hadoop分布式系统
一、hadoop
1.hadoop简介
Hadoop是一个由Apache基金会所开发的分布式系统基础架构。
Hadoop的框架最核心的设计就是：
- HDFS
- MapReduce
2.hadoop的技术主要模块
1）Hdfs主要模块：
- NameNode：是整个文件系统的管理节点。维护整个文件系统的文件目录树，文件/目录的元数据和
每个文件对应的数据块列表。接收用户的请求。 - DataNode：是HA（高可用性）的一个解决方案，是备用镜像，但不支持热备
HDFS为海量的数据提供了存储，而MapReduce则为海量的数据提供了计算。
2）Yarn主要模块：
- ResourceManager
- NodeManager
3.hadoop的优点
- 高可靠性。Hadoop按位存储和处理数据的能力值得人们信赖。
- 高扩展性。Hadoop是在可用的计算机集簇间分配数据并完成计算任务的，这些集簇可以方便地扩展到数以千计的节点中。
- 高效性。Hadoop能够在节点之间动态地移动数据，并保证各个节点的动态平衡，因此处理速度非常快。
- 高容错性。Hadoop能够自动保存数据的多个副本，并且能够自动将失败的任务重新分配。
- 低成本。与一体机、商用数据仓库以及QlikView、Yonghong Z-Suite等数据集市相比，hadoop是开源的，项目的软件成本因此会大大降低。
4.hadoop大数据处理的意义
Hadoop得以在大数据处理应用中广泛应用得益于其自身在数据提取、变形和加载(ETL)方面上的天然优势。Hadoop的分布式架构，将大数据处理引擎尽可能的靠近存储，对例如像ETL这样的批处理操作相对合适，因为类似这样操作的批处理结果可以直接走向存储。Hadoop的MapReduce功能实现了将单个任务打碎，并将碎片任务(Map)发送到多个节点上，之后再以单个数据集的形式加载(Reduce)到数据仓库里。
二、部署hadoop
| linux主机 | ip |
|---|---|
| hadoop1 | 172.25.1.1 |
| hadoop2 | 172.25.1.2 |
| hadoop3 | 172.25.1.3 |
| hadoop4 | 172.25.1.4 |
软件 =====> 点击下载 提取码: 6br1
1.搭建单机版hadoop
1.建立用户，设置密码
[[email protected] ~]# useradd -u 1000 hadoop
[[email protected] ~]# passwd hadoop
2.hadoop的安装配置
此时需要切换到hadoop用户操作，把需要用的压缩包放到hadoop家目录下
[[email protected] ~]# ls
hadoop-3.0.3.tar.gz jdk-8u181-linux-x64.tar.gz
[[email protected] ~]# mv * /home/hadoop/
[[email protected] ~]# su - hadoop
Last login: Mon Apr 15 08:46:38 EDT 2019 on pts/0
[[email protected] ~]$ ls
hadoop-3.0.3.tar.gz jdk-8u181-linux-x64.tar.gz
[[email protected] ~]$ tar zxf jdk-8u181-linux-x64.tar.gz
[[email protected] ~]$ ln -s jdk1.8.0_181/ java
[[email protected] ~]$ tar zxf hadoop-3.0.3.tar.gz
[[email protected] ~]$ ln -s hadoop-3.0.3 hadoop
[[email protected] ~]$ ls
hadoop hadoop-3.0.3.tar.gz jdk1.8.0_181
hadoop-3.0.3 java jdk-8u181-linux-x64.tar.gz
3.配置环境变量
[[email protected] ~]$ cd hadoop/etc/hadoop/
[[email protected] hadoop]$ vim hadoop-env.sh
54 export JAVA_HOME=/home/hadoop/java
[[email protected] ~]$ vim .bash_profile
PATH=$PATH:$HOME/.local/bin:$HOME/bin:$HOME/java/bin
[[email protected] ~]$ source .bash_profile
[[email protected] ~]$ jps
11629 Jps
4.测试
[[email protected] hadoop]$ pwd
/home/hadoop/hadoop
[[email protected] hadoop]$ mkdir input
[[email protected] hadoop]$ cp etc/hadoop/*.xml input/
[[email protected] hadoop]$ bin/hadoop jar share/hadoop/mapreduce/hadoop-mapreduce-examples-3.0.3.jar grep input output 'dfs[a-z.]+'
File Input Format Counters
Bytes Read=123
File Output Format Counters
Bytes Written=23 ##这就表示成功了
[[email protected] hadoop]$ cd output/
[[email protected] output]$ ls
part-r-00000 _SUCCESS
[[email protected] output]$ cat *
1 dfsadmin
2.搭建伪分布式hadoop
1.编辑配置文件
[[email protected] hadoop]$ pwd
/home/hadoop/hadoop/etc/hadoop
[[email protected] hadoop]$ vim core-site.xml
<configuration>
<property>
<name>fs.defaultFS</name>
<value>hdfs://localhost:9000</value>
</property>
</configuration>
[[email protected] hadoop]$ vim hdfs-site.xml
<configuration>
<property>
<name>dfs.replication</name>
<value>1</value> ##自己充当节点
</property>
</configuration>
2.生成**做ssh免密
[[email protected] hadoop]$ ssh-******
[[email protected] hadoop]$ ssh-copy-id localhost
3.格式化，并开启服务
[[email protected] hadoop]$ bin/hdfs namenode -format
[[email protected] hadoop]$ cd sbin/
[[email protected] sbin]$ ./start-dfs.sh
Starting namenodes on [localhost]
Starting datanodes
Starting secondary namenodes [hadoop1]
hadoop1: Warning: Permanently added 'hadoop1,fe80::5054:ff:fe9a:f1f5%eth0' (ECDSA) to the list of known hosts.
4.浏览器查看http://172.25.1.1:9870
5.测试并上传
[[email protected] hadoop]$ bin/hdfs dfs -mkdir -p /user/hadoop
[[email protected] hadoop]$ bin/hdfs dfs -ls
[[email protected] hadoop]$ bin/hdfs dfs -put input
[[email protected] hadoop]$ bin/hdfs dfs -ls
Found 1 items
drwxr-xr-x - hadoop supergroup 0 2019-04-15 09:20 input
删除input和output文件，重新执行命令
[[email protected] hadoop]$ rm -fr input/ output/
[[email protected] hadoop]$ bin/hadoop jar share/hadoop/mapreduce/hadoop-mapreduce-examples-3.0.3.jar grep input output 'dfs[a-z.]+'
[[email protected] hadoop]$ ls
bin include libexec logs README.txt share
etc lib LICENSE.txt NOTICE.txt sbin
此时input和output不会出现在当前目录下，而是上传到了分布式文件系统中，网页上可以看到
[[email protected] hadoop]$ bin/hdfs dfs -cat output/*
1 dfsadmin
[[email protected] hadoop]$ bin/hdfs dfs -get output ##从分布式系统中get下来output目录
[[email protected] hadoop]$ cd output/
[[email protected] output]$ ls
part-r-00000 _SUCCESS
[[email protected] output]$ cat *
1 dfsadmin
三、分布式hadoop
1.首先停掉服务
[[email protected] hadoop]$ sbin/stop-dfs.sh ##我这里因为是关机以后第二天写的文章，所以不需要关闭该服务
[[email protected] ~]$ jps
2112 Jps
[[email protected] ~]$ cd /tmp/
[[email protected] tmp]$ rm -fr * ##删除所有的hadoop文件
2.新开hadoop2和hadoop3，新建用户
[[email protected] ~]# useradd -u 1000 hadoop
[[email protected] ~]# useradd -u 1000 hadoop
安装nfs-utils
[[email protected] ~]# yum install -y nfs-utils
[[email protected] ~]# yum install -y nfs-utils
[[email protected] ~]# yum install -y nfs-utils
开启服务
[[email protected] ~]# systemctl start rpcbind
[[email protected] ~]# systemctl start rpcbind
[[email protected] ~]# systemctl start rpcbind
3.hadoop1开启nfs存储，并共享
[[email protected] ~]# systemctl start nfs-server
[[email protected] ~]# vim /etc/exports
/home/hadoop *(rw,anonuid=1000,anongid=1000)
[[email protected] ~]# exportfs -rv
exporting *:/home/hadoop
[[email protected] ~]# showmount -e
Export list for hadoop1:
/home/hadoop *
4.hadoop3、hadoop4挂载
[[email protected] ~]# mount 172.25.1.1:/home/hadoop /home/hadoop
[[email protected] ~]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/rhel-root 17811456 1098404 16713052 7% /
devtmpfs 497300 0 497300 0% /dev
tmpfs 508264 0 508264 0% /dev/shm
tmpfs 508264 13072 495192 3% /run
tmpfs 508264 0 508264 0% /sys/fs/cgroup
/dev/sda1 1038336 141476 896860 14% /boot
tmpfs 101656 0 101656 0% /run/user/0
172.25.1.1:/home/hadoop 17811456 2796416 15015040 16% /home/hadoop
[[email protected] ~]# mount 172.25.1.1:/home/hadoop /home/hadoop
[[email protected] ~]# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/rhel-root 17811456 1097584 16713872 7% /
devtmpfs 497300 0 497300 0% /dev
tmpfs 508264 0 508264 0% /dev/shm
tmpfs 508264 13072 495192 3% /run
tmpfs 508264 0 508264 0% /sys/fs/cgroup
/dev/sda1 1038336 141476 896860 14% /boot
tmpfs 101656 0 101656 0% /run/user/0
172.25.1.1:/home/hadoop 17811456 2796416 15015040 16% /home/hadoop
这样我们就可以免密登陆hadoop2和hadoop3了,因为我们共享了家目录，里面含有ssh的公钥和私钥
5.重新编辑配置文件
[[email protected] ~]# su - hadoop
Last login: Mon Apr 15 23:12:09 EDT 2019 on pts/0
[[email protected] ~]$ cd hadoop/etc/hadoop/
[[email protected] hadoop]$ vim core-site.xml
<configuration>
<property>
<name>fs.defaultFS</name>
<value>hdfs://172.25.1.1:9000</value>
</property>
</configuration>
[[email protected] hadoop]$ vim hdfs-site.xml
<configuration>
<property>
<name>dfs.replication</name>
<value>2</value> ##改为两个节点
</property>
</configuration>
[[email protected] hadoop]$ vim workers
172.25.1.2
172.25.1.3
此时我们会发现其他节点的也有了这个文件，因为是共享的
6.格式化，并启动hadoop服务
[[email protected] hadoop]$ bin/hdfs namenode -format
[[email protected] hadoop]$ sbin/start-dfs.sh
Starting namenodes on [hadoop1]
Starting datanodes
172.25.1.2: Warning: Permanently added '172.25.1.2' (ECDSA) to the list of known hosts.
172.25.1.3: Warning: Permanently added '172.25.1.3' (ECDSA) to the list of known hosts.
Starting secondary namenodes [hadoop1]
[[email protected] hadoop]$ jps
2902 SecondaryNameNode
2682 NameNode
3021 Jps
我们在hadoop2上也可以看到进程
7.测试
[[email protected] hadoop]$ bin/hdfs dfs -mkdir -p /user/hadoop
[[email protected] hadoop]$ bin/hdfs dfs -mkdir input
[[email protected] hadoop]$ bin/hdfs dfs -put etc/hadoop/*.xml input
[[email protected] hadoop]$ bin/hadoop jar share/hadoop/mapreduce/hadoop-mapreduce-examples-3.0.3.jar grep input output 'dfs[a-z.]+'
网页上查看，有两个数据节点
这里我的节点一直初步来，后来发现是我的解析有问题，我怕是感冒脑子坏了
hadoop4模拟客户端
[[email protected] hadoop]$ useradd -u 1000 hadoop
[[email protected] hadoop]$ yum install -y nfs-utils
[[email protected] hadoop]$ systemctl start rpcbind
[[email protected] hadoop]$ mount 172.25.14.1:/home/hadoop /home/hadoop
[[email protected] hadoop]$ su - hadoop
[[email protected] hadoop]$ vim /home/hadoop/hadoop/etc/hadoop/workers
172.25.1.2
172.25.1.3
172.25.1.4
然后在hadoop1把hadoop重启
新建并上传文件
然后在网页端查看
ok～