zookeeper

1  修改 /etc/hosts
192.168.1.10    nn01
192.168.1.20    nn02
192.168.1.11    node1 (zookeeper)
192.168.1.12    node2 (zookeeper)
192.168.1.13    node3 (zookeeper)

2  配置 nn02 ssh 免密码登录，包含本机
/etc/ssh/ssh_config   修改  StrictHostKeyChecking no

3  停止 hadoop 集群，及其  kafka，并初始化 hdfs 集群
rm  -rf  /var/hadoop/*
nn02  创建  mkdir /var/hadoop

4  验证 zookeeper 集群
5  nn02 禁用 selinux ，卸载 firewalld ,安装 java openjdk
#----------------------------------------------------------------------#
目录  /usr/local/hadoop/etc/hadoop
配置文件 hadoop-env.sh
JAVA_HOME
HADOOP_CONF_DIR

配置文件 slaves
node1
node2
node3

xml 文件配置格式
    <property>
        <name></name>
        <value></value>
    </property>

配置文件 core-stie.xml
n: fs.defaultFS
v: hdfs://nsd1802
n: hadoop.tmp.dir
v: /var/hadoop
n: ha.zookeeper.quorum
v: node1:2181,node2:2181,node3:2181

配置文件 hdfs-site.xml
n: dfs.nameservices
v: nsd1802
n: dfs.ha.namenodes.nsd1802
v: nn1,nn2
n: dfs.namenode.rpc-address.nsd1802.nn1
v: nn01:8020
n: dfs.namenode.rpc-address.nsd1802.nn2
v: nn02:8020
n: dfs.namenode.http-address.nsd1802.nn1
v: nn01:50070
n: dfs.namenode.http-address.nsd1802.nn2
v: nn02:50070
n: dfs.namenode.shared.edits.dir
v: qjournal://node1:8485;node2:8485;node3:8485/nsd1802
n: dfs.journalnode.edits.dir
v: /var/hadoop/journal
n: dfs.client.failover.proxy.provider.nsd1802
v: org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider
n: dfs.ha.fencing.methods
v: sshfence
n: dfs.ha.fencing.ssh.private-key-files
v: /root/.ssh/id_rsa
n: dfs.ha.automatic-failover.enabled
v: true
n: dfs.replication
v: 2

配置文件 mapred-site.xml
n: mapreduce.framework.name
v: yarn

配置文件 yarn-site.xml
n: yarn.resourcemanager.ha.enabled
v: true
n: yarn.resourcemanager.ha.rm-ids
v: rm1,rm2
n: yarn.resourcemanager.recovery.enabled
v: true
n: yarn.resourcemanager.store.class
v: org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore
n: yarn.resourcemanager.zk-address
v: node1:2181,node2:2181,node3:2181
n: yarn.resourcemanager.cluster-id
v: yarn-ha
n: yarn.resourcemanager.hostname.rm1
v: nn01
n: yarn.resourcemanager.hostname.rm2
v: nn02
n: yarn.nodemanager.aux-services
v: mapreduce_shuffle

配置文件下载地址
http://118.144.89.240/hadoop_conf.tar.gz

#------------------------------------------------------------------------#
ALL: 所有机器
nodeX： node1    node2    node3

ALL:  同步配置到所有集群机器
NN1: 初始化ZK集群  ./bin/hdfs zkfc -formatZK
nodeX:  启动 journalnode 服务
        ./sbin/hadoop-daemon.sh start journalnode

NN1: 格式化  ./bin/hdfs  namenode  -format

NN2: 数据同步到本地 /var/hadoop/dfs

NN1: 初始化 JNS
        ./bin/hdfs namenode -initializeSharedEdits

nodeX: 停止 journalnode 服务
        ./sbin/hadoop-daemon.sh stop journalnode

#----------------------------------------------------------------#
启动集群
NN1: ./sbin/start-all.sh
NN2: ./sbin/yarn-daemon.sh start resourcemanager

查看集群状态
./bin/hdfs haadmin -getServiceState nn1  
./bin/hdfs haadmin -getServiceState nn2
./bin/yarn rmadmin -getServiceState rm1
./bin/yarn rmadmin -getServiceState rm2

./bin/hdfs dfsadmin -report
./bin/yarn  node  -list

访问集群：
./bin/hadoop  fs -ls  /
./bin/hadoop  fs -mkdir hdfs://nsd1802/input

验证高可用，关闭 active namenode
./sbin/hadoop-daemon.sh stop namenode