准备工作
- 1.安装包版本以及功能
软件名 | 版本号 | 功能 |
---|---|---|
Hadoop | hadoop-2.7.3.tar.gz | 为海量数据提供分布式存 储(HDFS)和分布式计算 (YARN)。 |
ZooKeeper | zookeeper-3.4.10.tar.gz | 一个分布式应用程序协调 服务,为应用提供一致性服 务,是 Hadoop 和 Hbase 的 重要组件。 |
JDK | jdk-8u144-linux-x64.tar.gz | JAVA 运行环境 |
- 2.节点分配
HostName | IP | Hadoop | ZooKeeper |
---|---|---|---|
ActiveNN | 192.168.142.133 | NameNode | / |
mini1 | 192.168.142.135 | DataNode | QuorumPeerMain |
mini2 | 192.168.142.134 | DataNode | QuorumPeerMain |
mini3 | 192.168.142.137 | DataNode | QuorumPeerMain |
StandbyNN | 192.168.142.136 | NameNode | / |
- 3.逻辑结构图
NameNode(NN) HA 实现方式: 一种是将 NN 维护的元数据保存一份到 NFS 上,当 NN 故障,可以通过另一台 NNe 读 取 NFS 目录中的元数据备份进行恢复工作,需要手动进行操作,并不是真正意义上的 HA 方案。 另一种是准备一台备用 NN 节点,通过定期下载 NN 的元数据和日志文件来备份,当 NN 故障时,可以通过这台进行恢复,由于主备节点元数据和日志并不是实时同步,所以 会丢失一些数据。 前两种方案都不是很理想,社区提供一种更好的方案,基于 QJM(Qurom Journal Manager)的共享日志方案。QJM 的基本原理是 NN(Active)把日志写本地和 2N+1(奇 数)台 JournalNode 上,当数据操作返回成功时才写入日志,这个日志叫做 editlog, 而元数据存在 fsimage 文件中,NN(Standby)定期从 JournalNode 上读取 editlog 到 本地。在这手动切换的基础上有开发了基于 Zookeeper 的 ZKFC(ZookeeperFailover Controller)自动切换机制,Active 和 Standby 节点各有 ZKFC 进程监控 NN 监控状况, 定期发送心跳,当 Active 节点故障时 Standby 会自动切换为 ActiveNode,我们这次就 用的此方案,如下图所示。
ResourceManager(RM) HA 实现方式: RM 将状态信息存储在 Zookeeper 中,当 Active 故障,Standby 切换为 Active 后, 从 ZK 读取相应的作业信息,重新构建作业的内存信息,然后开始接受 NodeManager 心 跳,并接受客户端提交作业的请求等。
搭建高可用Hadoop集群
- 1.基础环境配置
1.1 修改/etc/hostname(主机名)和/etc/hosts(ip和主机名的映射)两个文件
修改主机名
#vim /etc/hostname //填写节点分配中的HostName
实现ip和主机名的映射
#vim /etc/hosts
添加如下内容
192.168.142.133 ActiveNN
192.168.142.135 mini1
192.168.142.134 mini2
192.168.142.137 mini3
192.168.142.136 StandbyNN
注:每台虚拟机都要做相应的操作并且重启虚拟机,最好将/etc/hosts中原有的东西注释掉。
1.2 配置SSH免密码登陆
生成密钥对
#ssh-keygen //一直回车生成密钥对
复制公钥给集群中的所有虚拟机(包括自己)
#ssh-copy-id root@目的虚拟机的主机名或者ip地址
1.3 安装配置JDK
#tar -zxvf jdk-8u144-linux-x64.tar.gz -C /usr/local/jvm
#vim /etc/profile
注:可以通过“scp -r /usr/local/jvm/jdk1.8.0_144 主机名:目录”将解压的jdk文件夹复制到其他虚拟机的相应目录。
如:scp -r /usr/local/jvm/jdk1.8.0_144 StandbyNN:/usr/local/jvm
添加如下记录
export JAVA_HOME=/usr/local/jvm/jdk1.8.0_144
export PATH=$PATH:$JAVA_HOME/bin:$JAVA_HOME/jre/bin
export CLASSPATH=$JAVA_HOME/lib:$JAVA_HOME/jre/lib
使配置文件生效
#source /etc/profile
验证JDK是否配置成功
#java -version
如果打印出如下信息则表示JDK安装配置成功
[root@ActiveNN ~]# java -version
java version "1.8.0_144"
Java(TM) SE Runtime Environment (build 1.8.0_144-b01)
Java HotSpot(TM) 64-Bit Server VM (build 25.144-b01, mixed mode)
[root@ActiveNN ~]#
- 2.ZooKeeper集群的安装配置(这里使用三台虚拟机来设置ZooKeeper集群,分别是mini1,mini2以及mini3,三台虚拟机同样的安装配置)
2.1 安装ZooKeeper
#tar -zxvf zookeeper-3.4.10.tar.gz -C /usr/local/hadoop/
#cd /usr/local/hadoop/zookeeper-3.4.10/conf
#cp zoo_sample.cfg zoo.cfg
#vim zoo.cfg
配置文件中有的参数就修改,没有的参数就添加,如下:
tickTime=2000
initLimit=10
syncLimit=5
dataDir=/usr/local/zookeeper-3.4.10/zookeeperdata
clientPort=2181
server.1=mini1:2888:3888
server.2=mini2:2888:3888
server.3=mini3:2888:3888
参数说明:
tickTime:zk服务器之间或者客户端与服务器之间间隔多长时间发送一个心跳,单位是毫秒
initLimit:zk服务器集群中连接Leader服务器与Follower服务器能容忍的最长的时间间隔(5*2000=10s)
syncLimit:表示Leader与Follower同步消息,如果超过时间(5*2000=10s)未完成同步,将剔除这个节点,所有连接此Follower服务器的客户端将连接到另一个Follower服务器上
dataDir:zk保存数据的目录,默认情况下zk也会将日志文件保存在此目录中
clientPort:客户端连接zk的服务器端口
server.X=A:B:C:其中X是一个数字(myid中的数字),代表这是第几号zk服务器。A是该server所在的IP地址或者主机名,B配置该server和集群中的leader交换消息所使用的端口,C配置选举leader时所使用的端口。
配置ZooKeeper环境变量
#vim /etc/profile
添加如下记录
ZOOKEEPER_HOME=/usr/local/hadoop/zookeeper-3.4.10
export PATH=$ZOOKEEPER_HOME/bin:$PATH
使配置文件生效
#source /etc/profile
2.2 创建目录和id文件(三台需要对应的配置,特别是myid的对应)
#cd /usr/local/hadoop/zookeeper-3.4.10
#mkdir zookeeperdata
#cd zookeeperdata
#vim myid
1 //必须创建id号,否则启动会报错。集群中的每一台虚拟机都要创建myid号,myid号对应zoo.cfg中配置的server后面1、2和3这个zk号。
2.3分别启动三台虚拟机的ZooKeeper(mini1,mini2以及mini3)
#/usr/local/hadoop/zookeeper-3.4.10/bin/zkServer.sh start
//分别启动zk服务器
2.4验证是否启动成功
#/usr/local/hadoop/zookeeper-3.4.10/bin/zkServer.sh status
ZooKeeper JMX enabled by default
Using config: /usr/local/hadoop/zookeeper-3.4.10/bin/../conf/zoo.cfg
Mode: leader(或follower)
//三台虚拟机中有一台leader和两台follower
另外,集群中的每一台虚拟机都会启动一个进程:
[root@mini3 bin]# jps
5122 QuorumPeerMain
5460 Jps
[root@mini3 bin]#
- 3.Hadoop的安装配置(每台虚拟机同样的安装配置)
3.1 安装与配置
#tar -zxvf hadoop-2.7.3.tar.gz -C /usr/local/hadoop/
#cd /usr/local/hadoop/hadoop-2.7.3/etc/hadoop
修改hadoop-env.sh
将export JAVA_HOME=${JAVA_HOME}修改为自己的JDK安装路径
export JAVA_HOME=/usr/local/jvm/jdk1.8.0_144
修改core-site.xml
<configuration>
<!--这里的NN表示两个NameNode组成的逻辑名-->
<property>
<name>fs.defaultFS</name>
<value>hdfs://nn/</value>
<description>The name of the default file system</description>
</property>
<!--指定Hadoop临时目录-->
<property>
<name>hadoop.tmp.dir</name>
<value>/usr/local/hadoop/hadoop-2.7.3/hadoopdata/tmp</value>
<description>A base for other temporary directories</description>
</property>
<!--以下是HDFS HA的配置-->
<property>
<name>ha.zookeeper.quorum</name>
<value>mini1:2181,mini2:2181,mini3:2181</value>
</property>
</configuration>
修改hdfs-site.xml
<configuration>
<property>
<name>dfs.replication</name>
<value>3</value>
</property>
<property>
<name>dfs.permissions</name>
<value>false</value>
</property>
<property>
<name>dfs.namenode.name.dir</name>
<value>/usr/local/hadoop/hadoop-2.7.3/hadoopdata/dfs/name</value>
</property>
<property>
<name>dfs.datanode.data.dir</name>
<value>/usr/local/hadoop/hadoop-2.7.3/hadoopdata/dfs/data</value>
</property>
<!--以下是HDFS HA的配置-->
<!--指定HDFS的nameservices名称为nn,需要和core-site.xml中保持一致-->
<property>
<name>dfs.nameservices</name>
<value>nn</value>
</property>
<!--指定mycluster的两个NameNode的名称分别位ActiveNN和StandbyNN-->
<property>
<name>dfs.ha.namenodes.nn</name>
<value>ActiveNN,StandbyNN</value>
</property>
<!--配置ActiveNN和StandbyNN的rpc通信端口-->
<property>
<name>dfs.namenode.rpc-address.nn.ActiveNN</name>
<value>ActiveNN:9000</value>
</property>
<property>
<name>dfs.namenode.rpc-address.nn.StandbyNN</name>
<value>StandbyNN:9000</value>
</property>
<!--配置ActiveNN和StandbyNN的http通信端口-->
<property>
<name>dfs.namenode.http-address.nn.ActiveNN</name>
<value>ActiveNN:50070</value>
</property>
<property>
<name>dfs.namenode.http-address.nn.StandbyNN</name>
<value>StandbyNN:50070</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://mini1:8485;mini2:8485;mini3:8485/nn</value>
</property>
<!--JournalNode上元数据和日志文件存放位置-->
<property>
<name>dfs.journalnode.edits.dir</name>
<value>/usr/local/hadoop/hadoop-2.7.3/hadoopdata/journal</value>
</property>
<!--开启NameNode失败自动切换-->
<property>
<name>dfs.ha.automatic-failover.enabled</name>
<value>true</value>
</property>
<!--NameNode失败自动切换实现方式-->
<property>
<name>dfs.client.failover.proxy.provider.nn</name>
<value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value>
</property>
<!--隔离机制方式,确保任何时间只有一个NameNode处于活动状态-->
<property>
<name>dfs.ha.fencing.methods</name>
<value>sshfence
shell(/bin/true)</value>
</property>
<!--使用sshfence隔离机制需要SSH免密码认证-->
<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>
</configuration>
修改mapred-site.xml
复制mapred-site.xml.template为mapred-site.xml
<configuration>
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
<!--配置MapReduce JobHistory Server地址,默认端口10020-->
<property>
<name>mapreduce.jobhistory.address</name>
<value>0.0.0.0:10020</value>
<description>MapReduce JobHistory Server IPC host:port</description>
</property>
<!--配置MapReduce JobHistory Server HTTP地址,默认端口19888-->
<property>
<name>mapreduce.jobhistory.webapp.address</name>
<value>0.0.0.0:19888</value>
<description>MapReduce JobHistory Server Web UI host:port</description>
</property>
</configuration>
启动historyserver:mr-jobhistory-daemon.sh start historyserver
停止historyserver:mr-jobhistory-daemon.sh sop historyserver
jobhistoryserver的webUI地址:主机名:19888
当我们启动jobhistoryserver服务之后,在HDFS上/tmp/hadoop-yarn/staging/history路径下会生成两个文件夹:done和done_intermediate。done文件夹下存放已经完成的job,done_intermediate文件夹下存放正在进行的job信息。
修改yarn-site.xml
<configuration>
<!-- Site specific YARN configuration properties -->
<!--启用RM高可用-->
<property>
<name>yarn.resourcemaneger.ha.enabled</name>
<value>true</value>
</property>
<!--指定RM的cluster-id-->
<property>
<name>yarn.resourcemaneger.cluster-id</name>
<value>looc</value>
</property>
<!--指定两台RM主机名标识符-->
<property>
<name>yarn.resourcemaneger.ha.rm-ids</name>
<value>RM1,RM2</value>
</property>
<!--RM主机1-->
<property>
<name>yarn.resourcemaneger.hostname.RM1</name>
<value>ActiveNN</value>
</property>
<!--RM主机2-->
<property>
<name>yarn.resourcemaneger.hostname.RM2</name>
<value>StandbyNN</value>
</property>
<!--RM故障自动切换-->
<property>
<name>yarn.resourcemaneger.ha.automatic-failover.recover.enabled</name>
<value>true</value>
</property>
<!--RM故障自动恢复-->
<property>
<name>yarn.resourcemaneger.recovery.enabled</name>
<value>true</value>
</property>
<!--指定ZK集群地址-->
<property>
<name>yarn.resourcemaneger.zk-address</name>
<value>mini1:2181,mini2:2181,mini3:2181</value>
</property>
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
<!--启用日志聚集功能-->
<property>
<name>yarn.log-aggregation-enable</name>
<value>true</value>
</property>
<property>
<name>yarn.log.server.url</name>
<value>http://0.0.0.0:19888/jobhistory/logs</value>
</property>
<!--向RM调度资源地址-->
<property>
<name>yarn.resourcemaneger.scheduler.address.RM1</name>
<value>ActiveNN:8030</value>
</property>
<property>
<name>yarn.resourcemaneger.scheduler.address.RM2</name>
<value>StandbyNN:8030</value>
</property>
<!--NodeManeger通过该地址交换信息-->
<property>
<name>yarn.resourcemaneger.resource-tracker.address.RM1</name>
<value>ActiveNN:8031</value>
</property>
<property>
<name>yarn.resourcemaneger.resource-tracker.address.RM2</name>
<value>StandbyNN:8031</value>
</property>
<!--客户端通过该地址向RM提交对应用程序的操作-->
<property>
<name>yarn.resourcemaneger.address.RM1</name>
<value>ActiveNN:8032</value>
</property>
<property>
<name>yarn.resourcemaneger.address.RM2</name>
<value>StandbyNN:8032</value>
</property>
<!--管理员通过该地址向RM发送管理命令-->
<property>
<name>yarn.resourcemaneger.admin.address.RM1</name>
<value>ActiveNN:8033</value>
</property>
<property>
<name>yarn.resourcemaneger.admin.address.RM2</name>
<value>StandbyNN:8033</value>
</property>
<!--RM HTTP访问地址,查看集群信息-->
<property>
<name>yarn.resourcemaneger.webapp.address.RM1</name>
<value>ActiveNN:8088</value>
</property>
<property>
<name>yarn.resourcemaneger.webapp.address.RM2</name>
<value>StandbyNN:8088</value>
</property>
</configuration>
修改slaves
mini1
mini2
mini3
配置Hadoop环境变量
#vim /etc/profile
添加如下记录
HADOOP_HOME=/usr/local/hadoop/hadoop-2.7.3
export PATH=$PATH:$HADOOP_HOME/bin
export PATH=$PATH:$HADOOP_HOME/sbin
export HADOOP_CONF_DIR=${HADOOP_HOME}/etc/hadoop
export HADOOP_HDFS_HOME=${HADOOP_HOME}
export HADOOP_YARN_HOME=${HADOOP_HOME}
使配置文件生效
#source /etc/profile
3.2 Hadoop集群的初始化
启动ZooKeeper集群(分别在mini1,mini2以及mini3上执行)
#zkServer.sh start
格式化ZKFC(在ActiveNN上执行)
#cd /usr/local/hadoop/hadoop-2.7.3
#bin/hdfs zkfc -fotmatZK
出现相应的格式化信息
17/08/29 22:29:04 INFO tools.DFSZKFailoverController:
Failover controller configured for NameNode NameNode at ActiveNN/192.168.142.133:9000
17/08/29 22:29:06 INFO zookeeper.ZooKeeper:
Client environment:zookeeper.version=3.4.6-1569965, built on 02/20/2014 09:09 GMT
17/08/29 22:29:06 INFO zookeeper.ZooKeeper:
Client environment:host.name=ActiveNN
17/08/29 22:29:06 INFO zookeeper.ZooKeeper:
Client environment:java.version=1.8.0_144
17/08/29 22:29:06 INFO zookeeper.ZooKeeper:
Client environment:java.vendor=Oracle Corporation
17/08/29 22:29:06 INFO zookeeper.ZooKeeper:
Client environment:java.home=/usr/local/jvm/jdk1.8.0_144/jre
.
.
.
===============================================
The configured parent znode /hadoop-ha/nn already exists.
Are you sure you want to clear all failover information from
ZooKeeper?
WARNING: Before proceeding, ensure that all HDFS services and
failover controllers are stopped!
===============================================
Proceed formatting /hadoop-ha/nn
? (Y or N) 17/08/30 03:36:50 INFO ha.ActiveStandbyElector: Session connected.
Y
17/08/29 22:29:09 INFO ha.ActiveStandbyElector: Recursively deleting /hadoop-ha/nn from ZK...
17/08/29 22:29:09 INFO ha.ActiveStandbyElector: Successfully deleted /hadoop-ha/nn from ZK.
17/08/29 22:29:09 INFO ha.ActiveStandbyElector: Successfully created /hadoop-ha/nn in ZK.
17/08/29 22:29:09 INFO zookeeper.ZooKeeper: Session: 0x25e2f63c4cb0002 closed
17/08/29 22:29:09 INFO zookeeper.ClientCnxn: EventThread shut down
启动journalnode(分别在mini1,mini2以及mini3上执行)
#hadoop-daemon.sh start journalnode
出现如下提示,之后再使用jps查看当前运行着的进程
[root@mini3 ~]# hadoop-daemon.sh start journalnode
starting journalnode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-journalnode-mini3.out
[root@mini3 hadoop]# jps
5956 QuorumPeerMain
6197 JournalNode
6246 Jps
[root@mini3 ~]#
格式化HDFS(在ActiveNN上执行)
#cd /usr/local/hadoop/hadoop-2.7.3
#bin/hdfs namenode -fotmat
出现相应的格式化信息
[root@ActiveNN hadoop-2.7.3]# bin/hdfs namenode -format
17/08/30 00:13:29 INFO namenode.NameNode: STARTUP_MSG:
/************************************************************
STARTUP_MSG: Starting NameNode
STARTUP_MSG: host = ActiveNN/192.168.142.133
STARTUP_MSG: args = [-format]
STARTUP_MSG: version = 2.7.3
.
.
.
17/08/30 00:13:29 INFO namenode.NameNode: registered UNIX signal handlers for [TERM, HUP, INT]
17/08/30 00:13:29 INFO namenode.NameNode: createNameNode [-format]
17/08/30 00:13:30 WARN common.Util: Path /usr/local/hadoop/hadoop-2.7.3/hadoopdata/dfs/name should be sp
17/08/30 00:13:30 WARN common.Util: Path /usr/local/hadoop/hadoop-2.7.3/hadoopdata/dfs/name should be sp
Formatting using clusterid: CID-338fb5d5-c871-46a7-a066-356fd8fd6102
17/08/30 00:13:30 INFO namenode.FSNamesystem: No KeyProvider found.
17/08/30 00:13:30 INFO namenode.FSNamesystem: fsLock is fair:true
17/08/30 00:13:31 INFO blockmanagement.DatanodeManager: dfs.block.invalidate.limit=1000
17/08/30 00:13:31 INFO blockmanagement.DatanodeManager: dfs.namenode.datanode.registration.ip-hostname-c
17/08/30 00:13:31 INFO blockmanagement.BlockManager: dfs.namenode.startup.delay.block.deletion.sec is se
17/08/30 00:13:31 INFO blockmanagement.BlockManager: The block deletion will start around 2017 Aug 30 0
.
.
.
17/08/30 00:13:33 INFO namenode.NNStorageRetentionManager: Going to retain 1 images with txid >= 0
17/08/30 00:13:33 INFO util.ExitUtil: Exiting with status 0
17/08/30 00:13:33 INFO namenode.NameNode: SHUTDOWN_MSG:
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at ActiveNN/192.168.142.133
************************************************************/
将格式化之后的ActiveNN节点hadoop工作目录中的元数据目录复制到StandbyNN节点
[root@ActiveNN ~]# scp -r /usr/local/hadoop/hadoop-2.7.3/hadoopdata/ StandbyNN:/usr/local/hadoop/hadoop-2.7.3/
VERSION 100% 207 0.2KB/s 00:00
seen_txid 100% 2 0.0KB/s 00:00
fsimage_0000000000000000000.md5 100% 62 0.1KB/s 00:00
fsimage_0000000000000000000 100% 351 0.3KB/s 00:00
[root@ActiveNN ~]#
初始化完毕之后可以关闭journalnode(分别在mini1,mini2以及mini3上执行)(之后在ActiveNN上启动dfs会随之启动全部的journalnode)
[root@mini3 ~]# hadoop-daemon.sh stop journalnode
stopping journalnode
[root@mini3 ~]# jps
5956 QuorumPeerMain
6317 Jps
[root@mini3 ~]#
启动 HDFS(在ActiveNN上执行)
[root@ActiveNN ~]# start-all.sh
This script is Deprecated. Instead use start-dfs.sh and start-yarn.sh
Starting namenodes on [ActiveNN StandbyNN]
StandbyNN: starting namenode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-namenode-StandbyNN.out
ActiveNN: starting namenode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-namenode-ActiveNN.out
mini2: starting datanode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-datanode-mini2.out
mini3: starting datanode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-datanode-mini3.out
mini1: starting datanode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-datanode-mini1.out
Starting journal nodes [mini1 mini2 mini3]
mini2: starting journalnode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-journalnode-mini2.out
mini1: starting journalnode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-journalnode-mini1.out
mini3: starting journalnode, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-journalnode-mini3.out
Starting ZK Failover Controllers on NN hosts [ActiveNN StandbyNN]
ActiveNN: starting zkfc, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-zkfc-ActiveNN.out
StandbyNN: starting zkfc, logging to /usr/local/hadoop/hadoop-2.7.3/logs/hadoop-root-zkfc-StandbyNN.out
starting yarn daemons
starting resourcemanager, logging to /usr/local/hadoop/hadoop-2.7.3/logs/yarn-root-resourcemanager-ActiveNN.out
mini1: starting nodemanager, logging to /usr/local/hadoop/hadoop-2.7.3/logs/yarn-root-nodemanager-mini1.out
mini2: starting nodemanager, logging to /usr/local/hadoop/hadoop-2.7.3/logs/yarn-root-nodemanager-mini2.out
mini3: starting nodemanager, logging to /usr/local/hadoop/hadoop-2.7.3/logs/yarn-root-nodemanager-mini3.out
[root@ActiveNN ~]#
启动之后查看每台虚拟机的进程
ActiveNN
[root@ActiveNN ~]# jps
12612 DFSZKFailoverController
12742 ResourceManager
12327 NameNode
13002 Jps
[root@ActiveNN ~]#
StandbyNN
[root@StandbyNN ~]# jps
8196 DFSZKFailoverController
8103 NameNode
8253 Jps
[root@StandbyNN ~]#
mini*
[root@mini* ~]# jps
8692 Jps
8485 JournalNode
8397 DataNode
8573 NodeManager
6830 QuorumPeerMain
[root@mini* ~]# //*表示三台DataNode虚拟机
- 4.Web UI
http://ActiveNN:50070
http://StandbyNN:50070
测试是否为HA,可以先 kill 这个 active NN,然后另外一个 standby NN就会变成 active NN。
[root@ActiveNN ~]# hdfs haadmin -getServiceState ActiveNN
active //此时ActiveNN是active状态
[root@ActiveNN ~]# hdfs haadmin -getServiceState StandbyNN
standby //此时StandbyNN是standby状态
[root@ActiveNN ~]# jps
12612 DFSZKFailoverController
12742 ResourceManager
12327 NameNode
13145 Jps
[root@ActiveNN ~]# kill 12327 //kill掉NameNode进程
[root@ActiveNN ~]# jps
12612 DFSZKFailoverController
12742 ResourceManager
13160 Jps
[root@ActiveNN ~]# hdfs haadmin -getServiceState ActiveNN
17/08/30 00:46:24 INFO ipc.Client: Retrying connect to server: ActiveNN/192.168.142.133:9000. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=1, sleepTime=1000 MILLISECONDS)
Operation failed: Call From ActiveNN/192.168.142.133 to ActiveNN:9000 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused
//之前active的NameNode现在死掉了,可以使用hadoop-daemon.sh start namenode启动
[root@ActiveNN ~]# hdfs haadmin -getServiceState StandbyNN
active //此时StandbyNN变为了active状态
[root@ActiveNN ~]#
http://ActiveNN:50070
http://StandbyNN:50070