搭建部署Hadoop 之 HDFS
HDFS Hadoop 分布式文件系统
分布式文件系统
分布式文件系统可以有效解决数据的存储和管理难题
- 将固定于某个地点的某个文件系统,扩展到任意多个地点/多个文件系统
- 众多的节点组成一个文件系统网络
- 每个节点可以分布在不同的地点,通过网络进行节点间的通信和数据传输
- 人们在使用分布式文件系统时,无需关心数据是存储在哪个节点上、或者是从哪个节点从获取的,只需要像使用本地文件系统一样管理和存储文件系统中的数据
HDFS 角色及概念
• 是Hadoop体系中数据存储管理的基础。它是一个高度容错的系统,用于在低成本的通用硬件上运行。
• 角色和概念
- Client
- Namenode
- Secondarynode
- Datanode
• NameNode
- Master节点,管理HDFS的名称空间和数据块映射信息,配置副本策略,处理所有客户端请求。
• Secondary NameNode
- 定期合并 fsimage 和fsedits,推送给NameNode
- 紧急情况下,可辅助恢复NameNode,
• 但Secondary NameNode并非NameNode的热备。
• DataNode
- 数据存储节点,存储实际的数据
- 汇报存储信息给NameNode。
• Client
- 切分文件
- 访问HDFS
- 与NameNode交互,获取文件位置信息
- 与DataNode交互,读取和写入数据。
• Block
- 每块缺省64MB大小
- 每块可以多个副本
搭建部署 HDFS 分布式文件系统
实验环境准备：
# vim /etc/hosts
.. ..
192.168.4.1master
192.168.4.2node1
192.168.4.3node2
192.168.4.4node3
# sed -ri "/Host */aStrictHostKeyChecking no" /etc/ssh/ssh_config
# ssh-keygen
# for i in {1..4}
> do
> ssh-copy-id 192.168.4.${i}
> done
# for i in {1..4} //同步本地域名
> do
> rsync -a /etc/hosts 192.168.4.${i}:/etc/hosts
> done
# rm -rf /etc/yum.repos.d/*
# vim /etc/yum.repos.d/yum.repo //配置网络yum
[yum]
name=yum
baseurl=http://192.168.4.254/rhel7
gpgcheck=0
# for i in {2..4}
> do
> ssh 192.168.4.${i} "rm -rf /etc/yum.repos.d/*"
> rsync -a /etc/yum.repos.d/yum.repo 192.168.4.${i}:/etc/yum.repos.d/
> done
# for i in {1..4}
> do
> ssh 192.168.4.${i} 'sed -ri "s/^(SELINUX=).*/\1disabled/" /etc/selinux/config ; yum -y remove firewalld'
> done
//所有机器重启
搭建完全分布式
系统规划：
主机 角色 软件
192.168.4.1 master NameNode SecondaryNameNode HDFS
192.168.4.2 node1 DataNode HDFS
192.168.4.3 node2 DataNode HDFS
192.168.4.4 node3 DataNode HDFS
在所有系统上安装java 环境和调试工具jtarps
# for i in {1..4}
> do
> ssh 192.168.4.${i} "yum -y install java-1.8.0-openjdk-devel.x86_64"
> done
# which java
/usr/bin/java
# readlink -f /usr/bin/java
/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.65-3.b17.el7.x86_64/jre/bin/java
安装 hadoop
# tar -xf hadoop-2.7.3.tar.gz
# mv hadoop-2.7.3 /usr/local/hadoop
修改配置
# cd /usr/local/hadoop/
# sed -ri "s;(export JAVA_HOME=).*;\1/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.65-3.b17.el7.x86_64/jre;" etc/hadoop/hadoop-env.sh
# sed -ri "s;(export HADOOP_CONF_DIR=).*;\1/usr/local/hadoop/etc/hadoop;" etc/hadoop/hadoop-env.sh
# sed -n "25p;33p" etc/hadoop/hadoop-env.sh
export JAVA_HOME=/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.65-3.b17.el7.x86_64/jre
export HADOOP_CONF_DIR=/usr/local/hadoop/etc/hadoop
//配置参数说明 网站http://hadoop.apache.org/docs/r2.7.5/hadoop-project-dist/hadoop-common/core-default.xml
# vim etc/hadoop/core-site.xml
.. ..
//所有机器上创建 根目录
# for i in {1..4}
> do
> ssh 192.168.4.${i} "mkdir /var/hadoop"
> done
//配置参数说明 网站http://hadoop.apache.org/docs/r2.7.5/hadoop-project-dist/hadoop-hdfs/hdfs-default.xml
# vim etc/hadoop/hdfs-site.xml
# vim etc/hadoop/slaves //配置去那些主机上寻找 DataNode
node1
node2
node3
配置完成以后,把 hadoop 的文件夹拷贝到所有机器
# for i in {2..4}
> do
> rsync -azSH --delete /usr/local/hadoop 192.168.4.${i}:/usr/local/ -e "ssh"
> done
//在 NameNode 下执行格式化 Hadoop
# ./bin/hdfs namenode -format
看见 successfully formatted. 说明 格式化成功了
//在没有报错的情况下 启动集群
# ./sbin/start-dfs.sh
启动以后分别在 namenode 和 datanode执行命令
# for i in master node{1..3}
> do
> echo $i
> ssh ${i} "jps"
> done
master
4562 SecondaryNameNode
4827 NameNode
5149 Jps
node1
3959 DataNode
4105 Jps
node2
3957 Jps
3803 DataNode
node3
3956 Jps
3803 DataNode
# ./bin/hdfs dfsadmin -report //查看注册成功的节点
Configured Capacity: 160982630400 (149.93 GB)
Present Capacity: 150644051968 (140.30 GB)
DFS Remaining: 150644039680 (140.30 GB)
DFS Used: 12288 (12 KB)
DFS Used%: 0.00%
Under replicated blocks: 0
Blocks with corrupt replicas: 0
Missing blocks: 0
Missing blocks (with replication factor 1): 0
-------------------------------------------------
Live datanodes (3):
Name: 192.168.4.2:50010 (node1)
Hostname: node1
Decommission Status : Normal
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 4096 (4 KB)
Non DFS Used: 3446755328 (3.21 GB)
DFS Remaining: 50214117376 (46.77 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.58%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Mon Jan 29 21:17:39 EST 2018
Name: 192.168.4.4:50010 (node3)
Hostname: node3
Decommission Status : Normal
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 4096 (4 KB)
Non DFS Used: 3445944320 (3.21 GB)
DFS Remaining: 50214928384 (46.77 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.58%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Mon Jan 29 21:17:39 EST 2018
Name: 192.168.4.3:50010 (node2)
Hostname: node2
Decommission Status : Normal
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 4096 (4 KB)
Non DFS Used: 3445878784 (3.21 GB)
DFS Remaining: 50214993920 (46.77 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.58%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Mon Jan 29 21:17:39 EST 2018
namenode
secondarynamenode
datanode
HDFS 基本使用
HDFS 基本命令 几乎和shell命令相同
# ./bin/hadoop fs -ls hdfs://master:9000/
# ./bin/hadoop fs -mkdir /test
# ./bin/hadoop fs -ls /
Found 1 items
drwxr-xr-x - root supergroup 0 2018-01-29 21:35 /test
# ./bin/hadoop fs -rmdir /test
# ./bin/hadoop fs -mkdir /input
# ./bin/hadoop fs -put *.txt /input //上传文件
# ./bin/hadoop fs -ls /input
Found 3 items
-rw-r--r-- 2 root supergroup 84854 2018-01-29 21:37 /input/LICENSE.txt
-rw-r--r-- 2 root supergroup 14978 2018-01-29 21:37 /input/NOTICE.txt
-rw-r--r-- 2 root supergroup 1366 2018-01-29 21:37 /input/README.txt
# ./bin/hadoop fs -get /input/README.txt /root/ //下载文件
# ls /root/README.txt
/root/README.txt
HDFS 增加节点
- 1. 配置所有hadoop环境,包括主机名、ssh免密码登录、禁用 selinux、iptables、安装 java 环境
[root@newnode ~]# yum -y install java-1.8.0-openjdk-devel.x86_64
[root@master ~] # cat /etc/hosts
192.168.4.1 master
192.168.4.2 node1
192.168.4.3 node2
192.168.4.4 node3
192.168.4.5 newnode
- 2. 修改namenode的slaves文件增加该节点
[root@master ~]# cd /usr/local/hadoop/etc/hadoop/
[root@master hadoop]# echo newnode >> slaves
- 3. 把namnode的配置文件复制到配置文件目录下
# cat /root/rsyncfile.sh
#!/bin/bash
for i in node{2..4}
do
rsync -azSH --delete /usr/local/hadoop/etc/hadoop ${i}:/usr/local/hadoop/etc/ -e 'ssh' &
done
wait
[root@master hadoop]# bash /root/rsyncfile.sh
[root@newnode ~]# rsync -azSH --delete master:/usr/local/hadoop /usr/local
- 5. 在该节点启动Datanode
[root@newnode ~]# cd /usr/local/hadoop/
[root@newnode hadoop]# ./sbin/hadoop-daemon.sh start datanode
[root@newnode hadoop]# jps
4007 Jps
3705 DataNode
- 6. 查看集群状态
[root@master hadoop]# cd /usr/local/hadoop/
[root@master hadoop]# ./bin/hdfs dfsadmin -report
Safe mode is ON
Configured Capacity: 268304384000 (249.88 GB)
Present Capacity: 249863049216 (232.70 GB)
DFS Remaining: 249862311936 (232.70 GB)
DFS Used: 737280 (720 KB)
DFS Used%: 0.00%
Under replicated blocks: 0
Blocks with corrupt replicas: 0
Missing blocks: 0
Missing blocks (with replication factor 1): 0
-------------------------------------------------
Live datanodes (5):
...
Name: 192.168.4.5:50010 (newnode)
Hostname: newnode
Decommission Status : Normal
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 4096 (4 KB)
Non DFS Used: 3662835712 (3.41 GB)
DFS Remaining: 49998036992 (46.56 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.17%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Sun Jan 28 20:30:23 EST 2018
...
- 7. 设置同步带宽,并同步数据
[root@master hadoop]# ./bin/hdfs dfsadmin -setBalancerBandwidth 67108864
[root@master hadoop]# ./sbin/start-balancer.sh -threshold 5
缩减节点
- 配置NameNode的hdfs-site.xml
- dfs.replication 副本数量
- 增加 dfs.hosts.exclude 配置
[root@master hadoop]# vim etc/hadoop/hdfs-site.xml
...
...
- 增加 exclude 配置文件,写入要删除的节点 ip
[root@master hadoop]# vim etc/hadoop/slaves
node1
node2
node3
[root@master hadoop]# vim etc/hadoop/exclude
newnode
# cat /root/rsyncfile.sh
#!/bin/bash
for i in node{1..5}
do
rsync -azSH --delete /usr/local/hadoop/etc/hadoop ${i}:/usr/local/hadoop/etc/ -e 'ssh' &
done
wait
[root@master hadoop]# bash /root/rsyncfile.sh
[root@master hadoop]# ./bin/hdfs dfsadmin -refreshNodes
[root@master hadoop]# ./bin/hdfs dfsadmin -report
...
Name: 192.168.4.6:50010 (newnode)
Hostname: newnode
Decommission Status : Decommission in progress //数据迁移状态
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 12288 (12 KB)
Non DFS Used: 3662950400 (3.41 GB)
DFS Remaining: 49997914112 (46.56 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.17%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Sun Jan 28 20:52:01 EST 2018
...
[root@master hadoop]# ./bin/hdfs dfsadmin -report
...
Name: 192.168.4.6:50010 (newnode)
Hostname: newnode
Decommission Status : Decommissioned //最终状态
Configured Capacity: 53660876800 (49.98 GB)
DFS Used: 12288 (12 KB)
Non DFS Used: 3662950400 (3.41 GB)
DFS Remaining: 49997914112 (46.56 GB)
DFS Used%: 0.00%
DFS Remaining%: 93.17%
Configured Cache Capacity: 0 (0 B)
Cache Used: 0 (0 B)
Cache Remaining: 0 (0 B)
Cache Used%: 100.00%
Cache Remaining%: 0.00%
Xceivers: 1
Last contact: Sun Jan 28 20:52:43 EST 2018
...
//当节点状态变为 Decommissioned 状态时 才能停止节点
[root@newnode hadoop]# ./sbin/hadoop-daemon.sh stop datanode
[root@newnode hadoop]# jps
4045 Jps
