hadoop3.1.1 HA高可用分布式集群安装部署

 

1、环境介绍

涉及到软件下载地址：https://pan.baidu.com/s/1hpcXUSJe85EsU9ara48MsQ

服务器：CentOS 6.8 其中：2 台 namenode、3 台 datanode

zookeeper集群地址：192.168.67.11:2181,192.168.67.12:2181

JDK：jdk-8u191-linux-x64.tar.gz

hadoop：hadoop-3.1.1.tar.gz

节点信息：

节点	IP	namenode	datanode	resourcemanager	journalnode
namenode1	192.168.67.101	√		√	√
namenode2	192.168.67.102	√		√	√
datanode1	192.168.67.103		√		√
datanode2	192.168.67.104		√		√
datanode3	192.168.67.105		√		√

2、配置ssh免密登陆

2.1 在每台机器上执行 ssh-keygen -t rsa

2.2 vim ~/.ssh/id_rsa.pub 将所有机器上的公钥内容汇总到 authorized_keys 文件并分发到每台机器上。

2.3 授权 chmod 600 ~/.ssh/authorized_keys

3、配置hosts： 

vim /etc/hosts

#增加如下配置
192.168.67.101 namenode1
192.168.67.102 namenode2
192.168.67.103 datanode1
192.168.67.104 datanode2
192.168.67.105 datanode3

#将hosts文件分发至其他机器
scp -r /etc/hosts namenode2:/etc/hosts
scp -r /etc/hosts datanode1:/etc/hosts
scp -r /etc/hosts datanode2:/etc/hosts
scp -r /etc/hosts datanode3:/etc/hosts

4、关闭防火墙

service iptables stop
chkconfig iptables off

5、安装JDK

tar -zxvf /usr/local/soft/jdk-8u191-linux-x64.tar.gz -C /usr/local/

vim /etc/profile

#增加JDK环境变量内容
export JAVA_HOME=/usr/local/jdk1.8.0_191
export JRE_HOME=${JAVA_HOME}/jre
export CLASSPATH=.:${JAVA_HOME}/lib:${JRE_HOME}/lib
export PATH=${JAVA_HOME}/bin:$PATH

使环境变量生效：source /etc/profile

 6、安装hadoop

tar -zxvf /usr/local/soft/hadoop-3.1.1.tar.gz -C /usr/local/
vim /etc/profile

#增加hadoop环境变量内容
export HADOOP_HOME=/usr/local/hadoop-3.1.1
export PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$HADOOP_HOME/lib

使环境变量生效：source /etc/profile

#修改 start-dfs.sh 和 stop-dfs.sh 两个文件，增加配置
vim /usr/local/hadoop-3.1.1/sbin/start-dfs.sh
vim /usr/local/hadoop-3.1.1/sbin/stop-dfs.sh

#增加启动用户
HDFS_DATANODE_USER=root
HDFS_DATANODE_SECURE_USER=root
HDFS_NAMENODE_USER=root
HDFS_SECONDARYNAMENODE_USER=root
HDFS_JOURNALNODE_USER=root
HDFS_ZKFC_USER=root

#修改 start-yarn.sh 和 stop-yarn.sh 两个文件，增加配置
vim /usr/local/hadoop-3.1.1/sbin/start-yarn.sh
vim /usr/local/hadoop-3.1.1/sbin/stop-yarn.sh

#增加启动用户
YARN_RESOURCEMANAGER_USER=root
HDFS_DATANODE_SECURE_USER=root
YARN_NODEMANAGER_USER=root

vim /usr/local/hadoop-3.1.1/etc/hadoop/hadoop-env.sh

#增加内容
export JAVA_HOME=/usr/local/jdk1.8.0_191
export HADOOP_HOME=/usr/local/hadoop-3.1.1

#修改 workers 文件内容
vim /usr/local/hadoop-3.1.1/etc/hadoop/workers

#替换内容为
datanode1
datanode2
datanode3

vim /usr/local/hadoop-3.1.1/etc/hadoop/core-site.xml

#修改为如下配置
<configuration>
    <!-- 指定hdfs的nameservice为nameservice -->
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://mycluster/</value>
    </property>

    <!-- 指定hadoop临时目录 -->
    <property>
        <name>hadoop.tmp.dir</name>
        <value>file:/usr/local/hadoop-3.1.1/hdfs/temp</value> 
    </property>

    <!-- 指定zookeeper地址 -->
    <property>
        <name>ha.zookeeper.quorum</name>
        <value>192.168.67.1:2181</value>
    </property>
</configuration>

vim /usr/local/hadoop-3.1.1/etc/hadoop/hdfs-site.xml

#修改为如下配置
<configuration>
    <property>
        <name>dfs.namenode.name.dir</name>
        <value>file:/usr/local/hadoop-3.1.1/hdfs/name</value>
    </property>
    
    <property>
        <name>dfs.datanode.data.dir</name>
        <value>file:/usr/local/hadoop-3.1.1/hdfs/data</value>
    </property>
    
    <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>namenode1:9000</value>
    </property>
    
    <property>
        <name>dfs.namenode.rpc-address.mycluster.nn2</name>
        <value>namenode2:9000</value>
    </property>
    
    <property>
        <name>dfs.namenode.http-address.mycluster.nn1</name>
        <value>namenode1:50070</value>
    </property>
    
    <property>
        <name>dfs.namenode.http-address.mycluster.nn2</name>
        <value>namenode2:50070</value>
    </property>
    
    <!--HA故障切换 -->
    <property>
        <name>dfs.ha.automatic-failover.enabled</name>
        <value>true</value>
    </property>
    
    <!-- journalnode 配置 -->
    <property>
        <name>dfs.namenode.shared.edits.dir</name>
        <value>qjournal://namenode1:8485;namenode2:8485;datanode1:8485;datanode2:8485;datanode3:8485/mycluster</value>
    </property>
    
    <property>
        <name>dfs.client.failover.proxy.provider.mycluster</name>
        <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
    </property>
    
    <!--发生failover时，Standby的节点要执行一系列方法把原来那个Active节点中不健康的NameNode服务给杀掉，
    这个叫做fence过程。sshfence会通过ssh远程调用fuser命令去找到Active节点的NameNode服务并杀死它-->
    <property>
        <name>dfs.ha.fencing.methods</name>
        <value>shell(/bin/true)</value>
    </property>
    
    <!--SSH私钥 -->
    <property>
        <name>dfs.ha.fencing.ssh.private-key-files</name>
        <value>/root/.ssh/id_rsa</value>
    </property>
    
    <!--SSH超时时间 -->
    <property>
        <name>dfs.ha.fencing.ssh.connect-timeout</name>
        <value>30000</value>
    </property>
    
    <!--Journal Node文件存储地址 -->
    <property>
        <name>dfs.journalnode.edits.dir</name>
        <value>/usr/local/hadoop-3.1.1/hdfs/journaldata</value>
    </property>
    
    <property>
        <name>dfs.qjournal.write-txns.timeout.ms</name>
        <value>60000</value>
    </property>
</configuration>

vim /usr/local/hadoop-3.1.1/etc/hadoop/mapred-site.xml

#修改为如下配置
<configuration>
    <!-- 指定mr框架为yarn方式 -->
    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
    </property>
</configuration>

vim /usr/local/hadoop-3.1.1/etc/hadoop/yarn-site.xml

#修改为如下配置
<configuration>
    <!-- Site specific YARN configuration properties -->
    <!-- 开启RM高可用 -->
    <property>
        <name>yarn.resourcemanager.ha.enabled</name>
        <value>true</value>
    </property>

    <!-- 指定RM的cluster id -->
    <property>
        <name>yarn.resourcemanager.cluster-id</name>
        <value>yrc</value>
    </property>

    <!-- 指定RM的名字 -->
    <property>
        <name>yarn.resourcemanager.ha.rm-ids</name>
        <value>rm1,rm2</value>
    </property>

    <!-- 分别指定RM的地址 -->
    <property>
        <name>yarn.resourcemanager.hostname.rm1</name>
        <value>namenode1</value>
    </property>

    <property>
        <name>yarn.resourcemanager.hostname.rm2</name>
        <value>namenode2</value>
    </property>

    <!-- 指定zk集群地址 -->
    <property>
        <name>yarn.resourcemanager.zk-address</name>
        <value>192.168.67.11:2181,192.168.67.12:2181</value>
    </property>

    <property>
        <name>yarn.nodemanager.aux-services</name>
        <value>mapreduce_shuffle</value>
    </property>
</configuration>

#将这些修改的文件分发至其他4台服务器中
/usr/local/hadoop-3.1.1/sbin/start-dfs.sh
/usr/local/hadoop-3.1.1/sbin/stop-dfs.sh
/usr/local/hadoop-3.1.1/sbin/start-yarn.sh
/usr/local/hadoop-3.1.1/sbin/stop-yarn.sh
/usr/local/hadoop-3.1.1/etc/hadoop/hadoop-env.sh
/usr/local/hadoop-3.1.1/etc/hadoop/workers
/usr/local/hadoop-3.1.1/etc/hadoop/core-site.xml
/usr/local/hadoop-3.1.1/etc/hadoop/hdfs-site.xml
/usr/local/hadoop-3.1.1/etc/hadoop/mapred-site.xml
/usr/local/hadoop-3.1.1/etc/hadoop/yarn-site.xml

首次启动顺序
1、确保配置的zookeeper服务器已经运行
2、在所有journalnode机器上启动：hdfs --daemon start journalnode
3、namenode1中执行格式化zkfc：hdfs zkfc -formatZK
4、namenode1中格式化主节点：hdfs namenode -format
5、启动namenode1中的主节点：hdfs --daemon start namenode
6、namenode2副节点同步主节点格式化：hdfs namenode -bootstrapStandby
7、启动集群：start-all.sh

7、验证

7.1 访问地址：

http://192.168.67.101/50070/

http://192.168.67.102/50070/

http://192.168.67.101/8088/

http://192.168.67.102/8088/

7.2 关闭 namenode 为 active 对应的服务器，观察另一台 namenode 状态是否由 standby 变更为 active