Arm64v8 cpu + Centos7 aarch64中安装 Ambari 2.7.3 和 HDP 3.1.0

# 下载不存在的资源的方法

使用迅雷云盘，添加下载任务到云盘，有一定的概率下载到已经被删除的资源。
比如下载HDP相关的资源：
<http://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/HDP-GPL/3.x/updates/3.1.0.0/HDP-GPL-3.1.0.0-centos7-gpl.tar.gz>
<http://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/HDP-UTILS-1.1.0.22/repos/HDP-UTILS-1.1.0.22-centos7.tar.gz>
<http://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/ambari/2.x/updates/2.7.3.0/ambari-2.7.3.0-centos7.tar.gz>
<http://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/HDP/3.x/updates/3.1.0.0/HDP-3.1.0.0-centos7-rpm.tar.gz>
<https://github.com/AdoptOpenJDK/openjdk8-binaries/releases/download/jdk8u222-b10/OpenJDK8U-jdk_x64_linux_hotspot_8u222b10.tar.gz>

# 1.软件使用的技术栈(编程语言列表、运行时、打包方法<可选>)

## 编程语言列表

C、C++、Java、HTML、Javascript、Shell

## 运行时

Oracle JDK1.8

## 编译环境

On Linux, you need the tools to create the native libraries: LZO headers,zlib headers, gcc, OpenSSL headers, cmake, protobuf dev tools, and libtool, and the GNU autotools (automake, autoconf, etc).  
For RHEL (and hence also CentOS):
 yum -y install  lzo-devel  zlib-devel  gcc gcc-c++ autoconf automake libtool openssl-devel fuse-devel cmake
For Debian and Ubuntu:
 apt-get -y install maven build-essential autoconf automake libtool cmake zlib1g-dev pkg-config libssl-dev libfuse-dev
Windows：
CygWin、jdk1.6+，Maven3.0+、Cmake 2.6+、Windows SDK or Visual Studio 2010 Professional、Cygwin、zlib headers
更全面的编译环境指导：
<https://svn.apache.org/viewvc/hadoop/common/branches/branch-2/BUILDING.txt?view=markup>

## 打包方法

使用Maven打包，但是需要非常高配置的硬件，但是官方并没有给出具体的最低配置

# 2.能否运行于 ARM64v8 架构下

Arm64v8下的编译参考：<https://blog.csdn.net/github_39577257/article/details/106211612>
华为已经释出相关编译后的景象和repo：
<https://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/>

# 3.是否已有中文语言包，没有的话能否汉化?

可以进行汉化，但是HDP组件无需汉化，因为其主要为后台进程，

# 4.如何进行二次开发，需要使用哪些技术，需要哪些软硬件环境，能否打包成安装 包<可选>?

不建议进行二次开发，如果非要二次开发，需要熟悉C、C++、Java、HTML、Javascript、Shell、深入研究java虚拟机和Linux操作系统

# 5.确定统一管理⻚面的需求边界

此处不涉及

# 6.调研过程中需要记录过程文档

文档在此处

# 7.列出面临的挑战与各种资源需求

## 面临的挑战

对Arm64V8的交叉编译环境不熟悉，对Linux生态尤其是国产OS的现状不了解

## 资源需求

更高配置的编辑服务器（x86_64）和测试开发环境服务器（Arm64V8）

# 8.著作权问题及法律⻛险

Apache License v2.0，可以商用，但是需要Notice文件标明引用了HDP

# 9.由于硬件暂时不具备，确定工作重心优先定制化;按照任务内容细化调研任务拆 分(大数据、k8s、数据库等)

# 10.统一用户管理，了解 apache kerberos(可选)，kuboard 的组、⻆色和用户

# 11.分任务确定各组件移植的工作量和移植后环境的稳定性(含压测《侧重⻓时间而 非数据量》)

# 12.部署步骤

## 下载离线部署包

<https://mirrors.huaweicloud.com/kunpeng/yum/el/7/bigdata/>
## 安全前置准备，包括安装操作系统、关闭防火墙、同步服务器时钟等；

### 机器名

vi /etc/hostname
hostnamectl set-hostname xxx
vi /etc/hosts

### 关闭防火墙

systemctl stop firewalld
systemctl disable firewalld

### 同步服务器时钟

### selinux

在所有节点执行setenforce 0 命令，此处使用批处理shell执行：
setenforce 0
集群所有节点修改 /etc/selinux/config 文件的如下内容：
SELINUX=disabled

### 禁用虚拟内存和内存大页面

vim /etc/rc.d/rc.local

if test -f /sys/kernel/mm/transparent_hugepage/enabled;
 then
  echo never > /sys/kernel/mm/transparent_hugepage/enabled
fi
if test -f /sys/kernel/mm/transparent_hugepage/defrag;
then
 echo never > /sys/kernel/mm/transparent_hugepage/defrag
fi

禁用虚拟内存和内存大页面
echo vm.swappiness = 10 >> /etc/sysctl.conf
echo never > /sys/kernel/mm/transparent_hugepage/enabled
echo never > /sys/kernel/mm/transparent_hugepage/defrag

### ntp

All server:
yum -y remove chrony
yum -y install ntp

master(如果有自己的ntp上游就填写，否则就用本机器):
vim /etc/ntp.conf

server  127.127.1.0     # local clock
fudge   127.127.1.0 stratum 10

其他节点：
vim /etc/ntp.conf

server 172.31.13.10

开机自启动和启动服务
systemctl enable ntpd
systemctl start ntpd


### repo

master：
 yum -y install httpd
 systemctl enable httpd
 systemctl start httpd

将下载的tar.gz拷贝到 /var/www/html 目录，然后解压缩各文件，删除原文件　tar.gz。

cd /var/www/html
createrepo .

所有节点：

vim /etc/yum.repos.d/ambari.repo

[ambarirepo]
name = ambari_repo
baseurl = http://master/ambari/centos7/2.7.3.0-139
enable = true
gpgcheck = false

vi /etc/yum.repos.d/hdp.repo
[hdprepo]
name = hdp_repo
baseurl = http://master/HDP/centos7/3.1.0.0-78
enable = true
gpgcheck = false

vi /etc/yum.repos.d/hdp_gpl.repo
[hdpgplrepo]
name = hdp_gpl_repo
baseurl = http://master/HDP-GPL/centos7/3.1.0.0-78
enable = true
gpgcheck = false

vi /etc/yum.repos.d/hdp_utls_repo.repo
[hdputilsrepo]
name = hdp_utls_repo
baseurl = http://master/HDP-UTILS/centos7/1.1.0.22
enable = true
gpgcheck = false

## 外部数据库如MySQL安装

yum -y install mariadb mariadb-server mysql-connector-java
 systemctl enable mariadb
 systemctl start mariadb
 mysql_secure_installation
 
create database metastore default character set utf8;
 CREATE USER 'hive'@'%' IDENTIFIED BY '1qaz2wsx';
  GRANT ALL PRIVILEGES ON metastore. * TO 'hive'@'%';
  FLUSH PRIVILEGES;
  create database ambari default character set utf8;
 
  CREATE USER 'ambari'@'%' IDENTIFIED BY '1qaz2wsx';
   CREATE USER 'ambari'@'localhost' IDENTIFIED BY '1qaz2wsx';
   CREATE USER 'ambari'@'master.test.fnii' IDENTIFIED BY '1qaz2wsx';
   CREATE USER 'ambari'@'master' IDENTIFIED BY '1qaz2wsx';

    GRANT ALL PRIVILEGES ON ambari.* TO 'ambari'@'localhost';
 GRANT ALL PRIVILEGES ON ambari.*TO 'ambari'@'master.test.fnii';
 GRANT ALL PRIVILEGES ON ambari.* TO 'ambari'@'master';
  GRANT ALL PRIVILEGES ON ambari.* TO 'ambari'@'%';
  FLUSH PRIVILEGES;

## 安装Ambari

### 安装ambari

yum -y install ambari-server

### 建立Ambari与MySQL驱动的连接

ambari-server setup --jdbc-db=mysql --jdbc-driver=/usr/share/java/mysql-connector-java.jar

ambari-server setup
Customize user account for ambari-server daemon [y/n] (n)? n

自定义jdk，输入java——home：
/usr/java/jdk1.8.0_361-aarch64
Enter advanced database configuration [y/n] (n)? y
Enter choice (1): 3
输入数据库相关信息
Should ambari use existing default jdbc /usr/share/java/mysql-connector-java.jar [y/n] (y)? y

### 完毕后手动执行sql创建ambari相关的表

mysql -uambari -p -Dambari < /var/lib/ambari-server/resources/Ambari-DDL-MySQL-CREATE.sql

### 启动

ambari-server start
netstat -lnpt | grep 8080
## 安装HDP集群；

### centos7免密登录

ssh-keygen -t rsa
 ssh-copy-id -i .ssh/id_rsa.pub -p22 root@localhost

### 访问

<http://masterIP:8080>

如果出现“Cannot find a valid baseurl for repo: HDP-3.1-repo-1” 的错误，按照下面的方式解决：

Steps

1) go to /usr/lib/ambari-server/web/javascipts

cd /usr/lib/ambari-server/web/javascripts

2) take backup of app.js

 cp app.js app.js_backup

3) edit the app.js

find out the line(39892) : onNetworkIssuesExist: function () {

Change the line from :

  /**

* Use Local Repo if some network issues exist
   */
  onNetworkIssuesExist: function () {
    if (this.get('networkIssuesExist')) {
      this.get('content.stacks').forEach(function (stack) {
          stack.setProperties({
            usePublicRepo: false,
            useLocalRepo: true
          });
          stack.cleanReposBaseUrls();
      });
    }
  }.observes('networkIssuesExist'),

to

  /**

* Use Local Repo if some network issues exist
   */
  onNetworkIssuesExist: function () {
    if (this.get('networkIssuesExist')) {
      this.get('content.stacks').forEach(function (stack) {
        if(stack.get('useLocalRepo') != true){
          stack.setProperties({
            usePublicRepo: false,
            useLocalRepo: true
          });
          stack.cleanReposBaseUrls();
        }
      });
    }
  }.observes('networkIssuesExist'),

as per : <https://github.com/apache/ambari/pull/2743/files>

Later as you have already deployed the cluster we need to reset the cluster (Caution : this will erase all the configs you have created previously in Step6 and also the Hosts and services you have selected need to select again )

Command :

ambari-server reset

And hard reload the page and start the create cluster wizard again.

Incase you have already at Step 9 and cannot proceed with ambari-server reset (as it invovles lots of Configs being added again , the below steps are for you )

Preqrequesties : The cluster now is in Deployment step(step 9 ) and you have only retry button to press

steps

1) Stop ambari-server

2) login to Database

3) use the below command to list out all the contents in repo_definition table :

 select * from repo_definition;

4) you can see the base_url will be empty for the all the Rows in the table

5) Correct the base_url for every rows and update it using the command :

update repo_definition set base_url='<YOUR BASE URL>' where id=<THE CORESPONDING ID>;

for ex :

update repo_definition set base_url='http://asnaik.example.com/localrepo/HDP-3.1' where id=9;

update repo_definition set base_url='http://master/HDP/centos7/3.1.0.0-78' where id=25;
update repo_definition set base_url='http://master/HDP-UTILS/centos7/1.1.0.22' where id=26;

6) after correcting all the base_url columns in repo_definition table and also delete the empty repos created by ambari from location /etc/yum.repos.d

7) start ambari, Login to UI and press retry button, The Installation will work as smooth as it can be.

Hope this helps.

if HBase was start with failue, please check hbase-env and remove jvm para:
UseG1GC
then restart hbase process.
enjoy~
如果出现无法安装HBase的问题，请查看日志，如果是无法创建java虚拟机，请删除两处java虚拟机的参数即可，一个是env，一个在Ambari页面中。