hadoop+zookeeper集群高可用搭建
Senerity 发布于 2016/12/12 09:19
Hadoop集群搭建步骤
1. 架构图
2. 准备5台机器
centosa: 192.168.42.128
centosb: 192.168.42.129
centosc: 192.168.42.130
centosd: 192.168.42.133
centose: 192.168.42.134
namenode: centosa, centosb
datanode: centosc, centosd, centose
journalnode: centosc, centosd, centose
ResourceManager: centosb
NodeManager: centosc, centosd, centose
zookeeper:centosc ,centosd ,centose
3. 5台机器分别修改主机名和IP的映射
1) 修改主机名
2) IP映射
4. 所有机器关闭防火墙
[root@centosa ~]# service iptables stop
...
5. 安装JDK配置JAVA_HOME环境变量(推荐 .bashrc)
...
6. 所有机器彼此做SSH免密码登陆
方法一，手动配置:
1) 在centosa机器生成公私密钥对
[root@centosa ~]# ssh-keygen -t dsa -P "" -f ~/.ssh/id_dsa
2) 把公钥上传给其它机器
[root@centosa ~]# scp .ssh/id_dsa.pub centosb: ~/
[root@centosa ~]# scp .ssh/id_dsa.pub centosc: ~/
[root@centosa ~]# scp .ssh/id_dsa.pub centosd: ~/
[root@centosa ~]# scp .ssh/id_dsa.pub centose: ~/
3) 在所有机器上把公钥添加到自己的信任列表
centosa:
[root@centosa ~]# cat ~/.ssh/id_dsa.pub>> .ssh/authorized_keys
centosb:
[root@centosb ~]# cat ~/id_dsa.pub>> .ssh/authorized_keys
centosc:
[root@centosc ~]# cat ~/id_dsa.pub>> .ssh/authorized_keys
centosd:
[root@centosd~]# cat ~/id_dsa.pub>> .ssh/authorized_keys
centose:
[root@centose ~]# cat ~/id_dsa.pub>> .ssh/authorized_keys
（到此，centosa可以免密码登录自己和其它机器）
4) 在其它机器上重复以上(1)(2)(3)步
方法二，使用脚本(推荐)，脚本如下:
|shellscripts
|bin
|add_auth.sh:
#!/bin/bash
cat ~/id_dsa.pub >> ~/.ssh/authorized_keys
rm -rf ~/id_dsa.pub
exit
|ssh.sh:
#!/bin/bash
rm -rf ~/.ssh
ssh-keygen -t dsa -P '' -f ~/.ssh/id_dsa
cat ~/.ssh/id_dsa.pub >> ~/.ssh/authorized_keys
exit
|sbin
|add_auth.sh:
#!/bin/bash
for i in $@; do
if [ ! $HOSTNAME = $i ] ;then
echo 'upload id_dsa.pub file to '$i'~/'
scp ~/.ssh/id_dsa.pub $i:~/
echo 'login on '$i
ssh root@$i ./shellscripts/bin/add_auth.sh
fi
done
|ssh.sh:
#!/bin/bash
rm -rf .ssh
for i in $@; do
echo ' Add authentication files to other files...'
ssh root@$i ./shellscripts/bin/ssh.sh
done
1) 把脚本文件夹shellscripts上传到centosa
2) 修改脚本权限
[root@centosa ~]# chmod -R 777 ~/shellscripts/
3) 把shellscripts上传给其它机器
[root@centosa ~]# scp -r ~/shellscripts centosb:~/
[root@centosa ~]# scp -r ~/shellscripts centosc:~/
[root@centosa ~]# scp -r ~/shellscripts centosd:~/
[root@centosa ~]# scp -r ~/shellscripts centose:~/
4) 执行脚本, 实现自己免登录自己,只需要在一台机器上执行该命令即可
[root@centosa ~]#
./shellscripts/sbin/ssh.sh centosa centosb centosc centosd centose
5) 执行脚本, 实现自己免登录其它机器，需要在各机器上分别执行该命令
[root@centosa ~]#
./shellscripts/sbin/add_auth.sh centosb centosc centosd centose
[root@centosb ~]#
./shellscripts/sbin/add_auth.sh centosa centosc centosd centose
[root@centosc ~]#
./shellscripts/sbin/add_auth.sh centosa centosb centosd centose
[root@centosd ~]#
./shellscripts/sbin/add_auth.sh centosa centosb centosc centose
[root@centose ~]#
./shellscripts/sbin/add_auth.sh centosa centosb centosc centosd
7. 解压hadoop-2.6.0到/usr
1) 配置core-site.xml(注意: hadoop配置文件不支持中文注释，请删除)
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop /core-site.xml
<property>
<name>fs.defaultFS</name>
<value>hdfs://mycluster</value> //mycluster是对namenode做命名服务
</property>
<property>
<name>fs.trash.interval</name> //开启回收站
<value>10</value>
</property>
<property>
<name>hadoop.tmp.dir</name>
<value>/usr/hadoop-2.6.0/tmp-${user.name}</value>
</property>
<property>
<name>net.topology.script.file.name</name>
<value>/usr/hadoop-2.6.0/etc/hadoop/rack.sh</value>//机架，需要创建
</property>
创建机架文件:
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/rack.sh
while [ $# -gt 0 ] ; do
nodeArg=$1
exec</usr/hadoop-2.6.0/etc/hadoop/topology.data
result=""
while read line ; do
ar=( $line )
if [ "${ar[0]}" = "$nodeArg" ] ; then
result="${ar[1]}"
fi
done
shift
if [ -z "$result" ] ; then
echo -n "/default_rack"
else
echo -n "$result "
fi
done
给机架添加可执行权限:
[root@centosa ~]# chmod u+x /usr/hadoop-2.6.0/etc/hadoop/rack.sh
创建topology.data文件
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/topology.data
测试机架:
[root@centosa hadoop-2.6.0]# ./etc/hadoop/rack.sh 192.168.42.130
2) 配置hdfs-site.xml
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/hdfs-site.xml
<property>
<name>dfs.replication</name>
<value>3</value>
</property>
<property>
<name>dfs.nameservices</name>
<value>mycluster</value>
</property>
<property>
< name>dfs.ha.namenodes.mycluster</name>
<value>nn1,nn2</value>
</property>
<property>
< name>dfs.namenode.rpc-address.mycluster.nn1</name>
<value>centosa:8020</value>
</property>
<property>
< name>dfs.namenode.rpc-address.mycluster.nn2</name>
<value>centosb:8020</value>
</property>
<property>
< name>dfs.namenode.http-address.mycluster.nn1</name>
<value>centosa:50070</value>
</property>
<property>
< name>dfs.namenode.http-address.mycluster.nn2</name>
<value>centosb:50070</value>
</property>
<property>
< name>dfs.namenode.shared.edits.dir</name>
<value>
qjournal://centosb:8485;centosc:8485;centosd:8485/mycluster
</value>
</property>
<property>
< name>dfs.client.failover.proxy.provider.mycluster</name>
<value>
org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
</value>
</property>
<property>
< name>dfs.ha.fencing.methods</name>
<value>sshfence</value>
</property>
<property>
< name>dfs.ha.fencing.ssh.private-key-files</name>
<value>/root/.ssh/id_dsa</value>
</property>
3) 配置slaves文件
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/slaves
centosc
centosd
centose
4) 配置yarn-site.xml
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/yarn-site.xml
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
<property>
<name>yarn.resourcemanager.hostname</name>
<value>centosb</value>
</property>
5) 配置mapred-site.xml
[root@centosa ~]# vi /usr/hadoop-2.6.0/etc/hadoop/mapred-site.xml
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
6) 把./etc/hadoop/*下的所有文件同步到其它机器
[root@centosa hadoop-2.6.0]#
scp ./etc/hadoop/* centosb:/usr/hadoop-2.6.0/etc/hadoop/
scp ./etc/hadoop/* centosc:/usr/hadoop-2.6.0/etc/hadoop/
scp ./etc/hadoop/* centosd:/usr/hadoop-2.6.0/etc/hadoop/
scp ./etc/hadoop/* centose:/usr/hadoop-2.6.0/etc/hadoop/
到此，配置完成。
8. 启动步骤
1) 先启动centosc centosd centose 上的journalnode
[root@centosc hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
[root@centosd hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
[root@centose hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
2) 格式化centosa上的namenode
[root@centosa hadoop-2.6.0]# ./bin/hdfs namenode -format
3) 启动centosa上的namenode
[root@centosa hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start namenode
4) 引导格式化centosb上的namenode
[root@centosb hadoop-2.6.0]# ./bin/hdfs namenode bootstrapStandby
5) 启动centosb上的namenode
[root@centosb hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start namenode
6) 启动centosc centosd centose上的dataname
[root@centosc hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
[root@centosd hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
[root@centose hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
到此为止：HDFS的集群搭建成功，但是默认情况下namenode都处于standby模式
该模式下是不允许用户对hdfs做任何修改动作。
9. [root@centosa hadoop-2.6.0]# ./bin/hdfs haadmin –help
Usage: DFSHAAdmin [-ns <nameserviceId>]
[-transitionToActive< serviceId> [--forceactive]]
[-transitionToStandby< serviceId>]
[-failover [--forcefence] [--forceactive] <serviceId> <serviceId>]
[-getServiceState< serviceId>]
[-checkHealth< serviceId>]
[-help <command>]
1) 让centosa(namenode nn1)变成Active活跃节点
[root@centosa hadoop-2.6.0]# ./bin/hdfs haadmin -transitionToActive nn1
目前 centosa(nn1)处于Active模式, centosb(nn2)处于Standby模式
2) 将centosa变成Standy模式 centosb 变成Active模式
[root@centosa hadoop-2.6.0]# ./bin/hdfs haadmin -failover --forceactive nn1 nn2
10. 在centosb上启动 yarn
[root@centosb hadoop-2.6.0]# ./sbin/start-yarn.sh
11. 停止hadoop集群
1) 在centosa上执行./sbin/stop-dfs.sh
2) 在centosb上执行./sbin/stop-yarn.sh
12. 搭建zookeeper (centosc ,centosd ,centose)
1) 上传zookeeper-3.4.6.tar.gz到centosc,centosd,centose
2) 解压zookeeper-3.4.6.tar.gz
[root@centosc ~]# tar -zxf zookeeper-3.4.6.tar.gz -C /usr
[root@centosd ~]# tar -zxf zookeeper-3.4.6.tar.gz -C /usr
[root@centose ~]# tar -zxf zookeeper-3.4.6.tar.gz -C /usr
3) 在centosc上配置zoo.cfg
4) 同步./conf/zoo.cfg文件到centosd, centose
[root@centosc zookeeper-3.4.6]# scp ./conf/zoo.cfg centosd:/usr/zookeeper-3.4.6/conf/
[root@centosc zookeeper-3.4.6]# scp ./conf/zoo.cfg centose:/usr/zookeeper-3.4.6/conf/
5) 创建/usr/zookeeper-3.4.6/zkdata目录
[root@centosc zookeeper-3.4.6]# mkdir /usr/zookeeper-3.4.6/zkdata
[root@centosd zookeeper-3.4.6]# mkdir /usr/zookeeper-3.4.6/zkdata
[root@centose zookeeper-3.4.6]# mkdir /usr/zookeeper-3.4.6/zkdata
6) 分别在centosc, centosd, centose上执行：
echo '服务器编号'>> /usr/zookeeper/zkdata/myid
[root@centosc zookeeper-3.4.6]# echo '1' >> ./zkdata/myid
[root@centosd zookeeper-3.4.6]# echo '2' >> ./zkdata/myid
[root@centose zookeeper-3.4.6]# echo '3' >> ./zkdata/myid
13. 在前面的基础上,在hadoop的hdfs-site.xml中添加如下配置,并同步到其它机器
[root@centosa hadoop-2.6.0]# vi ./etc/hadoop/hdfs.site.xml
<property>
<name>dfs.ha.automatic-failover.enabled</name>
<value>true</value>
</property>
<property>
<name>ha.zookeeper.quorum</name>
<value>centosc:2181,centosd:2181,centose:2181</value>
</property>
同步到其它机器:
到此, 集群配置完成。
14. 集群启动一
1) 首先启动zookeeper
[root@centosc zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
[root@centosd zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
[root@centose zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
2) 在centosa上执行./sbin/start-dfs.sh
[root@centosd hadoop-2.6.0]# ./sbin/start-dfs.sh
此时namenode, datanode, journalnode, zkfc会自动启动,但是ZK还没有启动
3) 注册namenode到zookeeper 只需要在任意一台namenode上执行
[root@centosa hadoop-2.6.0]# ./bin/hdfs zkfc –formatZK
4) 在centosa上执行./sbin/stop-dfs.sh 再重新启动,此时ZK才会启动
[root@centosd hadoop-2.6.0]# ./sbin/stop-dfs.sh
[root@centosd hadoop-2.6.0]# ./sbin/start-dfs.sh
5) 启动yarn, 必须登录centosb启动
[root@centosb hadoop-2.6.0]# ./sbin/start-yarn.sh
如果集群启动不成功，可能是namenode数据不一致导致，请登录centosa,centosb删除/usr/hadoop-2.6.0/tmp-${user.name}目录之后，参考集群启动二。
15. 集群启动二
1) 分别登陆centosc, centosd, centose启动zookeeper服务
[root@centosc zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
[root@centosd zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
[root@centose zookeeper-3.4.6]# ./bin/zkServer.sh start zoo.cfg
2) 分别登陆centosc,centosd,centose启动journalnode
[root@centosc hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
[root@centosd hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
[root@centose hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start journalnode
3) 格式化centosa上的namenode
[root@centosa hadoop-2.6.0]# ./bin/hdfs namenode –format
4) 启动centosa上的namenode
[root@centosa hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start namenode
5) 引导格式化centosb上的namenode
[root@centosb hadoop-2.6.0]# ./bin/hdfs namenode bootstrapStandby
6) 启动centosb上的namenode
[root@centosb hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start namenode
7) 注册namenode到zookeeper 只需要在任意一台namenode上执行
[root@centosa hadoop-2.6.0]# ./bin/hdfs zkfc -formatZK
8) 分别在两个namenode上启动zkfc监视器
[root@centosa hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start zkfc
[root@centosb hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start zkfc
9) 分别登陆centosc,centosd,centose启动datanode
[root@centosc hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
[root@centosd hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
[root@centose hadoop-2.6.0]# ./sbin/hadoop-daemon.sh start datanode
10) 启动yarn, 必须登录centosb启动
[root@centosb hadoop-2.6.0]# ./sbin/start-yarn.sh
到此,zookeeper+hadoop集群搭建完毕