[集群自动化四] 来用Docker装个HADOOP集群

前言：

本篇写一个Hadoop自动安装脚本，以5台KVM虚拟机和docker为基础，来自动安装Hadoop集群（HDFS、YARN、HBASE、HIVE、SPARK、HUE、Jupyter-notebook、Scope等）

各组件版本

HDFS 3.1.1

YARN 3.1.1

HBASE 2.1.1

HIVE 3.1.1

SPARK 2.4.0

HUE 4.3.0

Jupyter-notebook 4.4.0

Scope 1.10.1

Mysql：mariadb 10.3.11

Solr: 7.6.0

Livy: 0.5.0

OOZIE: 5.0.0

Myweb: 自编，服务索引与服务状态监控

架构与服务分配

nn1	nn2	dn1	dn2	dn3
		zookeeper1	zookeeper2	zookeeper3
		JournalNode	JournalNode	JournalNode
NameNode	NameNode	DataNode	DataNode	DataNode
NodeManager	NodeManager	NodeManager	NodeManager	NodeManager
ResourceManager	ResourceManager
HbaseMaster&Rest&Thrift	HiveMetadata&HiveServer2	HbaseRegion	HbaseRegion	HbaseRegion
	SparkMaster	SparkWorker	SparkWorker	SparkWorker
YarnHistory&WebProxy	SparkHistory&Livy&Solr&Oozie
Myweb	Hue&Mysql&Jupyter
WeaveScope	WeaveScope	WeaveScope	WeaveScope	WeaveScope

[图片上传失败...(image-26fe0f-1549602339382)]

准备

克隆5台虚拟机，详见前篇[基础架构十四：来试试把Centos7改造成Coreos]，nn2需要调整到8G内存
下载所需docker镜相（包含9个安装用到的docker image, 解压到下一步clone的目录里）

配置

cd hadoopHa-autoins; vim CONFIG #修改5台机器的IP

4.png

安装

sh install all # 大约10分钟后完成安装

3.png

验证

1.png

服务端口状态

2.png

HDFS

5.png

YARN

6.png

HBASE

7.png

SPARK

8.png

HUE

10.png

Solr

9.png

Oozie

13.png

Jupyter-notebook

11.png

WeaveScope

12.png

install.sh 更多用法：

usage: install.sh [option]

option:

 p0|pimages     :cp&load all tgz&images to all nodes.
 p|pconfig      :cp config&shell to all nodes.
 zookeeper      :install zookeeper cluster on ZKX
 journalnode    :install JN on JNX
 format         :format ZKFC&Nodename on nn1
 startnn1       :start NN/ZKFC/RM on nn1
 standby        :sync namenode info on nn2
 startnn2       :start NN/ZKFC/RM on nn2
 datanode       :start datanode on DNX
 nodemanager    :start nodemanager on all nodes
 hmaster        :HBASE master
 hregion        :HBASE region
 spark          :start spark master
 sparkslave     :start spark slaves
 oozie          :oozie for schedule jobs
 hue            :HUE manager page
 scope          :weavescope monitor
 myweb          :index for all services
 genindex       :generate svc-hadoop.html 
 finish         :print finish page
 status         :get status of NameNode&ResourceManager(NN&RM) & zookeeper
 timezone8      :set timezone CST-8
 route          :add route temporally
 all            :install all units