Hadoop的HA(ZooKeeper)安装与部署

非HA的安装步骤

https://www.cnblogs.com/live41/p/15467263.html

一、部署设定

1.服务器

c1   192.168.100.105    zk、namenode、zkfc、datanode、nodemanager
c2   192.168.100.110    zk、namenode、zkfc、datanode、nodemanager、resourcemanager、journalnode
c3   192.168.100.115    zk、datanode、nodemanager、resourcemanager、journalnode
c4   192.168.100.120    zk、datanode、nodemanager、resourcemanager、journalnode

2.目录

Hadoop NameNode目录    /home/hadoop/hdfs/name
Hadoop DataNode目录      /home/hadoop/hdfs/data
Hadoop JournalNode目录  /home/hadoop/hdfs/journal
Hadoop临时目录                /home/hadoop/tmp
ZooKeeper数据目录           /home/zookeeper/data/
ZooKeeper日志目录           /home/zookeeper/logs/

* 以上目录在每台机都要手动创建

二、系统配置

* 每台机都要执行

https://www.cnblogs.com/live41/p/15467263.html

执行这篇文章的第二点的步骤

三、下载安装包及配置系统环境变量

* 每台机都要执行

1.下载及解压

https://downloads.apache.org/hadoop/common/

http://zookeeper.apache.org/releases.html

解压到/home/目录

2.配置环境变量

vim ~/.bashrc

在末尾加入以下内容：

export HADOOP_HOME=/home/hadoop
export PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:/home/zookeeper/bin

export HADOOP_COMMON_HOME=/home/hadoop
export HADOOP_HDFS_HOME=/home/hadoop
export HADOOP_MAPRED_HOME=/home/hadoop
export HADOOP_CONF_DIR=/home/hadoop/etc/hadoop

export HDFS_DATANODE_USER=root
export HDFS_NAMENODE_USER=root
export HDFS_SECONDARYNAMENODE_USER=root

export YARN_HOME=/home/hadoop
export YARN_RESOURCEMANAGER_USER=root
export YARN_NODEMANAGER_USER=root

更新环境变量

source ~/.bashrc

四、安装和配置ZooKeeper

* 每台机都要执行

1.创建data和logs目录

mkdir -pv /home/zookeeper/{data,logs} 

2.编辑zoo.cfg

cd /home/zookeeper/conf
cp zoo_sample.cfg zoo.cfg
vim zoo.cfg

修改属性或添加内容：

tickTime=2000
initLimit=10
syncLimit=5
clientPort=2181

dataDir=/home/zookeeper/data/
dataLogDir=/home/zookeeper/logs

server.1=c1:2888:3888
server.2=c2:2888:3888
server.3=c3:2888:3888
server.4=c4:2888:3888

3.创建myid

echo 1 > /home/zookeeper/data/myid

* 其中数字是每台机不同，例如第1台是1，第2台是2，第3台是3，第4台是4.

4.启动ZooKeeper节点

zkServer.sh start

5.测试状态

zkServer.sh status

不同节点会不同，有1个leader和多个follower

五、安装和配置Hadoop

* 不用每台机执行，只在c1机执行，再通过scp命令同步配置文件给其它机

1.进入配置目录

cd $HADOOP_HOME/etc/hadoop

2.编辑hadoop-env.sh

vim hadoop-env.sh

添加以下行（已有的不用改）：

export JAVA_HOME=/usr/bin/jdk1.8.0
export HADOOP_OS_TYPE=${HADOOP_OS_TYPE:-$(uname -s)}

* 这里的JAVA_HOME根据你安装的路径来修改。

3.编辑core-site.xml

<configuration>
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://ns6/</value> <!--该属性对应的是hdfs-site.xml的dfs.nameservices属性-->
    </property>
    <property>
        <name>hadoop.tmp.dir</name>
        <value>/home/hadoop/tmp</value>
    </property>
    <property>
        <name>io.file.buffer.size</name>
        <value>131072</value>
    </property>
    <property>
        <name>ha.zookeeper.quorum</name>
        <value>c1:2181,c2:2181,c3:2181,c4:2181</value>
    </property>
    <property>
        <name>ha.zookeeper.session-timeout.ms</name>
        <value>1000</value>
    </property>
</configuration>

4.编辑hdfs-site.xml

<configuration>
    <property>
        <name>dfs.replication</name>
        <value>3</value>
    </property>
    <property>
        <name>dfs.namenode.name.dir</name>
        <value>file:///home/hadoop/hdfs/name</value>
    </property>
    <property>
        <name>dfs.datanode.data.dir</name>
        <value>file:///home/hadoop/hdfs/data</value>
    </property>
    <property>
        <name>dfs.webhdfs.enabled</name>
        <value>true</value>
    </property>

    <!-- core-site.xml中使用的是这里的配置值 -->
    <property>
        <name>dfs.nameservices</name>
        <value>ns6</value>
    </property>

    <property>
        <name>dfs.ha.namenodes.ns6</name>
        <value>nn1,nn2</value>
    </property>

    <property>
        <name>dfs.namenode.rpc-address.ns6.nn1</name>
        <value>c1:9000</value>
    </property>

    <property>
        <name>dfs.namenode.http-address.ns6.nn1</name>
        <value>c1:50070</value>
    </property>

    <property>
        <name>dfs.namenode.rpc-address.ns6.nn2</name>
        <value>c2:9000</value>
    </property>
    <property>
        <name>dfs.namenode.http-address.ns6.nn2</name>
        <value>c2:50070</value>
    </property>

    <!-- 就是JournalNode列表，url格式：
        qjournal://host1:port1;host2:port2;host3:port3/journalId
        journalId推荐使用nameservice，默认端口号是：8485 -->
    <property>
        <name>dfs.namenode.shared.edits.dir</name>
        <value>qjournal://c1:8485;c2:8485;c3:8485;c4:8485/ns6</value>
    </property>
    <property>
        <name>dfs.journalnode.edits.dir</name>
        <value>/home/hadoop/hdfs/journal</value>
    </property>
    <property>
        <name>dfs.ha.automatic-failover.enabled</name>
        <value>true</value>
    </property>
    <property>
        <name>dfs.client.failover.proxy.provider.ns6</name>
        <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
    </property>
    <property>
        <name>dfs.ha.fencing.methods</name>
        <value>
            sshfence
            shell(/bin/true)
        </value>
    </property>
    <property>
        <name>dfs.ha.fencing.ssh.private-key-files</name>
        <value>~/.ssh/id_rsa</value> <!-- 或/root/.ssh/id_rsa -->
    </property>
    <property>
        <name>dfs.ha.fencing.ssh.connect-timeout</name>
        <value>30000</value>
    </property>
    <property>
        <name>ha.failover-controller.cli-check.rpc-timeout.ms</name>
        <value>60000</value>
    </property>
</configuration>

5.编辑mapred-site.xml

<configuration>
    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
    </property>
    <property>
        <name>mapreduce.jobtracker.http.address</name>
        <value>c1:50030</value>
    </property>
    <property>
        <name>mapreduce.jobhistory.address</name>
        <value>c1:10020</value>
    </property>
    <property>
        <name>mapreduce.jobhistory.webapp.address</name>
        <value>c1:19888</value>
    </property>
    <property>
        <name>mapred.job.tracker</name>
        <value>http://c1:9001</value>
    </property>
</configuration>

6.编辑yarn-site.xml

<configuration>
    <property>
        <name>yarn.resourcemanager.ha.enabled</name>
        <value>true</value>
    </property>

    <property>
        <name>yarn.resourcemanager.cluster-id</name>
        <value>yn6</value> <!-- 可自行定义cluster-id -->
    </property>

    <property>
        <name>yarn.resourcemanager.ha.rm-ids</name>
        <value>rm1,rm2</value>
    </property>

    <property>
        <name>yarn.resourcemanager.hostname.rm1</name>
        <value>c1</value>
    </property>
    <property>
        <name>yarn.resourcemanager.hostname.rm2</name>
        <value>c2</value>
    </property>
    <property>
        <name>yarn.resourcemanager.webapp.address.rm1</name>
        <value>c1:8088</value>
    </property>
    <property>
        <name>yarn.resourcemanager.webapp.address.rm2</name>
        <value>c2:8088</value>
    </property>

    <property>
        <name>yarn.nodemanager.aux-services</name>
        <value>mapreduce_shuffle</value>
    </property>
    <property>
        <name>yarn.log-aggregation-enable</name>
        <value>true</value>
    </property>
    <property>
        <name>yarn.log-aggregation.retain-seconds</name>
        <value>86400</value>
    </property>

    <property>
        <name>yarn.resourcemanager.zk-address</name>
        <value>c1:2181,c2:2181,c3:2181,c4:2181</value>
    </property>
    <property>
        <name>yarn.resourcemanager.recovery.enabled</name>
        <value>true</value>
    </property>
    <property>
        <name>yarn.resourcemanager.store.class</name>
        <value>org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore</value>
    </property>

    <property>
        <name>yarn.application.classpath</name>
        <value>/home/hadoop/etc/hadoop:/home/hadoop/share/hadoop/common/lib/*:/home/hadoop/share/hadoop/common/*:/home/hadoop/share/hadoop/hdfs:/home/hadoop/share/hadoop/hdfs/lib/*:/home/hadoop/share/hadoop/hdfs/*:/home/hadoop/share/hadoop/mapreduce/*:/home/hadoop/share/hadoop/yarn:/home/hadoop/share/hadoop/yarn/lib/*:/home/hadoop/share/hadoop/yarn/*</value>
    </property>
</configuration>

7.配置workers

vim workers

添加以下内容：

c1
c2
c3
c4

8.同步配置文件

由于前面已经配置过免密登录，所以可以先配置c1机的文件，再通过scp同步给其它机。

cd /home/hadoop/etc/hadoop
scp *.* c2:/home/hadoop/etc/hadoop
scp *.* c3:/home/hadoop/etc/hadoop
scp *.* c4:/home/hadoop/etc/hadoop

六、启动Hadoop

* 只在c1机执行

1.启动JournalNode

hdfs --workers --daemon start journalnode

旧版命令如下：

hadoop-daemons.sh start journalnode

旧版命令也能执行，但会报警告信息。

* 有2个相似文件：hadoop-daemon.sh和hadoop-daemons.sh

* 前者(没带s)的是只执行本机的journalnode，后者（有带s）是执行所有机器的journalnode。

* 注意不要输错～！

2.格式化NameNode

hadoop namenode -format

格式化后同步namenode的信息文件给c2机。因为有2个NameNode节点，c1和c2

scp -r /home/hadoop/hdfs/name/current/ c2:/home/hadoop/hdfs/name/

3.格式化zkfc

hdfs zkfc -formatZK

zkfc = ZKFailoverController = ZooKeeper Failover Controller

zkfc用于监控NameNode状态信息，并进行自动切换。

4.启动HDFS和Yarn

start-dfs.sh
start-yarn.sh

* 这步是最经常报错的，部分错误及解决方法见下面附录。

5.检查进程

jps

6.检查节点状态

hdfs haadmin -getServiceState nn1
hdfs haadmin -getServiceState nn2
yarn rmadmin -getServiceState rm1
yarn rmadmin -getServiceState rm2

7.关闭Hadoop

stop-yarn.sh
stop-dfs.sh

七、使用Hadoop

* 只在c1机执行

1.Web页面

http://192.168.100.105:50070/

2.使用命令

https://www.cnblogs.com/xiaojianblogs/p/14281445.html

附录

* 每台机都要执行

1.没有配置用root启动Hadoop

错误提示信息：

Starting namenodes on [master]
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 [slave1]
ERROR: Attempting to operate on hdfs secondarynamenode as root
ERROR: but there is no HDFS_SECONDARYNAMENODE_USER defined. Aborting operation.

解决方法：

(1) 修改start-dfs.sh和stop-dfs.sh，在头部添加以下内容：

#!/usr/bin/env bash
HDFS_DATANODE_USER=root
HADOOP_SECURE_DN_USER=root
HDFS_NAMENODE_USER=root
HDFS_SECONDARYNAMENODE_USER=root

(2) 修改start-yarn.sh和stop-yarn.sh，在头部添加以下内容：

#!/usr/bin/env bash
HADOOP_SECURE_DN_USER=root
YARN_RESOURCEMANAGER_USER=root
YARN_NODEMANAGER_USER=root

* 可以先在c1机修改，用scp同步给其它机器

2.重新格式化后的IO错误

错误提示信息（多种）：

Incompatible clusterIDs in /home/hadoop/hdfs/data

Failed to add storage directory [DISK]file

Directory /home/hadoop/hdfs/journal/ns6 is in an inconsistent state: Can't format the storage directory because the current directory is not empty.

解决方法：

在格式化之前，先把name、data、logs等目录里面的文件先清除掉。注意只是删除里面的文件，不删除目录