hadoop 3.2.x 高可用集群搭建

2020-02-14 本文已影响0人陈sir的知识图谱

本环境使用centos 8

HOSTNAME	作用	IP	role
hadoop301	hdfs namenode,hdfs datanode, yarn resourcemanager, yarn node manager, journal node, zookeeeper	192.168.142.101	yarn and hdfs master, worker
hadoop302	hdfs namenode, hdfs datanode, yarn resourcemanager, yarn node manager, journal node, zookeeeper	192.168.142.102	yarn and hdfs master, worker
hadoop303	hdfs namenode, hdfs datanode, yarn resourcemanager, journal node, zookeeeper	192.168.142.103	hdfs master , worker

前置条件

先设置好 linux 通用设置
配置好三台机器之间的免密码登陆
下载好hadoop-3.2.1，zookeeper-3.5.6 的压缩包

安装步骤

1 在所有机器上执行

安装jdk

1.1. 安装jdk
yum install -y java-1.8.0-openjdk-devel.x86_64 java-1.8.0-openjdk.x86_64
1.2 编辑/etc/profile, 在末尾添加

export HADOOP_HOME=/opt/hadoop-3.2.1
export PATH=$PATH:$HADOOP_HOME/bin
export PATH=$PATH:$HADOOP_HOME/sbin
export JAVA_HOME=/usr/lib/jvm/jre-1.8.0
export ZOOKEEPER_HOME=/opt/zookeeper-3.5.6
export PATH=$PATH:$ZOOKEEPER_HOME/bin

1.3 创建文件夹,至于为什么是这些路径，和后面的配置文件有关。

mkdir -p /tmp/hadoop/tmpdir
mkdir -p /tmp/hadoop/journalnode/data
mkdir -p /tmp/hadoop/hdfs/namenode
mkdir -p /tmp/hadoop/hdfs/datanode
mkdir -p /tmp/zookeeper
echo > 1 /tmp/zookeeper/myid #hadoop301
echo > 2 /tmp/zookeeper/myid #hadoop302
echo > 3 /tmp/zookeeper/myid #hadoop303

1.4 设置好hosts

192.168.142.101 hadoop301
192.168.142.102 hadoop302
192.168.142.103 hadoop303

2 在 hadoop301 上执行

2.1安装ZK

2.1.1 将zk 解压

tar -zxf zookeeper-3.5.6.tar.gz

2.1.2 配置zoo.cfg

cd zookeeper-3.5.6/conf
mv zoo_sample.cfg zoo.cfg
vim zoo.cfg

zoo.cfg 内容如下

# The number of milliseconds of each tick 心跳基本时间单位，毫秒级，ZK基本上所有的时间都是这个时间的整数倍。
tickTime=2000
# The number of ticks that the initial 
# synchronization phase can take 
# tickTime的个数，表示在leader选举结束后，followers与leader同步需要的时间，如果followers比较多或者说leader的数据灰常多时，同步时间相应可能会增加，那么这个值也需要相应增加。当然，这个值也是follower和observer在开始同步leader的数据时的最大等待时间(setSoTimeout)
initLimit=10
# The number of ticks that can pass between 
# sending a request and getting an acknowledgement
# tickTime的个数，这时间容易和上面的时间混淆，它也表示follower和observer与leader交互时的最大等待时间，只不过是在与leader同步完毕之后，进入正常请求转发或ping等消息交互时的超时时间
syncLimit=5
# the directory where the snapshot is stored.
# do not use /tmp for storage, /tmp here is just 
# example sakes.
# 内存数据库快照存放地址，如果没有指定事务日志存放地址(dataLogDir)，默认也是存放在这个路径下，建议两个地址分开存放到不同的设备上
dataDir=/tmp/zookeeper
# the port at which the clients will connect
clientPort=2181
# the maximum number of client connections.
# increase this if you need to handle more clients
#maxClientCnxns=60
#
# Be sure to read the maintenance section of the 
# administrator guide before turning on autopurge.
#
# http://zookeeper.apache.org/doc/current/zookeeperAdmin.html#sc_maintenance
#
# The number of snapshots to retain in dataDir
#autopurge.snapRetainCount=3
# Purge task interval in hours
# Set to "0" to disable auto purge feature
#autopurge.purgeInterval=1
server.1=hadoop301:2888:3888
server.2=hadoop302:2888:3888
server.3=hadoop303:2888:3888

2.1.3 同步到 hadoop302 hadoop303

yum install -y rsync
rsync -auvp /opt/zookeeper-3.5.6 root@hadoop302:/opt 
rsync -auvp /opt/zookeeper-3.5.6 root@hadoop303:/opt

2.2 安装 hadoop

2.2.1 将 hadoop-3.2.1.tar.gz 上传到/opt 目录下, 并解压
2.2.2 编辑 /opt/hadoop-3.2.1/etc/hadoop/core-site.xml

<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
<configuration>
  <property>
    <name>fs.defaultFS</name>
    <value>hdfs://mycluster</value>
  </property>
  <property>
    <name>hadoop.tmp.dir</name>
    <value>/tmp/hadoop/tmpdir</value>
  </property>
  <property>
    <name>ha.zookeeper.quorum</name>
    <value>hadoop301:2181,hadoop302:2181,hadoop303:2181</value>
  </property>
</configuration>

2.2.3 /opt/hadoop-3.2.1/etc/hadoop/hdfs-site.xml

<?xml version="1.0" encoding="UTF-8"?>
<?xml-stylesheet type="text/xsl" href="configuration.xsl"?>

<configuration>
  <!-- hdfs HA configuration-->
  <!-- all default configuration can be found at https://hadoop.apache.org/docs/stable|<can be a version liek r3.2.1></can>/hadoop-project-dist/hadoop-hdfs//hdfs-default.xml -->
  
  <property>
    <name>dfs.ha.automatic-failover.enabled</name>
    <value>true</value>
  </property>
  <!-- dfs.nameservices 这里需要与core-site.xml 中fs.defaultFS 的名称一致-->
  <property>
    <name>dfs.nameservices</name>
    <value>mycluster</value>
  </property>
  <!-- 定义集群中 namenode 列表，这里定义了三个namenode，分别是nn1,nn2,nn3-->
  <property>
    <name>dfs.ha.namenodes.mycluster</name>
    <value>nn1,nn2,nn3</value>
  </property>
  <!-- namenode nn1的具体定义，这里要和 dfs.ha.namenodes.mycluster 定义的列表对应 -->
  <property>
    <name>dfs.namenode.rpc-address.mycluster.nn1</name>
    <value>hadoop301:8020</value>
  </property>
  <property>
    <name>dfs.namenode.rpc-address.mycluster.nn2</name>
    <value>hadoop302:8020</value>
  </property>
  <property>
    <name>dfs.namenode.rpc-address.mycluster.nn3</name>
    <value>hadoop303:8020</value>
  </property>
  <!-- namenode nn1的具体定义，这里要和 dfs.ha.namenodes.mycluster 定义的列表对应 -->
  <property>
    <name>dfs.namenode.http-address.mycluster.nn1</name>
    <value>hadoop301:9870</value>
  </property>
  <property>
    <name>dfs.namenode.http-address.mycluster.nn2</name>
    <value>hadoop302:9870</value>
  </property>
  <property>
    <name>dfs.namenode.http-address.mycluster.nn3</name>
    <value>hadoop303:9870</value>
  </property>
<!-- 指定NameNode的元数据在JournalNode上的存放位置 -->
  <property>
    <name>dfs.namenode.shared.edits.dir</name>
    <value>qjournal://hadoop301:8485;hadoop302:8485;hadoop303:8485/mycluster</value>
  </property>
  <!-- 指定JournalNode在本地磁盘存放数据的位置 -->
  <property>
    <name>dfs.journalnode.edits.dir</name>
    <value>/tmp/hadoop/journalnode/data</value>
  </property>
  <!-- 配置失败自动切换实现方式 -->
  <property>
    <name>dfs.client.failover.proxy.provider.mycluster</name>
    <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
  </property>
  <!-- 配置隔离机制方法，多个机制用换行分割，即每个机制暂用一行-->
  <property>
    <name>dfs.ha.fencing.methods</name>
    <value>sshfence</value>
  </property>
  <!-- 使用sshfence隔离机制时需要ssh免登陆 -->
  <property>
    <name>dfs.ha.fencing.ssh.private-key-files</name>
    <value>/root/.ssh/id_rsa</value>
  </property>
  <!-- 配置sshfence隔离机制超时时间 -->
  <property>
    <name>dfs.ha.fencing.ssh.connect-timeout</name>
    <value>30000</value>
  </property>
  <property>
    <name>dfs.journalnode.http-address</name>
    <value>0.0.0.0:8480</value>
  </property>
  <property>
    <name>dfs.journalnode.rpc-address</name>
    <value>0.0.0.0:8485</value>
  </property>
  <!-- hdfs HA configuration end-->

  <property>
    <name>dfs.replication</name>
    <value>1</value>
  </property>
  <property>
    <name>dfs.namenode.name.dir</name>
    <value>/tmp/hadoop/hdfs/namenode</value>
  </property>
  <property>
    <name>dfs.datanode.data.dir</name>
    <value>/tmp/hadoop/hdfs/datanode</value>
  </property>
  <!--开启webhdfs接口访问-->
  <property>
    <name>dfs.webhdfs.enabled</name>
    <value>true</value>
  </property>
<!-- 关闭权限验证，hive可以直连 -->
  <property>
    <name>dfs.permissions.enabled</name>
    <value>false</value>
  </property>
</configuration>

2.2.4 编辑 /opt/hadoop-3.2.1/etc/hadoop/yarn-site.xml

<?xml version="1.0"?>
<configuration>

  <!-- yarn ha configuration-->
  <property>
    <name>yarn.resourcemanager.ha.enabled</name>
    <value>true</value>
  </property>
  <!-- 定义集群名称 -->
  <property>
    <name>yarn.resourcemanager.cluster-id</name>
    <value>cluster1</value>
  </property>
  <!-- 定义本机在在高可用集群中的id 要与 yarn.resourcemanager.ha.rm-ids 定义的值对应，如果不作为resource manager 则删除这项配置。-->
  <property>
    <name>yarn.resourcemanager.ha.id</name>
    <value>rm1</value>
  </property>
  <!-- 定义高可用集群中的 id 列表 -->
  <property>
    <name>yarn.resourcemanager.ha.rm-ids</name>
    <value>rm1,rm2</value>
  </property>
  <!-- 定义高可用RM集群具体是哪些机器 -->
  <property>
    <name>yarn.resourcemanager.hostname.rm1</name>
    <value>hadoop301</value>
  </property>
  <property>
    <name>yarn.resourcemanager.hostname.rm2</name>
    <value>hadoop302</value>
  </property>
  <property>
    <name>yarn.resourcemanager.webapp.address.rm1</name>
    <value>hadoop301:8088</value>
  </property>
  <property>
    <name>yarn.resourcemanager.webapp.address.rm2</name>
    <value>hadoop302:8088</value>
  </property>
  <property>
    <name>hadoop.zk.address</name>
    <value>hadoop301:2181,hadoop302:2181,hadoop303:2181</value>
  </property>

  <!-- Site specific YARN configuration properties -->
  <property>
    <name>yarn.nodemanager.aux-services</name>
    <value>mapreduce_shuffle</value>
  </property>
</configuration>

2.2.5 编辑 /opt/hadoop-3.2.1/etc/hadoop/mapred-site.xml

<?xml version="1.0"?>
<?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
<configuration>
  <property>
    <name>mapreduce.framework.name</name>
    <value>yarn</value>
  </property>
  <property>
    <name>mapreduce.application.classpath</name>
    <value>  
        /opt/hadoop-3.2.1/share/hadoop/common/*,
        /opt/hadoop-3.2.1/share/hadoop/common/lib/*,
        /opt/hadoop-3.2.1/share/hadoop/hdfs/*,
        /opt/hadoop-3.2.1/share/hadoop/hdfs/lib/*,
        /opt/hadoop-3.2.1/share/hadoop/mapreduce/*,
        /opt/hadoop-3.2.1/share/hadoop/mapreduce/lib/*,
        /opt/hadoop-3.2.1/share/hadoop/yarn/*,
        /opt/hadoop-3.2.1/share/hadoop/yarn/lib/*
    </value>
  </property>

</configuration>

2.2.6 编辑 /opt/hadoop-3.2.1/etc/hadoop/hadoop-env.sh

# The java implementation to use. By default, this environment
# variable is REQUIRED on ALL platforms except OS X!
# export JAVA_HOME=
export JAVA_HOME=/usr/lib/jvm/jre-1.8.0

# Some parts of the shell code may do special things dependent upon
# the operating system.  We have to set this here. See the next
# section as to why....
export HADOOP_OS_TYPE=${HADOOP_OS_TYPE:-$(uname -s)}
export HADOOP_PID_DIR=/opt/hadoop-3.2.1/pid
export HADOOP_LOG_DIR=/var/log/hadoop

2.2.7 编辑 /opt/hadoop-3.2.1/etc/hadoop/yarn-env.sh

# Specify the max heapsize for the ResourceManager.  If no units are
# given, it will be assumed to be in MB.
# This value will be overridden by an Xmx setting specified in either
# HADOOP_OPTS and/or YARN_RESOURCEMANAGER_OPTS.
# Default is the same as HADOOP_HEAPSIZE_MAX
#export YARN_RESOURCEMANAGER_HEAPSIZE=
export JAVA_HOME=/usr/lib/jvm/jre-1.8.0

2.2.7 编辑 /opt/hadoop-3.2.1/sbin/start-dfs.sh, /opt/hadoop-3.2.1/sbin/stop-dfs.sh，在脚本开始添加

HDFS_NAMENODE_USER=root
HDFS_DATANODE_USER=root
HDFS_JOURNALNODE_USER=root
HDFS_ZKFC_USER=root

2.2.8 编辑 /opt/hadoop-3.2.1/sbin/start-yarn.sh, /opt/hadoop-3.2.1/sbin/stop-yarn.sh，在脚本开始添加

YARN_RESOURCEMANAGER_USER=root
YARN_NODEMANAGER_USER=root

2.2.9 修改/opt/hadoop-3.2.1/etc/hadoo/workers 为如下内容

hadoop301
hadoop302
hadoop303

2.2.10 拷贝 hadoop-3.2.1 到 hadoop302 hadoop303
rsync -auvp /opt/hadoop-3.2.1 root@hadoop302:/opt
rsync -auvp /opt/hadoop-3.2.1 root@hadoop303:/opt

3 在hadoop302 上执行

需要修改yarn-site.xml的yarn.resourcemanager.ha.id，改为如下内容

  <property>
    <name>yarn.resourcemanager.ha.id</name>
    <value>rm2</value>
  </property>

4 在hadoop303上执行

删除如下property

  <property>
    <name>yarn.resourcemanager.ha.id</name>
    <value>rm1</value>
  </property>

5 启动

启动顺序 Zookeeper->JournalNode->格式化NameNode->创建命名空间zkfs->NameNode->Datanode->ResourceManager->NodeManager

5.1 启动zookeeper

在所有机器行上执行，顺序 hadoop301 hadoop302 hadoop303

# 注意，如果使用zsh 需要切换回bash
#chsh -s /usr/bin/bash
#如果想用zsh 直接执行，需要使用如下领命,emualte 命令必须安装 oh my zsh 才有。
# emulate sh -c '/opt/zookepper-3.5.6/bin/zkServer.sh start'
/opt/zookepper-3.5.6/bin/zkServer.sh start 
/opt/zookepper-3.5.6/bin/zkServer.sh status

5.2 启动journalnode

在所有机器行上执行，顺序 hadoop301 hadoop302 hadoop303

# 注意，如果使用zsh 需要切换回bash
#chsh -s /usr/bin/bash
/opt/hadoop-3.2.1/sbin/hadoop-daemon.sh start journalnode
# 或者通过 /opt/hadoop-3.2.1/bin/hdfs --daemon start journalnode

5.3 格式化 Namenode

在hadoop301上执行

# 注意，如果使用zsh 需要切换回bash
#chsh -s /usr/bin/bash
/opt/hadoop-3.2.1/bin/hadoop namenode -format
# 同步格式化之后的元数据到其他namenode，不然可能起不来
rsync -auvp /tmp/hadoop/hdfs/namenode/current root@hadoop302:/tmp/hadoop/hdfs/namenode
rsync -auvp /tmp/hadoop/hdfs/namenode/current root@hadoop303:/tmp/hadoop/hdfs/namenode
# 格式化ZK
hdfs zkfc -formatZK

5.4 停止 jounalnode

在所有机器上执行

/opt/hadoop-3.2.1/sbin/hadoop-daemon.sh stop journalnode
# 或者通过 /opt/hadoop-3.2.1/bin/hdfs --daemon stop journalnode

5.5 启动 hadoop

在hadoop301 上执行

# 必须在bash 环境下执行，zsh 兼容模式也不行
start-dfs.sh
start-yarn.sh
hdfs haadmin -getAllServiceState

# 正常启动后所看到的进程 jps 查看
2193 QuorumPeerMain
5252 JournalNode
4886 NameNode
5016 DataNode
5487 DFSZKFailoverController

Hadooop Classpath

很多其他的计算引擎都会使用hadoop的hdfs和yarn，他们使用的方式都是通过Hadoop class path。通过如下命令，可以看到hadoop的class path 又哪些

/opt/hadoop-3.2.1/bin/hadoop classpath

spark without hadoop 的安装包就会要求配置已经安装的hadoop 的classpath，可以再spark-env.sh中添加如下配置

### in conf/spark-env.sh ###

# If 'hadoop' binary is on your PATH
export SPARK_DIST_CLASSPATH=$(hadoop classpath)

# With explicit path to 'hadoop' binary
export SPARK_DIST_CLASSPATH=$(/path/to/hadoop/bin/hadoop classpath)

# Passing a Hadoop configuration directory
export SPARK_DIST_CLASSPATH=$(hadoop --config /path/to/configs classpath)

troubleshooting

遇到所有namenoe 都是standby
这种问题一般是DFSZKFailoverController 没起起来，没起来的原因一般是hdfs zkfc -formatZK 初始化失败或者后期操作破坏了数据，通过命令hdfs zkfc -formatZK重新初始化即可

参考资料
 参考资料
 参考资料