3.hadoop HA-QJM 安装

概述

环境：hadoop 3.3.6
jdk : 1.8
linux 环境：

[root@hadoop01 soft]# uname -a
Linux hadoop01 3.10.0-693.el7.x86_64 #1 SMP Tue Aug 22 21:09:27 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

相关文章链接如下：

文章名称	链接
hadoop安装基础环境安装一	地址
hadoop一主三从安装	地址

官网文档(单节点安装)
官网文档速递

安装需要的包：hadoop-3.3.6.tar.gz zookeeper-3.4.14.tar.gz

HA 各组件机器分配如下：

hostname	ip	组件
hadoop01	10.xx.3x.142	NN RM
hadoop02	10.xx.3x.143	NN
hadoop03	10.xx.3x.144	NM DN
hadoop04	10.xx.3x.145	NM DN

注意：原来的这些机器 hadoop01 与其它机器做了免密，jdk 还是原有的配置， 需要做 hadoop02 与其它机器做免密。

免密请参考 hadoop安装基础环境安装一

实践

一主两从

解压

# 解压
[root@hadoop01 soft]# tar -zxvf  hadoop-3.3.6.tar.gz 
[root@hadoop01 soft]# ls
hadoop-3.3.6  hadoop-3.3.6.tar.gz  zookeeper-3.4.14.tar.gz

配置文件

hadoop-env.sh

[root@hadoop01 hadoop]# pwd
/data/hadoop/soft/hadoop-3.3.6/etc/hadoop

[root@hadoop01 hadoop]# vi hadoop-env.sh 
export JAVA_HOME=/data/soft/jdk1.8

core-site.xml

# 创建存储 hadoop_repo
mkdir -p /data/hadoop/hadoop_repo
[root@hadoop01 hadoop]# vi core-site.xml 
加入下面的配置内容

<property>
    <name>fs.defaultFS</name>
    <value>hdfs://hadoop01:9000</value>
</property>
    <!-- 是 hadoop文件系统依赖的基本配置,比如跑MapReduce时生成的临时路径本质上其实就是生成在它的下面 -->
<property>
    <name>hadoop.tmp.dir</name>
    <value>/data/hadoop/hadoop_repo</value>
</property>
<!-- 删除文件存在时间 -->
<property>
    <name>fs.trash.interval</name>
    <value>1440</value>
</property>
<property>
    <name>hadoop.proxyuser.root.hosts</name>
    <value>*</value>
</property>
<property>
    <name>hadoop.proxyuser.root.groups</name>
    <value>*</value>
</property>

hdfs-site.xml

因为只有两个 DN 所以分片设置为 2

[root@hadoop01 hadoop]# vi hdfs-site.xml 

 <property>
     <name>dfs.replication</name>
     <value>2</value>
 </property>

yarn-site.xml

[root@hadoop01 hadoop]# vi yarn-site.xml 

yarn-resourcemanager.webapp.address hadoop01:37856 可以定制对外访问地址 8088易容易被挖矿

<property>
    <name>yarn.nodemanager.aux-services</name>
    <value>spark_shuffle,mapreduce_shuffle</value>
</property>
<property>
　　<name>yarn.nodemanager.aux-services.spark_shuffle.class</name>
　　<value>org.apache.spark.network.yarn.YarnShuffleService</value>
</property>
<property>
    <name>yarn.nodemanager.env-whitelist</name>
	<value>JAVA_HOME,HADOOP_COMMON_HOME,HADOOP_HDFS_HOME,HADOOP_CONF_DIR,CLASSPATH_PREPEND_DISTCACHE,HADOOP_YARN_HOME,HADOOP_MAPRED_HOME</value>
</property>
<property>
    <name>yarn.resourcemanager.hostname</name>
    <value>hadoop01</value>
</property>
<property>
    <name>yarn.log-aggregation-enable</name>
    <value>true</value>
</property>
<property>
    <name>yarn.log.server.url</name>
    <value>http://hadoop01:19888/jobhistory/logs/</value>
</property>

<property>
    <name>yarn.nodemanager.resource.memory-mb</name>
    <value>10240</value>
    <description>该节点上YARN可使用的物理内存总量</description>
</property>
<property>
    <name>yarn.nodemanager.resource.cpu-vcores</name>
    <value>8</value>
    <description>表示集群中每个节点可被分配的虚拟CPU个数</description>
</property>
<property>
    <name>yarn.scheduler.minimum-allocation-mb</name>
    <value>1024</value>
    <description>每个容器container请求被分配的最小内存</description>
</property>
<property>
    <name>yarn.scheduler.maximum-allocation-mb</name>
    <value>8096</value>
    <description>每个容器container请求被分配的最大内存,不能给太大</description>
</property>
<property>
    <name>yarn.scheduler.minimum-allocation-vcores</name>
    <value>1</value>
    <description>每个容器container请求被分配的最少虚拟CPU个数</description>
</property>
<property>
    <name>yarn.scheduler.maximum-allocation-vcores</name>
    <value>4</value>
    <description>每个容器container请求被分配的最多虚拟CPU个数</description>
</property>

mapred-site.xml

[root@hadoop01 hadoop]# vi mapred-site.xml 

<property>
    <name>mapreduce.framework.name</name>
    <value>yarn</value>
</property>

workers

[root@hadoop01 hadoop]# vi workers 
[root@hadoop01 hadoop]# cat workers 
hadoop03
hadoop04

分发

# 每个节点都要创建
mkdir -p /data/hadoop/soft 

scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop02:/data/hadoop/soft/hadoop-3.3.6
scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop03:/data/hadoop/soft/hadoop-3.3.6
scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop04:/data/hadoop/soft/hadoop-3.3.6

分发时，注意检查一下配置是否生效

环境变量

根据自己本机环境变量配置

vi /etc/profile 或 ~.bash_profile

格式化

[root@hadoop01 bin]# pwd
/data/hadoop/soft/hadoop-3.3.6/bin

./hdfs namenode -format

2024-01-11 11:32:53,304 INFO common.Storage: Storage directory /data/hadoop/hadoop_repo/dfs/name has been successfully formatted.

启动 hdfs

[root@hadoop01 sbin]# pwd
/data/hadoop/soft/hadoop-3.3.6/sbin

# 报错，忘记配置 用户了
[root@hadoop01 sbin]# ./start-dfs.sh 
Starting namenodes on [hadoop01]
ERROR: Attempting to operate on hdfs namenode as root
ERROR: but there is no HDFS_NAMENODE_USER defined. Aborting operation.
Starting datanodes
ERROR: Attempting to operate on hdfs datanode as root
ERROR: but there is no HDFS_DATANODE_USER defined. Aborting operation.
Starting secondary namenodes [hadoop01]
ERROR: Attempting to operate on hdfs secondarynamenode as root
ERROR: but there is no HDFS_SECONDARYNAMENODE_USER defined. Aborting operation.
2024-01-11 11:35:01,599 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable


[root@hadoop01 sbin]# vi start-dfs.sh
[root@hadoop01 sbin]# vi stop-dfs.sh 
[root@hadoop01 sbin]# vi start-yarn.sh 
[root@hadoop01 sbin]# vi stop-yarn.sh 

# 重新分发

// start-dfs.sh，stop-dfs.sh，start-yarn.sh，stop-yarn.sh 写在最前面

HDFS_ZKFC_USER=root
HDFS_JOURNALNODE_USER=root
export HDFS_NAMENODE_USER=root
export HDFS_DATANODE_USER=root
export HDFS_SECONDARYNAMENODE_USER=root
export YARN_RESOURCEMANAGER_USER=root
export YARN_NODEMANAGER_USER=root

启动后的结果

[root@hadoop01 sbin]# ./start-dfs.sh 
Starting namenodes on [hadoop01]
上一次登录：四 1月 11 13:08:21 CST 2024从 10.35.232.75pts/2 上
/etc/profile: line 62: hadoop: command not found
Starting datanodes
上一次登录：四 1月 11 13:08:54 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
hadoop04: WARNING: /data/hadoop/soft/hadoop-3.3.6/logs does not exist. Creating.
hadoop03: WARNING: /data/hadoop/soft/hadoop-3.3.6/logs does not exist. Creating.
Starting secondary namenodes [hadoop01]
上一次登录：四 1月 11 13:08:56 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
2024-01-11 13:09:03,358 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable

[root@hadoop01 sbin]# jps
12433 NameNode
2422 nacos-server.jar
12823 SecondaryNameNode
19911 Jps

[root@hadoop02 ~]# jps
21780 Jps
[root@hadoop02 ~]# 

[root@hadoop03 ~]# jps
12853 DataNode
13033 Jps

[root@hadoop04 ~]# jps
5771 DataNode
5951 Jps

启动 yarn

[root@hadoop01 sbin]# ./start-dfs.sh 
Starting namenodes on [hadoop01]
上一次登录：四 1月 11 13:39:35 CST 2024从 10.35.232.75pts/2 上
/etc/profile: line 62: hadoop: command not found
Starting datanodes
上一次登录：四 1月 11 13:49:32 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
Starting secondary namenodes [hadoop01]
上一次登录：四 1月 11 13:49:34 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
2024-01-11 13:49:41,008 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
[root@hadoop01 sbin]# ./start-yarn.sh
Starting resourcemanager
上一次登录：四 1月 11 13:49:37 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
Starting nodemanagers
上一次登录：四 1月 11 13:50:01 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
[root@hadoop01 sbin]# 


[root@hadoop03 lib]# jps
16288 Jps
16007 NodeManager
15372 DataNode


[root@hadoop04 lib]# jps
9348 DataNode
10343 Jps
9995 NodeManager

http://10.xx.3x.142:9870

http://10.xx.3x.142:8088

验证

[root@hadoop02 ~]# hdfs dfs -mkdir  /test
[root@hadoop02 ~]# 

bug

hadoop一主三从安装 中就有这个问题
注意：这问题是因为生产上用的是 spark

按步就搬解决。记住重新分发一下。

zookeeper

在此只部署单机，生产上部署集群

[root@hadoop01 soft]# tar -zxvf zookeeper-3.4.14.tar.gz 
[root@hadoop01 soft]# ls
hadoop-3.3.6  hadoop-3.3.6.tar.gz  zookeeper-3.4.14  zookeeper-3.4.14.tar.gz

[root@hadoop01 zookeeper-3.4.14]# cd conf/
[root@hadoop01 conf]# ls
configuration.xsl  log4j.properties  zoo_sample.cfg
[root@hadoop01 conf]# mv zoo_sample.cfg  zoo.cfg

[root@hadoop01 conf]# vi zoo.cfg 
[root@hadoop01 conf]# 

dataDir=/data/hadoop/zk_repo

[root@hadoop01 bin]# pwd
/data/hadoop/soft/zookeeper-3.4.14/bin
[root@hadoop01 bin]#  ./zkServer.sh start
ZooKeeper JMX enabled by default
Using config: /data/hadoop/soft/zookeeper-3.4.14/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED

HA

core-site.xml

<property>
    <name>fs.defaultFS</name>
    <value>hdfs://mycluster</value>
</property>
    <!-- 是 hadoop文件系统依赖的基本配置,比如跑MapReduce时生成的临时路径本质上其实就是生成在它的下面 -->
<property>
    <name>hadoop.tmp.dir</name>
    <value>/data/hadoop/hadoop_repo</value>
</property>
<!-- 删除文件存在时间 -->
<property>
    <name>fs.trash.interval</name>
    <value>1440</value>
</property>
<property>
    <name>hadoop.proxyuser.root.hosts</name>
    <value>*</value>
</property>
<property>
    <name>hadoop.proxyuser.root.groups</name>
    <value>*</value>
</property>

 <property>
   <name>ha.zookeeper.quorum</name>
   <value>hadoop01:2181</value>
 </property>

hdfs-site.xml

注意 ： 三个journal节点都需要创建. (mkdir -p 执行启动命令)

 <property>
     <name>dfs.replication</name>
     <value>2</value>
 </property>

<! -- 上面已有 -->
<property>
  <name>dfs.nameservices</name>
  <value>mycluster</value>
</property>
<property>
  <name>dfs.ha.namenodes.mycluster</name>
  <value>nn1,nn2</value>
</property>
<property>
  <name>dfs.namenode.rpc-address.mycluster.nn1</name>
  <value>hadoop01:8020</value>
</property>
<property>
  <name>dfs.namenode.rpc-address.mycluster.nn2</name>
  <value>hadoop02:8020</value>
</property>
<property>
  <name>dfs.namenode.http-address.mycluster.nn1</name>
  <value>hadoop01:9870</value>
</property>
<property>
  <name>dfs.namenode.http-address.mycluster.nn2</name>
  <value>hadoop02:9870</value>
</property>
<property>
  <name>dfs.namenode.shared.edits.dir</name>
  <value>qjournal://hadoop01:8485;hadoop02:8485;hadoop03:8485/mycluster</value>
</property>

<property>
  <name>dfs.journalnode.edits.dir</name>
  <value>/data/hadoop/journal</value>
</property>
 <property>
   <name>dfs.ha.automatic-failover.enabled</name>
   <value>true</value>
 </property>

<property>
  <name>dfs.client.failover.proxy.provider.mycluster</name>
  <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
</property>
<property>
  <name>dfs.ha.nn.not-become-active-in-safemode</name>
  <value>true</value>
</property>
<property>
  <name>dfs.ha.fencing.methods</name>
  <value>sshfence</value>
</property>

<property>
  <name>dfs.ha.fencing.ssh.private-key-files</name>
  <value>/root/.ssh/id_rsa</value>
</property>
<property>
  <name>dfs.ha.fencing.ssh.connect-timeout</name>
  <value>30000</value>
</property>

[root@hadoop01 hadoop]# mkdir journal
[root@hadoop01 hadoop]# cd journal/
[root@hadoop01 journal]# ls
[root@hadoop01 journal]# pwd
/data/hadoop/journal

mkdir -p /data/hadoop/journal

改为配置分发

# 除了修改了的机器，其它的都删除，重新分发

cd /data/hadoop/soft/ && rm -rf hadoop-3.3.6/
scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop02:/data/hadoop/soft/hadoop-3.3.6
scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop03:/data/hadoop/soft/hadoop-3.3.6
scp -rq  /data/hadoop/soft/hadoop-3.3.6 hadoop04:/data/hadoop/soft/hadoop-3.3.6

cd /data/hadoop/soft/hadoop-3.3.6/etc/hadoop  &&  cat hdfs-site.xml 

执行

# 三个 journal 节点依次执行并检查 
[root@hadoop01 bin]# pwd
/data/hadoop/soft/hadoop-3.3.6/bin
[root@hadoop01 bin]# ./hdfs --daemon start journalnode

cd /data/hadoop/soft/hadoop-3.3.6/bin && ./hdfs --daemon start journalnode
 
[root@hadoop03 bin]# jps
10393 Jps
10175 JournalNode


# hadoop01上重新格式化
./hdfs namenode -format

# namenode 启动
cd /data/hadoop/soft/hadoop-3.3.6/bin && ./hdfs --daemon start  namenode
[root@hadoop01 bin]# ./hdfs --daemon start  namenode
[root@hadoop01 bin]# jps
24352 NameNode
20529 QuorumPeerMain
24515 Jps
2422 nacos-server.jar
32683 JournalNode

# namenode 备用节点上执行
[root@hadoop02 bin]# ./hdfs namenode -bootstrapStandby
2024-01-11 15:08:25,601 INFO common.Storage: Storage directory /data/hadoop/hadoop_repo/dfs/name has been successfully formatted.


# 回到 hadoop01节点上 zkfc -formatZK
[root@hadoop01 bin]# ./hdfs zkfc -formatZK
2024-01-11 15:10:29,614 INFO ha.ActiveStandbyElector: Session connected.
2024-01-11 15:10:29,629 INFO ha.ActiveStandbyElector: Successfully created /hadoop-ha/mycluster in ZK.


## 启动 hdfs 
[root@hadoop01 sbin]# ./start-dfs.sh
Starting namenodes on [hadoop01 hadoop02]
上一次登录：四 1月 11 15:38:41 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
hadoop02: namenode is running as process 1949.  Stop it first and ensure /tmp/hadoop-root-namenode.pid file is empty before retry.
hadoop01: namenode is running as process 10787.  Stop it first and ensure /tmp/hadoop-root-namenode.pid file is empty before retry.
Starting datanodes
上一次登录：四 1月 11 15:45:44 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
Starting journal nodes [hadoop03 hadoop02 hadoop01]
上一次登录：四 1月 11 15:45:44 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
hadoop03: journalnode is running as process 2861.  Stop it first and ensure /tmp/hadoop-root-journalnode.pid file is empty before retry.
hadoop01: journalnode is running as process 11453.  Stop it first and ensure /tmp/hadoop-root-journalnode.pid file is empty before retry.
hadoop02: journalnode is running as process 2263.  Stop it first and ensure /tmp/hadoop-root-journalnode.pid file is empty before retry.
2024-01-11 15:45:48,859 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Starting ZK Failover Controllers on NN hosts [hadoop01 hadoop02]
上一次登录：四 1月 11 15:45:48 CST 2024pts/0 上
/etc/profile: line 62: hadoop: command not found
hadoop02: zkfc is running as process 2518.  Stop it first and ensure /tmp/hadoop-root-zkfc.pid file is empty before retry.
hadoop01: zkfc is running as process 11880.  Stop it first and ensure /tmp/hadoop-root-zkfc.pid file is empty before retry.
[root@hadoop01 sbin]# jps
20529 QuorumPeerMain
3681 Jps
10787 NameNode
11880 DFSZKFailoverController
11453 JournalNode
[root@hadoop01 sbin]# 


[root@hadoop02 hadoop_repo]# jps
2518 DFSZKFailoverController
2263 JournalNode
1949 NameNode
18126 Jps


[root@hadoop03 hadoop]# jps
10263 DataNode
12170 Jps
2861 JournalNode

[root@hadoop04 hadoop]# jps
30245 DataNode
32366 Jps

验证 HA

验证：直接 kill 掉 active 所在节点的 NameNode ，观察 备用 节点是否自动转正。

[root@hadoop02 sbin]# jps
11411 DFSZKFailoverController
6342 JournalNode
11896 Jps
10649 NameNode
[root@hadoop02 sbin]# kill 10649
[root@hadoop02 sbin]# jps
11411 DFSZKFailoverController
6342 JournalNode
19782 Jps

结束

hadoop HA-QJM 至此就结束了，如有疑问，欢迎评论区留言。