin/mr-jobhistory-daemon.sh –config ./etc/hadoopstop historyserver