HDFS distributed cluster structures [Basic]
Environment: A Case Study with CentOS6.5 (3 nodes)
- Close and close the firewall from the start
- selinux set to close
- Set the host name
- Setting up host mapping jdk1.7
- ssh-free secret landing
- hadoop-2.5.2
ready
Decompression hadoop, and created in the root directory hadoop data / tmp directory
Modify the configuration file
* Configuration file location: etc / hadoop hadoop directory in
hadoop-env.sh
export JAVA_HOME=/usr/java/jdk1.7.0_71 #jdk路径
core-site.xml
<!-- 用于设置namenode并且作为Java程序的访问入口 --->
<!-- hadoop1.msk.com 为主机名 -->
<property>
<name>fs.defaultFS</name>
<value>hdfs://hadoop1.msk.com:8020</value>
</property>
<!-- 存储NameNode持久化的数据,DataNode块数据 -->
<!-- 手工创建$HADOOP_HOME/data/tmp -->
<property>
<name>hadoop.tmp.dir</name>
<value>/opt/install/hadoop-2.5.2/data/tmp</value>
</property>
hdfs-site.xml
<!-- 设置副本数量 默认是3 可自行根据需求更改 -->
<property>
<name>dfs.replication</name>
<value>3</value>
</property>
<!-- 权限,可省略 -->
<property>
<name>dfs.permissions.enabled</name>
<value>false</value>
</property>
mapred-site.xml
<!-- yarn 与 MapReduce相关 -->
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
yarn-site.xml
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
slaves
configured here DataNode hostname of the machine is limited, there is also a node 1 is the DataNode NameNode
hadoop1.msk.com
hadoop2.msk.com
hadoop3.msk.com
NameNode format
Objective function: hdfs system format, and generates data blocks stored in the directory
bin/hdfs namenode -format
Start | stop running NameNode hadoop [node]
sbin/start-dfs.sh
sbin/stop-dfs.sh
I want to test whether a successful start jps command to view the available process
HDFS shell access
- Check the directory structure
bin/hdfs dfs -ls 路径
- Create a folder
bin/hdfs dfs -mkdir /a
bin/hdfs dfs -mkdir -p /a/b
- Upload files to the local hdfs
bin/hdfs dfs -put local_path hdfs_path
- View the file contents
bin/hdfs dfs -text /a/c
bin/hdfs dfs -cat /a/c
- delete
bin/hdfs dfs -rm /a/c
- Delete non-empty folders
bin/hdfs dfs -rmr /a
- From hdfs download files to local
bin/hdfs dfs -get hdfs_path local_path
HDFC browser access
http://hadoop1.msk.com:50070 access HDFS
http://hadoop1.msk.com:8088 access yarn