一、启动HDFS集群

1、启动Zookeeper

命令：runRemoteCmd.sh “/home/hadoop/app/zookeeper/bin/zkServer.sh start” all

[hadoop@hadoop60 app]$ runRemoteCmd.sh "/home/hadoop/app/zookeeper/bin/zkServer.sh start" all
*******************hadoop60***********************
ZooKeeper JMX enabled by default
Using config: /home/hadoop/app/zookeeper/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED
*******************hadoop61***********************
ZooKeeper JMX enabled by default
Using config: /home/hadoop/app/zookeeper/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED
*******************hadoop62***********************
ZooKeeper JMX enabled by default
Using config: /home/hadoop/app/zookeeper/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED
runRemoteCm
d.sh "

2、启动journalnode

命令：runRemoteCmd.sh “/home/hadoop/app/hadoop/bin/hdfs --daemon start journalnode” all

[hadoop@hadoop60 app]$ runRemoteCmd.sh  "/home/hadoop/app/hadoop/bin/hdfs --daemon start journalnode" all
*******************hadoop60***********************
journalnode is running as process 2303.  Stop it first.
*******************hadoop61***********************
journalnode is running as process 2781.  Stop it first.
*******************hadoop62***********************
journalnode is running as process 2763.  Stop it first.

3、格式化NameNode （在hadoop60节点）

命令：[hadoop@hadoop60 hadoop]$bin/hdfs namenode -format

[hadoop@hadoop60 hadoop]$ bin/hdfs namenode -format

注：如果出现下面划线的，说明格式化成功

4、格式化zkfc（hadoop60节点）

命令：[hadoop@hadoop60 hadoop]$ bin/hdfs zkfc -formatZK

[hadoop@hadoop60 hadoop]$ bin/hdfs zkfc -formatZK

注：如果出现下面划线的，说明格式化成功

5、备用节点NameNode同步主节点NameNode元数据

（1）主节点hadoop60启动NameNode

[hadoop@hadoop60 hadoop]$ bin/hdfs --daemon start namenode

（2）查看jps进程，发现namenode已启动

[hadoop@hadoop60 hadoop]$ jps
2724 Jps
2681 NameNode
2303 JournalNode
2463 QuorumPeerMain

6、查看主节点NameNode信息

[hadoop@hadoop60 hadoop]$ cat /home/hadoop/data/hadoop3/tmp/dfs/name/current/VERSION 
#Sat Oct 30 09:39:34 CST 2021
namespaceID=1264475431
clusterID=CID-631a92bd-c0c3-4cbd-9b0b-0af8d35f437e
cTime=1635557974151
storageType=NAME_NODE
blockpoolID=BP-1478213952-192.168.24.60-1635557974151
layoutVersion=-65
[hadoop@hadoop60 hadoop]$ 

7、hadoop62同步hadoop63节点NameNode元数据

[hadoop@hadoop61 hadoop]$ bin/hdfs namenode -bootstrapStandby
[hadoop@hadoop62 hadoop]$  bin/hdfs namenode -bootstrapStandby

注：如果出现下面划线的，说明同步成功

8、查看备用节点NameNode信息

注：各个节点对照一下看是否一致，如果一致，则namenode数据同步成功。

[hadoop@hadoop62 hadoop]$ cat /home/hadoop/data/hadoop3/tmp/dfs/name/current/VERSION 
#Sat Oct 30 10:08:55 CST 2021
namespaceID=1264475431
clusterID=CID-631a92bd-c0c3-4cbd-9b0b-0af8d35f437e
cTime=1635557974151
storageType=NAME_NODE
blockpoolID=BP-1478213952-192.168.24.60-1635557974151
layoutVersion=-65

9、关闭主节点NameNode进程

[hadoop@hadoop60 hadoop]$ bin/hdfs --daemon stop namenode

10、jps看一下进程，发现namenode关掉了

[hadoop@hadoop60 hadoop]$ jps
3041 Jps
2303 JournalNode
2463 QuorumPeerMain

11、关闭journalnode进程

[hadoop@hadoop60 hadoop]$ runRemoteCmd.sh "/home/hadoop/app/hadoop/bin/hdfs --daemon stop journalnode" all
*******************hadoop60***********************
*******************hadoop61***********************
*******************hadoop62***********************
[hadoop@hadoop60 hadoop]$ 

12、一键启动HDFS集群

（1）注：启动前zookeeper是不能关闭的，jps查看下，发现
QuorumPeerMain还在就OK

[hadoop@hadoop60 hadoop]$ jps
3355 Jps
2463 QuorumPeerMain

（2）一键启动HDFS集群

[hadoop@hadoop60 hadoop]$ sbin/start-dfs.sh
Starting namenodes on [hadoop60 hadoop61 hadoop62]
Starting datanodes
Starting journal nodes [hadoop60 hadoop62 hadoop61]
Starting ZK Failover Controllers on NN hosts [hadoop60 hadoop61 hadoop62]
[hadoop@hadoop60 hadoop]$ 

（3）查看各节点NameNode状态（一个active，两个standby，如果是的话，说明启动OK）

[hadoop@hadoop62 hadoop]$ bin/hdfs haadmin -getServiceState nn1
active
[hadoop@hadoop62 hadoop]$ bin/hdfs haadmin -getServiceState nn2
standby
[hadoop@hadoop62 hadoop]$ bin/hdfs haadmin -getServiceState nn3
standby

13、web界面查看HDFS（百度输入IP地址:9870）

14、HDFS shell

（1)查看hdfs文件系统目录，可以看出hdfs文件和目录为空。

[hadoop@hadoop60 hadoop]$ bin/hdfs dfs -ls /
[hadoop@hadoop60 hadoop]$ 

(2)本地新建一个文件wd.txt

[hadoop@hadoop60 hadoop]$ vi wd.txt
[hadoop@hadoop60 hadoop]$ cat wd.txt 
hadoop hadoop hadoop
[hadoop@hadoop60 hadoop]$ 

(3)新建目录qwq

[hadoop@hadoop60 hadoop]$ bin/hdfs dfs -mkdir /qwq

(4)查看hdfs文件系统目录，可以看到qwq已经存在

[hadoop@hadoop60 hadoop]$ bin/hdfs dfs -ls /
Found 1 items
-rw-r--r--   3 hadoop supergroup         21 2021-10-30 10:43 /qwq

(5)在web页面，也可以看到test已经存在

(6)将本地文件上传到hdfs qwq目录下

[hadoop@hadoop60 hadoop]$ bin/hdfs dfs -put wd.txt /qwq

(7)查看hdfs文件系统目录，可以看到wd.txt已经存在

[hadoop@hadoop60 hadoop]$ bin/hdfs dfs -ls /qwq
Found 1 items
-rw-r--r--   3 hadoop supergroup         21 2021-10-30 13:32 /qwq/wd.txt

(8)在web页面，也可以看到wd.txt已经存在