经过了一些细节调整，完全分布式环境算是实测通过了，现就自己搭建的环境的启动步骤做一些总结与说明

三台主机构成分布式集群环境：

• master 主节点
• slave01 子节点1
• slave02 子节点2

启动步骤

1. 在master、slave01、slave02上分别启动Zookeeper zkServer.sh start

2. 在主节点上运行 start-dfs.sh

3. 再在slave01上运行 start-yarn.sh

4. 再在slave02上单独启动resourcemanager yarn-daemon.sh start resourcemanager

5. 在master上启动HBase start-hbase.sh

6. 在slave01、slave02上分别启动HMaster进程 hbase-daemon.sh start master

   它会自动根据先前博客那样配置的文件，在集群中的各个机器上启动服务进程，具体见下面详细说明

各节点的进程

–> [elon@master[slave01、slave02] ~]$ zkServer.sh start
JMX enabled by default
Using config: /home/elon/opt/modules/zookeeper-3.4.5/bin/…/conf/zoo.cfg
Starting zookeeper … STARTED

–> [elon@master ~]$ start-dfs.sh
Starting namenodes on [master slave01]
slave01: starting namenode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-namenode-slave01.out
master: starting namenode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-namenode-master.out
slave02: starting datanode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-datanode-slave02.out
master: starting datanode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-datanode-master.out
slave01: starting datanode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-datanode-slave01.out
Starting journal nodes [master slave01 slave02]
slave02: starting journalnode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-journalnode-slave02.out
master: starting journalnode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-journalnode-master.out
slave01: starting journalnode, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-journalnode-slave01.out
Starting ZK Failover Controllers on NN hosts [master slave01]
master: starting zkfc, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-zkfc-master.out
slave01: starting zkfc, logging to /home/elon/app/hadoop-2.4.1/logs/hadoop-elon-zkfc-slave01.out

–> [elon@slave01 ~]$ start-yarn.sh
starting yarn daemons
starting resourcemanager, logging to /home/elon/app/hadoop-2.4.1/logs/yarn-elon-resourcemanager-slave01.out
slave02: starting nodemanager, logging to /home/elon/app/hadoop-2.4.1/logs/yarn-elon-nodemanager-slave02.out
master: starting nodemanager, logging to /home/elon/app/hadoop-2.4.1/logs/yarn-elon-nodemanager-master.out
slave01: starting nodemanager, logging to /home/elon/app/hadoop-2.4.1/logs/yarn-elon-nodemanager-slave01.out

–> [elon@slave02 ~]$ yarn-daemon.sh start resourcemanager
starting resourcemanager, logging to /home/elon/app/hadoop-2.4.1/logs/yarn-elon-resourcemanager-slave02.out
You have new mail in /var/spool/mail/elon

–> [elon@master ~]$ start-hbase.sh
starting master, logging to /home/elon/opt/modules/hbase-0.96.2-hadoop2/logs/hbase-elon-master-master.out
slave02: starting regionserver, logging to /home/elon/opt/modules/hbase-0.96.2-hadoop2/bin/…/logs/hbase-elon-regionserver-slave02.out
slave01: starting regionserver, logging to /home/elon/opt/modules/hbase-0.96.2-hadoop2/bin/…/logs/hbase-elon-regionserver-slave01.out

–> [elon@slave01[slave02] ~]$ hbase-daemon.sh start master
starting master, logging to /home/elon/opt/modules/hbase-0.96.2-hadoop2/logs/hbase-elon-master-slave01.out

进程启动明细

    **master：**
        [elon@master ~]$ jps
        5873 JournalNode
        5589 NameNode
        5688 DataNode
        6268 Jps
        6157 NodeManager
        6043 DFSZKFailoverController
        5430 QuorumPeerMain

    **slave01：**
        [elon@slave01 ~]$ jps
        3990 JournalNode
        4303 NodeManager
        3897 DataNode
        3828 NameNode
        4925 HMaster
        4104 DFSZKFailoverController
        3771 QuorumPeerMain
        4990 Jps
        4201 ResourceManager

    **slave02:**
        [elon@slave02 ~]$ jps
        3990 HMaster
        4064 Jps
        3779 HRegionServer
        3674 ResourceManager
        3287 QuorumPeerMain
        3434 JournalNode
        3344 DataNode
        3530 NodeManager

集群Web控制台界面

• NameNode

  • master:9000 (active) http://master:50070/dfshealth.html
  • slave01:9000 (standby) http://slave01:50070/dfshealth.html

• Resourcemanager

  • All Applications http://slave01:8088/cluster
  • 当登陆slave02的resourcemanager控制台时 http://slave02:8088/cluster 会显示：
    This is standby RM. Redirecting to the current active RM: http://slave01:8088/cluster

• HBase

  • HBase Master http://master:60010/master-status

• Hive

  • Hive的启动方式 http://elon33.com/2017/Hive/#Hive的启动方式
  • Hive Web Interface http://slave02:9999/hwi/