分享

Hadoop-2.3.0-cdh5.0.1完全分布式环境搭建(NameNode,ResourceManager HA)

nettman 发表于 2014-9-11 18:26:48 [显示全部楼层] 只看大图 回帖奖励 阅读模式 关闭右栏 7 102452
本帖最后由 nettman 于 2014-9-11 19:05 编辑
问题导读

1.YARN的基础架构是什么?
2.哪一进程防止Yarn早就web攻击?
3.Zookeeper主进程是哪个?
4.如何下载cdh包?





本文搭建Hadoop CDH5.0.1 分布式系统,包括NameNode ,ResourceManger HA,忽略了Web Application Proxy 和Job HistoryServer。
一概述(一)HDFS

1)基础架构

(1)NameNode(Master)
  • 命名空间管理:命名空间支持对HDFS中的目录、文件和块做类似文件系统的创建、修改、删除、列表文件和目录等基本操作。
  • 块存储管理

(2)DataNode(Slaver)
namenode和client的指令进行存储或者检索block,并且周期性的向namenode节点报告它存了哪些文件的block

2)HA架构

1.png


使用Active NameNode,Standby NameNode 两个结点解决单点问题,两个结点通过JounalNode共享状态,通过ZKFC 选举Active ,监控状态,自动备援。

(1)Active NameNode:
接受client的RPC请求并处理,同时写自己的Editlog和共享存储上的Editlog,接收DataNode的Block report, block location updates和heartbeat;

(2)Standby NameNode:
      同样会接到来自DataNode的Block report, block location updates和heartbeat,同时会从共享存储的Editlog上读取并执行这些log操作,使得自己的NameNode中的元数据(Namespcae information + Block locations map)都是和Active NameNode中的元数据是同步的。所以说Standby模式的NameNode是一个热备(Hot Standby NameNode),一旦切换成Active模式,马上就可以提供NameNode服务

(3)JounalNode:
用于Active NameNode , Standby NameNode 同步数据,本身由一组JounnalNode结点组成,该组结点基数个,支持Paxos协议,保证高可用,是CDH5唯一支持的共享方式(相对于CDH4 促在NFS共享方式)

(4)ZKFC:
监控NameNode进程,自动备援。

(二)YARN

1)基础架构

(1)ResourceManager(RM)
接收客户端任务请求,接收和监控NodeManager(NM)的资源情况汇报,负责资源的分配与调度,启动和监控ApplicationMaster(AM)。

(2)NodeManager
节点上的资源管理,启动Container运行task计算,上报资源、container情况给RM和任务处理情况给AM。

(3)ApplicationMaster
单个Application(Job)的task管理和调度,向RM进行资源的申请,向NM发出launch Container指令,接收NM的task处理状态信息。NodeManager

(4)Web Application Proxy
用于防止Yarn遭受Web攻击,本身是ResourceManager的一部分,可通过配置独立进程。ResourceManager Web的访问基于守信用户,当Application Master运行于一个非受信用户,其提供给ResourceManager的可能是非受信连接,Web Application Proxy可以阻止这种连接提供给RM。

(5)Job History Server
NodeManager在启动的时候会初始化LogAggregationService服务, 该服务会在把本机执行的container log (在container结束的时候)收集并存放到hdfs指定的目录下. ApplicationMaster会把jobhistory信息写到hdfs的jobhistory临时目录下, 并在结束的时候把jobhisoty移动到最终目录, 这样就同时支持了job的recovery.History会启动web和RPC服务, 用户可以通过网页或RPC方式获取作业的信息

2)HA架构

2.png


      ResourceManager HA 由一对Active,Standby结点构成,通过RMStateStore存储内部数据和主要应用的数据及标记。目前支持的可替代的RMStateStore实现有:基于内存的MemoryRMStateStore,基于文件系统的FileSystemRMStateStore,及基于zookeeper的ZKRMStateStore。
ResourceManager HA的架构模式同NameNode HA的架构模式基本一致,数据共享由RMStateStore,而ZKFC成为 ResourceManager进程的一个服务,非独立存在。

二 规划

(一)版本
组件名
版本
说明
JRE
java version "1.7.0_60"
Java(TM) SE Runtime Environment (build 1.7.0_60-b19)
Java HotSpot(TM) Client VM (build 24.60-b09, mixed mode)

Hadoop
hadoop-2.3.0-cdh5.0.1.tar.gz
主程序包

Zookeeper
zookeeper-3.4.5-cdh5.0.1.tar.gz
热切,Yarn 存储数据使用的协调服务



(二)主机规划
IP
Host
部署模块
进程
8.8.8.11
Hadoop-NN-01
NameNode
ResourceManager
NameNode
DFSZKFailoverController
ResourceManager
8.8.8.12
Hadoop-NN-02
NameNode
ResourceManager
NameNode
DFSZKFailoverController
ResourceManager
8.8.8.13
Hadoop-DN-01
Zookeeper-01
DataNode
NodeManager
Zookeeper
DataNode
NodeManager
JournalNode
QuorumPeerMain
8.8.8.14
Hadoop-DN-02
Zookeeper-02
DataNode
NodeManager
Zookeeper
DataNode
NodeManager
JournalNode
QuorumPeerMain
8.8.8.15
Hadoop-DN-03
Zookeeper-03
DataNode
NodeManager
Zookeeper
DataNode
NodeManager
JournalNode
QuorumPeerMain

各个进程解释:
  • NameNode
  • ResourceManager
  • DFSZKFC:DFS Zookeeper Failover Controller 激活Standby NameNode
  • DataNode
  • NodeManager
  • JournalNode:NameNode共享editlog结点服务(如果使用NFS共享,则该进程和所有启动相关配置接可省略)。
  • QuorumPeerMain:Zookeeper主进程


(三)目录规划
名称
路径
$HADOOP_HOME
/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1
Data
$ HADOOP_HOME/data
Log
$ HADOOP_HOME/logs


三 环境准备

1)关闭防火墙


root 用户:
  1. [root@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]# service iptables stop
  2. iptables: Flushing firewall rules: [  OK  ]
  3. iptables: Setting chains to policy ACCEPT: filter [  OK  ]
  4. iptables: Unloading modules: [  OK  ]
复制代码




验证:
  1. [root@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]# service  iptables status
  2. iptables: Firewall is not running.
复制代码





2)安装JRE:略
3)安装Zookeeper :参见Zookeeper-3.4.5-cdh5.0.1 单机模式、副本模式安装指导

4)配置SSH互信:

(1)Hadoop-NN-01创建密钥:
  1. [zero@CentOS-Cluster-01 ~]$ ssh-keygen
  2. Generating public/private rsa key pair.
  3. Enter file in which to save the key (/home/zero/.ssh/id_rsa):
  4. Created directory '/home/zero/.ssh'.
  5. Enter passphrase (empty for no passphrase):
  6. Enter same passphrase again:
  7. Your identification has been saved in /home/zero/.ssh/id_rsa.
  8. Your public key has been saved in /home/zero/.ssh/id_rsa.pub.
  9. The key fingerprint is:
  10. 28:0a:29:1d:98:56:55:db:ec:83:93:56:8a:0f:6c:ea zero@CentOS-Cluster-01
  11. The key's randomart image is:
  12. +--[ RSA 2048]----+
  13. |   .....         |
  14. | o.     +        |
  15. |o..    . +       |
  16. |.o .. ..*        |
  17. |+ . .=.*So       |
  18. |.. .o.+ . .      |
  19. |  ..   .         |
  20. |  .              |
  21. |   E             |
  22. +-----------------+
复制代码



(2)分发密钥:
  1. [zero@CentOS-Cluster-01 ~]$ ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-NN-01
  2. The authenticity of host 'hadoop-nn-01 (8.8.8.11)' can't be established.
  3. RSA key fingerprint is a6:11:09:49:8c:fe:b2:fb:49:d5:01:fa:13:1b:32:24.
  4. Are you sure you want to continue connecting (yes/no)? yes
  5. Warning: Permanently added 'hadoop-nn-01,8.8.8.11' (RSA) to the list of known hosts.
  6. puppet@hadoop-nn-01's password:
  7. Permission denied, please try again.
  8. puppet@hadoop-nn-01's password:
  9. [zero@CentOS-Cluster-01 ~][        DISCUZ_CODE_105        ]nbsp; ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-NN-01
  10. zero@hadoop-nn-01's password:
  11. Now try logging into the machine, with "ssh 'zero@Hadoop-NN-01'", and check in:
  12.   .ssh/authorized_keys
  13. to make sure we haven't added extra keys that you weren't expecting.
  14. [zero@CentOS-Cluster-01 ~][        DISCUZ_CODE_105        ]nbsp; ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-NN-02
  15. (…略…)
  16. [zero@CentOS-Cluster-01 ~][        DISCUZ_CODE_105        ]nbsp; ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-DN-01
  17. (…略…)
  18. [zero@CentOS-Cluster-01 ~][        DISCUZ_CODE_105        ]nbsp; ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-DN-02
  19. (…略…)
  20. [zero@CentOS-Cluster-01 ~][        DISCUZ_CODE_105        ]nbsp; ssh-copy-id -i ~/.ssh/id_rsa.pub zero@Hadoop-DN-03
  21. (…略…)
复制代码
分发密钥这里如出现问题,可以参考:
linux(ubuntu)ssh无密码互通、相互登录高可靠文档

CentOS6.4之图解SSH无验证双向登陆配置


(3)验证:
  1. [zero@CentOS-Cluster-01 ~]$ ssh Hadoop-NN-01
  2. Last login: Sun Jun 22 19:56:23 2014 from 8.8.8.1
  3. [zero@CentOS-Cluster-01 ~]$ exit
  4. logout
  5. Connection to Hadoop-NN-01 closed.
  6. [zero@CentOS-Cluster-01 ~]$ ssh Hadoop-NN-02
  7. Last login: Sun Jun 22 20:03:31 2014 from 8.8.8.1
  8. [zero@CentOS-Cluster-03 ~]$ exit
  9. logout
  10. Connection to Hadoop-NN-02 closed.
  11. [zero@CentOS-Cluster-01 ~]$ ssh Hadoop-DN-01
  12. Last login: Mon Jun 23 02:00:07 2014 from centos_cluster_01
  13. [zero@CentOS-Cluster-03 ~]$ exit
  14. logout
  15. Connection to Hadoop-DN-01 closed.
  16. [zero@CentOS-Cluster-01 ~]$ ssh Hadoop-DN-02
  17. Last login: Sun Jun 22 20:07:03 2014 from 8.8.8.1
  18. [zero@CentOS-Cluster-04 ~]$ exit
  19. logout
  20. Connection to Hadoop-DN-02 closed.
  21. [zero@CentOS-Cluster-01 ~]$ ssh Hadoop-DN-03
  22. Last login: Sun Jun 22 20:07:05 2014 from 8.8.8.1
  23. [zero@CentOS-Cluster-05 ~]$ exit
  24. logout
  25. Connection to Hadoop-DN-03 closed.
复制代码





5)配置/etc/hosts并分发:
  1. [root@CentOS-Cluster-01 zero]# vi /etc/hosts
  2. 127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4
  3. ::1         localhost localhost.localdomain localhost6 localhost6.localdomain6
  4. 8.8.8.10 CentOS-StandAlone
  5. 8.8.8.11 CentOS-Cluster-01 Hadoop-NN-01
  6. 8.8.8.12 CentOS-Cluster-02 Hadoop-NN-02
  7. 8.8.8.13 CentOS-Cluster-03 Hadoop-DN-01 Zookeeper-01
  8. 8.8.8.14 CentOS-Cluster-04 Hadoop-DN-02 Zookeeper-02
  9. 8.8.8.15 CentOS-Cluster-05 Hadoop-DN-03 Zookeeper-03
复制代码




6)配置环境变量:vi ~/.bashrc 然后 source ~/.bashrc
  1. [zero@CentOS-Cluster-01 ~]$ vi ~/.bashrc
  2. ……
  3. # hadoop cdh5
  4. export HADOOP_HOME=/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1
  5. export PATH=$PATH:$HADOOP_HOME/sbin:$HADOOP_HOME/bin
  6. [zero@CentOS-Cluster-01 ~]$ source ~/.bashrc
复制代码




四 安装

1)解压
  1. [puppet@BigData-01 cdh4.4]$ tar -xvf hadoop-2.0.0-cdh4.4.0.tar.gz
复制代码

下载地址:
CDH4包汇总
http://archive.cloudera.com/cdh4/
CDH5包汇总
http://archive.cloudera.com/cdh5/


2)修改配置文件
说明:

配置名称
类型
说明
hadoop-env.sh
Bash脚本
Hadoop运行环境变量设置
core-site.xml
xml
配置Hadoop core,如IO
hdfs-site.xml
xml
配置HDFS守护进程:NN、JN、DN
yarn-env.sh
Bash脚本
Yarn运行环境变量设置
yarn-site.xml
xml
Yarn框架配置环境
mapred-site.xml
xml
MR属性设置
capacity-scheduler.xml
xml
Yarn调度属性设置
container-executor.cfg
Cfg
Yarn Container配置
mapred-queues.xml
xml
MR队列设置
hadoop-metrics.properties
Java属性
Hadoop Metrics配置
hadoop-metrics2.properties
Java属性
Hadoop Metrics配置
slaves
Plain Text
DN节点配置
exclude
Plain Text
移除DN节点配置文件
log4j.properties

系统日志设置
configuration.xsl




(1)修改$HADOOP_HOME/etc/hadoop-env.sh:

  1. #--------------------Java Env------------------------------
  2. export JAVA_HOME="/usr/runtime/java/jdk1.7.0_60"
  3. #--------------------Hadoop Env------------------------------
  4. #export HADOOP_PID_DIR=
  5. export HADOOP_PREFIX="/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1"
  6. #--------------------Hadoop Daemon Options-----------------
  7. #export HADOOP_NAMENODE_OPTS="-XX:+UseParallelGC ${HADOOP_NAMENODE_OPTS}"
  8. #export HADOOP_DATANODE_OPTS=
  9. #--------------------Hadoop Logs---------------------------
  10. #export HADOOP_LOG_DIR=
复制代码



(2)修改$HADOOP_HOME/etc/hadoop-site.xml

  1. <?xml version="1.0" encoding="UTF-8"?>
  2. <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
  3. <configuration>
  4.         <!--Yarn 需要使用 fs.defaultFS 指定NameNode URI -->
  5.         <property>
  6.                 <name>fs.defaultFS</name>
  7.                 <value>hdfs://mycluster</value>
  8.         </property>
  9.         <!--HDFS超级用户 -->
  10.         <property>
  11.                 <name>dfs.permissions.superusergroup</name>
  12.                 <value>zero</value>
  13.         </property>
  14.         <!--==============================Trash机制======================================= -->
  15.         <property>
  16.                 <!--多长时间创建CheckPoint NameNode截点上运行的CheckPointer 从Current文件夹创建CheckPoint;默认:0 由fs.trash.interval项指定 -->
  17.                 <name>fs.trash.checkpoint.interval</name>
  18.                 <value>0</value>
  19.         </property>
  20.         <property>
  21.                 <!--多少分钟.Trash下的CheckPoint目录会被删除,该配置服务器设置优先级大于客户端,默认:0 不删除 -->
  22.                 <name>fs.trash.interval</name>
  23.                 <value>1440</value>
  24.         </property>
  25. </configuration>
复制代码

(3)修改$HADOOP_HOME/etc/hdfs-site.xml

  1. <?xml version="1.0" encoding="UTF-8"?>
  2. <?xml-stylesheet type="text/xsl" href="configuration.xsl"?>
  3. <configuration>
  4.         <!--开启web hdfs -->
  5.         <property>
  6.                 <name>dfs.webhdfs.enabled</name>
  7.                 <value>true</value>
  8.         </property>
  9.         <property>
  10.                 <name>dfs.namenode.name.dir</name>
  11.                 <value>/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/data/dfs/name</value>
  12.                 <description> namenode 存放name table(fsimage)本地目录(需要修改)</description>
  13.         </property>
  14.         <property>
  15.                 <name>dfs.namenode.edits.dir</name>
  16.                 <value>${dfs.namenode.name.dir}</value>
  17.                 <description>namenode粗放 transaction file(edits)本地目录(需要修改)</description>
  18.         </property>
  19.         <property>
  20.                 <name>dfs.datanode.data.dir</name>
  21.                 <value>/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/data/dfs/data</value>
  22.                 <description>datanode存放block本地目录(需要修改)</description>
  23.         </property>
  24.         <property>
  25.                 <name>dfs.replication</name>
  26.                 <value>1</value>
  27.         </property>
  28.         <!-- 块大小 (默认) -->
  29.         <property>
  30.                 <name>dfs.blocksize</name>
  31.                 <value>268435456</value>
  32.         </property>
  33.         <!--======================================================================= -->
  34.         <!--HDFS高可用配置 -->
  35.         <!--nameservices逻辑名 -->
  36.         <property>
  37.                 <name>dfs.nameservices</name>
  38.                 <value>mycluster</value>
  39.         </property>
  40.         <property>
  41.                 <!--设置NameNode IDs 此版本最大只支持两个NameNode -->
  42.                 <name>dfs.ha.namenodes.mycluster</name>
  43.                 <value>nn1,nn2</value>
  44.         </property>
  45.         <!-- Hdfs HA: dfs.namenode.rpc-address.[nameservice ID] rpc 通信地址 -->
  46.         <property>
  47.                 <name>dfs.namenode.rpc-address.mycluster.nn1</name>
  48.                 <value>Hadoop-NN-01:8020</value>
  49.         </property>
  50.         <property>
  51.                 <name>dfs.namenode.rpc-address.mycluster.nn2</name>
  52.                 <value>Hadoop-NN-02:8020</value>
  53.         </property>
  54.         <!-- Hdfs HA: dfs.namenode.http-address.[nameservice ID] http 通信地址 -->
  55.         <property>
  56.                 <name>dfs.namenode.http-address.mycluster.nn1</name>
  57.                 <value>Hadoop-NN-01:50070</value>
  58.         </property>
  59.         <property>
  60.                 <name>dfs.namenode.http-address.mycluster.nn2</name>
  61.                 <value>Hadoop-NN-02:50070</value>
  62.         </property>
  63.         <!--==================Namenode editlog同步 ============================================ -->
  64.         <!--保证数据恢复 -->
  65.         <property>
  66.                 <name>dfs.journalnode.http-address</name>
  67.                 <value>0.0.0.0:8480</value>
  68.         </property>
  69.         <property>
  70.                 <name>dfs.journalnode.rpc-address</name>
  71.                 <value>0.0.0.0:8485</value>
  72.         </property>
  73.         <property>
  74.                 <!--设置JournalNode服务器地址,QuorumJournalManager 用于存储editlog -->
  75.                 <!--格式:qjournal://<host1:port1>;<host2:port2>;<host3:port3>/<journalId> 端口同journalnode.rpc-address -->
  76.                 <name>dfs.namenode.shared.edits.dir</name>
  77.                 <value>qjournal://Hadoop-DN-01:8485;Hadoop-DN-02:8485;Hadoop-DN-03:8485/mycluster</value>
  78.         </property>
  79.         <property>
  80.                 <!--JournalNode存放数据地址 -->
  81.                 <name>dfs.journalnode.edits.dir</name>
  82.                 <value>/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/data/dfs/jn</value>
  83.         </property>
  84.         <!--==================DataNode editlog同步 ============================================ -->
  85.         <property>
  86.                 <!--DataNode,Client连接Namenode识别选择Active NameNode策略 -->
  87.                 <name>dfs.client.failover.proxy.provider.mycluster</name>
  88.                 <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
  89.         </property>
  90.         <!--==================Namenode fencing:=============================================== -->
  91.         <!--Failover后防止停掉的Namenode启动,造成两个服务 -->
  92.         <property>
  93.                 <name>dfs.ha.fencing.methods</name>
  94.                 <value>sshfence</value>
  95.         </property>
  96.         <property>
  97.                 <name>dfs.ha.fencing.ssh.private-key-files</name>
  98.                 <value>/home/zero/.ssh/id_rsa</value>
  99.         </property>
  100.         <property>
  101.                 <!--多少milliseconds 认为fencing失败 -->
  102.                 <name>dfs.ha.fencing.ssh.connect-timeout</name>
  103.                 <value>30000</value>
  104.         </property>
  105.         <!--==================NameNode auto failover base ZKFC and Zookeeper====================== -->
  106.         <!--开启基于Zookeeper及ZKFC进程的自动备援设置,监视进程是否死掉 -->
  107.         <property>
  108.                 <name>dfs.ha.automatic-failover.enabled</name>
  109.                 <value>true</value>
  110.         </property>
  111.         <property>
  112.                 <name>ha.zookeeper.quorum</name>
  113.                 <value>Zookeeper-01:2181,Zookeeper-02:2181,Zookeeper-03:2181</value>
  114.         </property>
  115.         <property>
  116.                 <!--指定ZooKeeper超时间隔,单位毫秒 -->
  117.                 <name>ha.zookeeper.session-timeout.ms</name>
  118.                 <value>2000</value>
  119.         </property>
  120. </configuration>
复制代码

(4)修改$HADOOP_HOME/etc/yarn-env.sh
  1. #Yarn Daemon Options
  2. #export YARN_RESOURCEMANAGER_OPTS
  3. #export YARN_NODEMANAGER_OPTS
  4. #export YARN_PROXYSERVER_OPTS
  5. #export HADOOP_JOB_HISTORYSERVER_OPTS
  6. #Yarn Logs
  7. export YARN_LOG_DIR=” /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs”
复制代码

(5)$HADOOP_HOEM/etc/mapred-site.xml
  1. <configuration>
  2.         <!-- 配置 MapReduce Applications -->
  3.         <property>
  4.                 <name>mapreduce.framework.name</name>
  5.                 <value>yarn</value>
  6.         </property>
  7.         <!-- JobHistory Server ============================================================== -->
  8.         <!-- 配置 MapReduce JobHistory Server 地址 ,默认端口10020 -->
  9.         <property>
  10.                 <name>mapreduce.jobhistory.address</name>
  11.                 <value>0.0.0.0:10020</value>
  12.         </property>
  13.         <!-- 配置 MapReduce JobHistory Server web ui 地址, 默认端口19888 -->
  14.         <property>
  15.                 <name>mapreduce.jobhistory.webapp.address</name>
  16.                 <value>0.0.0.0:19888</value>
  17.         </property>
  18. </configuration>
复制代码

(6)修改$HADOOP_HOME/etc/yarn-site.xml

  1. <configuration>
  2.         <!-- nodemanager 配置 ================================================= -->
  3.         <property>
  4.                 <name>yarn.nodemanager.aux-services</name>
  5.                 <value>mapreduce_shuffle</value>
  6.         </property>
  7.         <property>
  8.                 <name>yarn.nodemanager.aux-services.mapreduce.shuffle.class</name>
  9.                 <value>org.apache.hadoop.mapred.ShuffleHandler</value>
  10.         </property>
  11.         <property>
  12.                 <description>Address where the localizer IPC is.</description>
  13.                 <name>yarn.nodemanager.localizer.address</name>
  14.                 <value>0.0.0.0:23344</value>
  15.         </property>
  16.         <property>
  17.                 <description>NM Webapp address.</description>
  18.                 <name>yarn.nodemanager.webapp.address</name>
  19.                 <value>0.0.0.0:23999</value>
  20.         </property>
  21.         <!-- HA 配置 =============================================================== -->
  22.         <!-- Resource Manager Configs -->
  23.         <property>
  24.                 <name>yarn.resourcemanager.connect.retry-interval.ms</name>
  25.                 <value>2000</value>
  26.         </property>
  27.         <property>
  28.                 <name>yarn.resourcemanager.ha.enabled</name>
  29.                 <value>true</value>
  30.         </property>
  31.         <property>
  32.                 <name>yarn.resourcemanager.ha.automatic-failover.enabled</name>
  33.                 <value>true</value>
  34.         </property>
  35.         <!-- 使嵌入式自动故障转移。HA环境启动,与 ZKRMStateStore 配合 处理fencing -->
  36.         <property>
  37.                 <name>yarn.resourcemanager.ha.automatic-failover.embedded</name>
  38.                 <value>true</value>
  39.         </property>
  40.         <!-- 集群名称,确保HA选举时对应的集群 -->
  41.         <property>
  42.                 <name>yarn.resourcemanager.cluster-id</name>
  43.                 <value>yarn-cluster</value>
  44.         </property>
  45.         <property>
  46.                 <name>yarn.resourcemanager.ha.rm-ids</name>
  47.                 <value>rm1,rm2</value>
  48.         </property>
  49.     <!—这里RM主备结点需要单独指定,(可选)
  50.         <property>
  51.         <name>yarn.resourcemanager.ha.id</name>
  52.         <value>rm2</value>
  53. </property>
  54. -->
  55.         <property>
  56.                 <name>yarn.resourcemanager.scheduler.class</name>
  57.                 <value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.fair.FairScheduler</value>
  58.         </property>
  59.         <property>
  60.                 <name>yarn.resourcemanager.recovery.enabled</name>
  61.                 <value>true</value>
  62.         </property>
  63.         <property>
  64.                 <name>yarn.app.mapreduce.am.scheduler.connection.wait.interval-ms</name>
  65.                 <value>5000</value>
  66.         </property>
  67.         <!-- ZKRMStateStore 配置 -->
  68.         <property>
  69.                 <name>yarn.resourcemanager.store.class</name>
  70.                 <value>org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore</value>
  71.         </property>
  72.         <property>
  73.                 <name>yarn.resourcemanager.zk-address</name>
  74.                 <value>Zookeeper-01:2181,Zookeeper-02:2181,Zookeeper-03:2181</value>
  75.         </property>
  76.         <property>
  77.                 <name>yarn.resourcemanager.zk.state-store.address</name>
  78.                 <value>Zookeeper-01:2181,Zookeeper-02:2181,Zookeeper-03:2181</value>
  79.         </property>
  80.         <!-- Client访问RM的RPC地址 (applications manager interface) -->
  81.         <property>
  82.                 <name>yarn.resourcemanager.address.rm1</name>
  83.                 <value>Hadoop-NN-01:23140</value>
  84.         </property>
  85.         <property>
  86.                 <name>yarn.resourcemanager.address.rm2</name>
  87.                 <value>Hadoop-NN-02:23140</value>
  88.         </property>
  89.         <!-- AM访问RM的RPC地址(scheduler interface) -->
  90.         <property>
  91.                 <name>yarn.resourcemanager.scheduler.address.rm1</name>
  92.                 <value>Hadoop-NN-01:23130</value>
  93.         </property>
  94.         <property>
  95.                 <name>yarn.resourcemanager.scheduler.address.rm2</name>
  96.                 <value>Hadoop-NN-02:23130</value>
  97.         </property>
  98.         <!-- RM admin interface -->
  99.         <property>
  100.                 <name>yarn.resourcemanager.admin.address.rm1</name>
  101.                 <value>Hadoop-NN-01:23141</value>
  102.         </property>
  103.         <property>
  104.                 <name>yarn.resourcemanager.admin.address.rm2</name>
  105.                 <value>Hadoop-NN-02:23141</value>
  106.         </property>
  107.         <!--NM访问RM的RPC端口 -->
  108.         <property>
  109.                 <name>yarn.resourcemanager.resource-tracker.address.rm1</name>
  110.                 <value>Hadoop-NN-01:23125</value>
  111.         </property>
  112.         <property>
  113.                 <name>yarn.resourcemanager.resource-tracker.address.rm2</name>
  114.                 <value>Hadoop-NN-02:23125</value>
  115.         </property>
  116.         <!-- RM web application 地址 -->
  117.         <property>
  118.                 <name>yarn.resourcemanager.webapp.address.rm1</name>
  119.                 <value>Hadoop-NN-01:23188</value>
  120.         </property>
  121.         <property>
  122.                 <name>yarn.resourcemanager.webapp.address.rm2</name>
  123.                 <value>Hadoop-NN-02:23188</value>
  124.         </property>
  125.         <property>
  126.                 <name>yarn.resourcemanager.webapp.https.address.rm1</name>
  127.                 <value>Hadoop-NN-01:23189</value>
  128.         </property>
  129.         <property>
  130.                 <name>yarn.resourcemanager.webapp.https.address.rm2</name>
  131.                 <value>Hadoop-NN-02:23189</value>
  132.         </property>
  133. </configuration>
复制代码

(7)修改slaves

  1. [zero@CentOS-Cluster-01  hadoop]$ vi slaves
  2. Hadoop-DN-01
  3. Hadoop-DN-02
  4. Hadoop-DN-03
复制代码

3)分发程序
  1. [zero@CentOS-Cluster-01 ~]$ scp -r /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1 zero@Hadoop-NN-02: /home/zero/hadoop/
  2. …….
  3. [zero@CentOS-Cluster-01 ~]$ scp -r /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1 zero@Hadoop-DN-01: /home/zero/hadoop/
  4. …….
  5. [zero@CentOS-Cluster-01 ~]$ scp -r /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1 zero@Hadoop-DN-02: /home/zero/hadoop/
  6. …….
  7. [zero@CentOS-Cluster-01 ~]$ scp -r /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1 zero@Hadoop-DN-03: /home/zero/hadoop/
复制代码


4)启动HDFS

(1)启动JournalNode
格式化前需要在JournalNode结点上启动JournalNode:
  1. [zero@CentOS-Cluster-03 hadoop-2.3.0-cdh5.0.1]$ hadoop-daemon.sh start journalnode
  2. starting journalnode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-journalnode-BigData-03.out
复制代码

验证JournalNode:
  1. [zero@CentOS-Cluster-03 hadoop-2.3.0-cdh5.0.1]$ jps
  2. 25918 QuorumPeerMain
  3. 16728 JournalNode
  4. 16772 Jps
复制代码


(2)NameNode 格式化:
结点Hadoop-NN-01:hdfs namenode -format
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hdfs namenode -format
  2. 14/06/23 21:02:49 INFO namenode.NameNode: STARTUP_MSG:
  3. /************************************************************
  4. STARTUP_MSG: Starting NameNode
  5. STARTUP_MSG:   host = CentOS-Cluster-01/8.8.8.11
  6. STARTUP_MSG:   args = [-format]
  7. STARTUP_MSG:   version = 2.3.0-cdh5.0.1
  8. STARTUP_MSG:   classpath =…
  9. STARTUP_MSG:   build =…
  10. STARTUP_MSG:   java = 1.7.0_60
  11. ************************************************************/
  12. (…略…)
  13. 14/06/23 21:02:57 INFO common.Storage: Storage directory /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/data/dfs/name has been successfully formatted. (…略…)
  14. 14/06/23 21:02:59 INFO namenode.NameNode: SHUTDOWN_MSG:
  15. /************************************************************
  16. SHUTDOWN_MSG: Shutting down NameNode at CentOS-Cluster-01/8.8.8.11
  17. ************************************************************/
复制代码


(3)同步NameNode元数据:
同步Hadoop-NN-01元数据到Hadoop-NN-02
主要是:dfs.namenode.name.dir,dfs.namenode.edits.dir还应该确保共享存储目录下(dfs.namenode.shared.edits.dir ) 包含NameNode 所有的元数据。
  1. [puppet@BigData-01 hadoop-2.0.0-cdh4.4.0]$ scp -r data/ zero@Hadoop-NN-02:/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1
  2. seen_txid                           100%    2     0.0KB/s   00:00   
  3. VERSION                            100%  201     0.2KB/s   00:00   
  4. seen_txid                           100%    2     0.0KB/s   00:00   
  5. fsimage_0000000000000000000.md5    100%   62     0.1KB/s   00:00   
  6. fsimage_0000000000000000000         100%  121     0.1KB/s   00:00   
  7. VERSION                            100%  201     0.2KB/s   00:00
复制代码


(4)初始化ZFCK:
创建ZNode,记录状态信息。
结点Hadoop-NN-01:hdfs zkfc -formatZK
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hdfs zkfc -formatZK
  2. 14/06/23 23:22:28 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
  3. 14/06/23 23:22:28 INFO tools.DFSZKFailoverController: Failover controller configured for NameNode NameNode at Hadoop-NN-01/8.8.8.11:8020
  4. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:zookeeper.version=3.4.5-cdh5.0.1--1, built on 05/06/2014 18:50 GMT
  5. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:host.name=CentOS-Cluster-01
  6. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:java.version=1.7.0_60
  7. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:java.vendor=Oracle Corporation
  8. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:java.home=/usr/runtime/java/jdk1.7.0_60/jre
  9. 14/06/23 23:22:28 INFO zookeeper.ZooKeeper: Client environment:java.class.path=...
  10. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:java.library.path=/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/lib/native
  11. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:java.io.tmpdir=/tmp
  12. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:java.compiler=<NA>
  13. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:os.name=Linux
  14. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:os.arch=i386
  15. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:os.version=2.6.32-431.el6.i686
  16. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:user.name=zero
  17. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:user.home=/home/zero
  18. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Client environment:user.dir=/home/zero/hadoop/hadoop-2.3.0-cdh5.0.1
  19. 14/06/23 23:22:29 INFO zookeeper.ZooKeeper: Initiating client connection, connectString=Zookeeper-01:2181,Zookeeper-02:2181,Zookeeper-03:2181 sessionTimeout=2000 watcher=org.apache.hadoop.ha.ActiveStandbyElector$WatcherWithClientRef@150c2b0
  20. 14/06/23 23:22:29 INFO zookeeper.ClientCnxn: Opening socket connection to server CentOS-Cluster-03/8.8.8.13:2181. Will not attempt to authenticate using SASL (unknown error)
  21. 14/06/23 23:22:29 INFO zookeeper.ClientCnxn: Socket connection established to CentOS-Cluster-03/8.8.8.13:2181, initiating session
  22. 14/06/23 23:22:29 INFO zookeeper.ClientCnxn: Session establishment complete on server CentOS-Cluster-03/8.8.8.13:2181, sessionid = 0x146cc0517b30002, negotiated timeout = 4000
  23. 14/06/23 23:22:29 INFO ha.ActiveStandbyElector: Session connected.
  24. 14/06/23 23:22:30 INFO ha.ActiveStandbyElector: Successfully created /hadoop-ha/mycluster in ZK.
  25. 14/06/23 23:22:30 INFO zookeeper.ZooKeeper: Session: 0x146cc0517b30002 closed
复制代码



(5)启动
集群启动法:Hadoop-NN-01:
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ start-dfs.sh
  2. 14/04/23 01:54:58 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
  3. Starting namenodes on [Hadoop-NN-01 Hadoop-NN-02]
  4. Hadoop-NN-01: starting namenode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-namenode-BigData-01.out
  5. Hadoop-NN-02: starting namenode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-namenode-BigData-02.out
  6. Hadoop-DN-01: starting datanode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-datanode-BigData-03.out
  7. Hadoop-DN-02: starting datanode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-datanode-BigData-04.out
  8. Hadoop-DN-03: starting datanode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-datanode-BigData-05.out
  9. Starting journal nodes [Hadoop-DN-01 Hadoop-DN-02 Hadoop-DN-03]
  10. Hadoop-DN-01: starting journalnode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-journalnode-BigData-03.out
  11. Hadoop-DN-03: starting journalnode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-journalnode-BigData-05.out
  12. Hadoop-DN-02: starting journalnode, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-journalnode-BigData-04.out
  13. 14/04/23 01:55:29 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
  14. Starting ZK Failover Controllers on NN hosts [Hadoop-NN-01 Hadoop-NN-02]
  15. Hadoop-NN-01: starting zkfc, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-zkfc-BigData-01.out
  16. Hadoop-NN-02: starting zkfc, logging to /home/puppet/hadoop/cdh4.4/hadoop-2.0.0-cdh4.4.0/logs/hadoop-puppet-zkfc-BigData-02.out
复制代码



单进程启动法:
<1>NameNode(Hadoop-NN-01,Hadoop-NN-02):

  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hadoop-daemon.sh start namenode
复制代码





<2>DataNode(Hadoop-DN-01,Hadoop-DN-02,Hadoop-DN-03):

  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hadoop-daemon.sh start datanode
复制代码




<3>JournalNode(Hadoop-DN-01,Hadoop-DN-02,Hadoop-DN-03):
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hadoop-daemon.sh start journalnode
复制代码





<4>ZKFC(Hadoop-NN-01,Hadoop-NN-02):
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ hadoop-daemon.sh start zkfc
复制代码




(6)验证

<1>进程

NameNode:
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ jps
  2. 4001 NameNode
  3. 4290 DFSZKFailoverController
  4. 4415 Jps
复制代码





DataNode:
  1. [zero@CentOS-Cluster-03 hadoop-2.3.0-cdh5.0.1]$ jps
  2. 25918 QuorumPeerMain
  3. 19217 JournalNode
  4. 19143 DataNode
  5. 19351 Jps
复制代码





<2>页面:
Active结点:http://8.8.8.10:50070

3.png


JSP:
4.png


StandBy结点:
5.png


JSP:

6.png


(7)停止:stop-dfs.sh



5)启动Yarn

(1)启动

<1>集群启动
Hadoop-NN-01启动Yarn,命令所在目录:$HADOOP_HOME/sbin
  1. [zero@CentOS-Cluster-01 hadoop]$ start-yarn.sh
  2. starting yarn daemons
  3. starting resourcemanager, logging to /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs/yarn-zero-resourcemanager-CentOS-Cluster-01.out
  4. Hadoop-DN-02: starting nodemanager, logging to /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs/yarn-zero-nodemanager-CentOS-Cluster-04.out
  5. Hadoop-DN-03: starting nodemanager, logging to /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs/yarn-zero-nodemanager-CentOS-Cluster-05.out
  6. Hadoop-DN-01: starting nodemanager, logging to /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs/yarn-zero-nodemanager-CentOS-Cluster-03.out
复制代码





Hadoop-NN-02备机启动RM:
  1. [zero@CentOS-Cluster-02 hadoop]$ yarn-daemon.sh start resourcemanager
  2. starting resourcemanager, logging to /home/zero/hadoop/hadoop-2.3.0-cdh5.0.1/logs/yarn-zero-resourcemanager-CentOS-Cluster-02.o
复制代码





<2>单进程启动
ResourceManager(Hadoop-NN-01,Hadoop-NN-02):
  1. [zero@CentOS-Cluster-01 hadoop-2.3.0-cdh5.0.1]$ yarn-daemon.sh start resourcemanager
复制代码

DataNode(Hadoop-DN-01,Hadoop-DN-02,Hadoop-DN-03):
  1. [zero@CentOS-Cluster-03 hadoop-2.3.0-cdh5.0.1]$ yarn-daemon.sh start nodemanager
复制代码

(2)验证


<1>进程:
JobTracker:Hadoop-NN-01,Hadoop-NN-02
  1. [zero@CentOS-Cluster-01 hadoop]$ jps
  2. 7454 NameNode
  3. 14684 Jps
  4. 14617 ResourceManager
  5. 7729 DFSZKFailoverController
复制代码





TaskTracker:Hadoop-DN-01,Hadoop-DN-02,Hadoop-DN-03
  1. [zero@CentOS-Cluster-03 ~]$ jps
  2. 8965 Jps
  3. 5228 JournalNode
  4. 5137 DataNode
  5. 2514 QuorumPeerMain
  6. 8935 NodeManager
复制代码





<2>页面
ResourceManger(Active):8.8.8.11:23188

7.png

ResourceManager(Standby):8.8.8.12:23188

8.png

9.png

(3)停止
Hadoop-NN-01:stop-yarn.sh
Hadoop-NN-02:yarn-daeman.sh stop resourcemanager














转载:http://shihlei.iteye.com/
加微信w3aboutyun,可拉入技术爱好者群

已有(7)人评论

跳转到指定楼层
AlexFeng 发表于 2015-6-6 01:27:31
谢谢分享,学习了
回复

使用道具 举报

ijuee 发表于 2015-9-29 10:46:06
你好,问个问题,能用CM管理么
回复

使用道具 举报

pianai 发表于 2015-10-30 10:34:36
很详细,谢谢楼主分享,学习一下~
回复

使用道具 举报

caicai 发表于 2015-10-30 15:09:08
hadoop东西真多啊
回复

使用道具 举报

stevenNiT 发表于 2015-12-5 11:10:57
安装CDH直接在web UI里面就可以设置这些参数了,楼主为什么要手动设置呢?
回复

使用道具 举报

tang_ 发表于 2016-1-19 17:59:02
很不错的东西
回复

使用道具 举报

雷夫23 发表于 2016-6-29 09:18:19
很好的分享,楼主能给个下载链接么
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

关闭

推荐上一条 /2 下一条