转载请注明出处:http://blog.csdn.net/l1028386804/article/details/51353063
集群规划:
主机名 IP 安装的软件 运行的进程
liuyazhuang01 192.168.1.201 jdk、hadoop NameNode、DFSZKFailoverController
liuyazhuang02 192.168.1.202 jdk、hadoop NameNode、DFSZKFailoverController
liuyazhuang03 192.168.1.203 jdk、hadoop ResourceManager
liuyazhuang04 192.168.1.204 jdk、hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMain
liuyazhuang05 192.168.1.205 jdk、hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMain
liuyazhuang06 192.168.1.206 jdk、hadoop、zookeeper DataNode、NodeManager、JournalNode、QuorumPeerMain
在hadoop2.0中通常由两个NameNode组成,一个处于active状态,另一个处于standby状态。Active NameNode对外提供服务,而Standby NameNode则不对外提供服务,仅同步active namenode的状态,以便能够在它失败时快速进行切换。
hadoop2.0官方提供了两种HDFS HA的解决方案,一种是NFS,另一种是QJM。这里我们使用简单的QJM。在该方案中,主备NameNode之间通过一组JournalNode同步元数据信息,一条数据只要成功写入多数JournalNode即认为写入成功。通常配置奇数个JournalNode
这里还配置了一个zookeeper集群,用于ZKFC(DFSZKFailoverController)故障转移,当Active NameNode挂掉了,会自动切换Standby NameNode为standby状态
tar -zxvf zookeeper-3.4.5.tar.gz -C /liuyazhuang/
cd /liuyazhuang/zookeeper-3.4.5/conf/
cp zoo_sample.cfg zoo.cfg
vim zoo.cfg
修改:dataDir=/liuyazhuang/zookeeper-3.4.5/tmp
在最后添加:
server.1=liuyazhuang04:2888:3888
server.2=liuyazhuang05:2888:3888
server.3=liuyazhuang06:2888:3888
保存退出
然后创建一个tmp文件夹
mkdir /liuyazhuang/zookeeper-3.4.5/tmp
再创建一个空文件
touch /liuyazhuang/zookeeper-3.4.5/tmp/myid
最后向该文件写入ID
echo 1 > /liuyazhuang/zookeeper-3.4.5/tmp/myid
将配置好的zookeeper拷贝到其他节点(首先分别在liuyazhuang05、liuyazhuang06根目录下创建一个liuyazhuang目录:mkdir /liuyazhuang)
scp -r /liuyazhuang/zookeeper-3.4.5/ liuyazhuang05:/liuyazhuang/
scp -r /liuyazhuang/zookeeper-3.4.5/ liuyazhuang06:/liuyazhuang/
注意:修改liuyazhuang05、liuyazhuang06对应/liuyazhuang/zookeeper-3.4.5/tmp/myid内容
liuyazhuang05:
echo 2 > /liuyazhuang/zookeeper-3.4.5/tmp/myid
liuyazhuang06:
echo 3 > /liuyazhuang/zookeeper-3.4.5/tmp/myid
tar -zxvf hadoop-2.2.0.tar.gz -C /liuyazhuang/
#将hadoop添加到环境变量中
vim /etc/profile
export JAVA_HOME=/usr/java/jdk1.7.0_55
export HADOOP_HOME=/liuyazhuang/hadoop-2.2.0
export PATH=$PATH:$JAVA_HOME/bin:$HADOOP_HOME/bin
#hadoop2.0的配置文件全部在$HADOOP_HOME/etc/hadoop下
cd /liuyazhuang/hadoop-2.2.0/etc/hadoop
export JAVA_HOME=/usr/java/jdk1.7.0_55
fs.defaultFS
hdfs://ns1
hadoop.tmp.dir
/liuyazhuang/hadoop-2.2.0/tmp
ha.zookeeper.quorum
liuyazhuang04:2181,liuyazhuang05:2181,liuyazhuang06:2181
dfs.nameservices
ns1
dfs.ha.namenodes.ns1
nn1,nn2
dfs.namenode.rpc-address.ns1.nn1
liuyazhuang01:9000
dfs.namenode.http-address.ns1.nn1
liuyazhuang01:50070
dfs.namenode.rpc-address.ns1.nn2
liuyazhuang02:9000
dfs.namenode.http-address.ns1.nn2
liuyazhuang02:50070
dfs.namenode.shared.edits.dir
qjournal://liuyazhuang04:8485;liuyazhuang05:8485;liuyazhuang06:8485/ns1
dfs.journalnode.edits.dir
/liuyazhuang/hadoop-2.2.0/journal
dfs.ha.automatic-failover.enabled
true
dfs.client.failover.proxy.provider.ns1
org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
dfs.ha.fencing.methods
sshfence
shell(/bin/true)
dfs.ha.fencing.ssh.private-key-files
/root/.ssh/id_rsa
dfs.ha.fencing.ssh.connect-timeout
30000
mapreduce.framework.name
yarn
yarn.resourcemanager.hostname
liuyazhuang03
yarn.nodemanager.aux-services
mapreduce_shuffle
修改slaves(slaves是指定子节点的位置,因为要在liuyazhuang01上启动HDFS、在liuyazhuang03启动yarn,所以liuyazhuang01上的slaves文件指定的是datanode的位置,liuyazhuang03上的slaves文件指定的是nodemanager的位置)
liuyazhuang04
liuyazhuang05
liuyazhuang06
#首先要配置liuyazhuang01到liuyazhuang02、liuyazhuang03、liuyazhuang04、liuyazhuang05、liuyazhuang06的免密码登陆
#在liuyazhuang01上生产一对钥匙
ssh-keygen -t rsa
#将公钥拷贝到其他节点,包括自己
ssh-coyp-id liuyazhuang01
ssh-coyp-id liuyazhuang02
ssh-coyp-id liuyazhuang03
ssh-coyp-id liuyazhuang04
ssh-coyp-id liuyazhuang05
ssh-coyp-id liuyazhuang06
#配置liuyazhuang03到liuyazhuang04、liuyazhuang05、liuyazhuang06的免密码登陆
#在liuyazhuang03上生产一对钥匙
ssh-keygen -t rsa
#将公钥拷贝到其他节点
ssh-coyp-id liuyazhuang04
ssh-coyp-id liuyazhuang05
ssh-coyp-id liuyazhuang06
#注意:两个namenode之间要配置ssh免密码登陆,别忘了配置liuyazhuang02到liuyazhuang01的免登陆
在liuyazhuang02上生产一对钥匙
ssh-keygen -t rsa
ssh-coyp-id -i liuyazhuang01
scp -r /liuyazhuang/ liuyazhuang02:/
scp -r /liuyazhuang/ liuyazhuang03:/
scp -r /liuyazhuang/hadoop-2.2.0/ root@liuyazhuang04:/liuyazhuang/
scp -r /liuyazhuang/hadoop-2.2.0/ root@liuyazhuang05:/liuyazhuang/
scp -r /liuyazhuang/hadoop-2.2.0/ root@liuyazhuang06:/liuyazhuang/
###注意:严格按照下面的步骤
(分别在liuyazhuang04、liuyazhuang05、liuyazhuang06上启动zk)
cd /liuyazhuang/zookeeper-3.4.5/bin/
./zkServer.sh start
#查看状态:一个leader,两个follower
./zkServer.sh status
(在liuyazhuang01上启动所有journalnode,注意:是调用的hadoop-daemons.sh这个脚本,注意是复数s的那个脚本)
cd /liuyazhuang/hadoop-2.2.0
sbin/hadoop-daemons.sh start journalnode
#运行jps命令检验,liuyazhuang04、liuyazhuang05、liuyazhuang06上多了JournalNode进程
#在liuyazhuang01上执行命令:
hdfs namenode -format
#格式化后会在根据core-site.xml中的hadoop.tmp.dir配置生成个文件,这里我配置的是/liuyazhuang/hadoop-2.2.0/tmp,然后将/liuyazhuang/hadoop-2.2.0/tmp拷贝到liuyazhuang02的/liuyazhuang/hadoop-2.2.0/下。
scp -r tmp/ liuyazhuang02:/liuyazhuang/hadoop-2.2.0/
hdfs zkfc -formatZK
sbin/start-dfs.sh
(#####注意#####:是在liuyazhuang03上执行start-yarn.sh,把namenode和resourcemanager分开是因为性能问题,因为他们都要占用大量资源,所以把他们分开了,他们分开了就要分别在不同的机器上启动)
sbin/start-yarn.sh
http://192.168.1.201:50070
NameNode 'liuyazhuang01:9000' (active)
http://192.168.1.202:50070
NameNode 'liuyazhuang02:9000' (standby)
首先向hdfs上传一个文件
hadoop fs -put /etc/profile /profile
hadoop fs -ls /
然后再kill掉active的NameNode
kill -9
通过浏览器访问:http://192.168.1.202:50070
NameNode 'liuyazhuang02:9000' (active)
这个时候liuyazhuang02上的NameNode变成了active
在执行命令:
hadoop fs -ls /
-rw-r--r-- 3 root supergroup 1926 2014-02-06 15:36 /profile
刚才上传的文件依然存在!!!
手动启动那个挂掉的NameNode
sbin/hadoop-daemon.sh start namenode
通过浏览器访问:http://192.168.1.201:50070
NameNode 'liuyazhuang01:9000' (standby)
运行一下hadoop提供的demo中的WordCount程序:
hadoop jar share/hadoop/mapreduce/hadoop-mapreduce-examples-2.2.0.jar wordcount /profile /out
OK,大功告成!!!