失眠网 > ansible一键部署高可用集群项目实战最细教程

ansible一键部署高可用集群项目实战最细教程

时间：2021-01-21 16:49:27

文章目录

服务架构图环境配置IP规划和配置ssh免密登录开始搭建管理节点准备工作搭建数据库搭建NAS存储节点搭建备份节点搭建web节点搭建负载均衡节点配置keepalived高可用配置Prometheus监控准备二进制软件包配置Prometheus-server节点配置node-exporter探针配置alertmanager组件总剧本搭建博客总结一下

服务架构图

环境配置

IP规划和配置

负载均衡节点

nginx1：192.168.146.100

先查看本机网卡：

nmcli connection show

可以看到我的虚拟机上网卡连接名是ens33 根据连接名修改网卡IP

nmcli connection modify ens33 ipv4.address 192.168.146.100/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

修改网卡后记得重启该网卡，让配置生效

nmcli connection down ens33nmcli connection up ens33

nginx2：192.168.146.101

修改网卡IP，按理说虚拟机环境下一张网卡的网卡连接名都是一样的，如果不放心可以重复上述步骤查看，再修改网卡

nmcli connection modify ens33 ipv4.address 192.168.146.101/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

重启网卡，操作跟上面的相同，下面就不再显示了

web服务器

apache1：192.168.146.102

nmcli connection modify ens33 ipv4.address 192.168.146.102/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

apache2：192.168.146.103

nmcli connection modify ens33 ipv4.address 192.168.146.103/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

NAS存储节点：

NFS服务器：192.168.146.104

nmcli connection modify ens33 ipv4.address 192.168.146.104/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

MySQL主从集群

master：192.168.146.105

nmcli connection modify ens33 ipv4.address 192.168.146.105/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

slave：192.168.146.106

nmcli connection modify ens33 ipv4.address 192.168.146.106/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

Prometheus-server节点

prometheus：192.168.146.107

nmcli connection modify ens33 ipv4.address 192.168.146.107/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

rsync-server节点

rsyncd：192.168.146.108

nmcli connection modify ens33 ipv4.address 192.168.146.108/24 ipv4.gateway 192.168.146.2 ipv4.dns 114.114.114.114 ipv4.method manual

ssh免密登录

用一台管理节点，对上面配置的所有服务器节点配置成可以ssh免密登录，为了后续在管理节点上ansible一键搭建整个架构

在管理节点上，生成密钥对

ssh-keygen

出现提示信息一路敲击回车就好

检查一下是否成功生成RSA密钥对

将公钥id_rsa.pub发送至各个服务器，以下命令在每个服务器包括管理节点都执行一遍，IP换成对应的即可

ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.134 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.100 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.101 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.102 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.103 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.104 ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.105ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.106ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.107ssh-copy-id -i /root/.ssh/id_rsa.pub 192.168.146.108

出现以下信息即完成传输

可以在其他服务器上检查一下是否存在authorized_keys文件，存在该文件就说明公钥从管理节点接收成功了

在管理节点上，可以验证一下是否可以成功ssh免密登录到其他节点

[root@server1 ~]# ssh root@192.168.146.101Last login: Sat Apr 9 10:39:32 from 192.168.146.1 # 查看当前IP，发现已经成功登录到apache2节点上了[root@server1 ~]# ip a | grep -v 'LOOPBACK' |awk '/^[0-9]/{print $2;getline;getline;if ($0~/inet/){print $2}}'ens33:192.168.146.101/24

开始搭建

管理节点

准备工作

以下所有命令均在管理节点上使用ansible工具配置即可

使用yum安装ansible工具

# 安装epel扩展源yum install epel-release.noarch -y# 安装ansibleyum install -y ansible

编写ansible主机清单/etc/ansible/hosts文件，在文件末尾添加如下内容：

[all_ip]192.168.146.134 hostname=manager192.168.146.100 hostname=nginx1192.168.146.101 hostname=nginx2192.168.146.102 hostname=apache1192.168.146.103 hostname=apache2192.168.146.104 hostname=nas rsync_server=192.168.146.108192.168.146.105 hostname=master192.168.146.106 hostname=slave192.168.146.107 hostname=prometheus192.168.146.108 hostname=rsyncd[balancers]nginx1 mb=MASTER priority=100nginx2 mb=BACKUP priority=98[web]apache1apache2[mysql]master master=trueslave slave=true[mysql:vars]master_ip=192.168.146.105slave_ip=192.168.146.106[nfs]nas[nfs:vars]rsync_server=192.168.146.108[rsync]rsyncd[prometheus]prometheus[alertmanagers]prometheus[node-exporter]192.168.146.100192.168.146.101192.168.146.102192.168.146.103192.168.146.104192.168.146.105192.168.146.106192.168.146.108

检查连通性，所有服务器均响应pong才表示管理节点与所有服务器连通无误

为所有服务器设置主机名，顺便把selinux和防火墙关了，这里为了偷懒选择直接关掉防火墙和selinux（当然也可以为每类的角色做特定的放行策略和设置对应的安全上下文，更符合生产环境配置）

[root@server1 ansible]# vim prepare_work.yml - name: prepare workhosts: all_iptasks:- name: set hostnameshell: hostnamectl set-hostname {{hostname }}- name: stop firewalldservice:name: firewalldstate: stoppedenabled: no- name: disabled selinuxselinux:state: disabled# 执行剧本！[root@server1 ansible]# ansible-playbook prepare_work.yml

为所有服务器主机生成/etc/hosts/解析文件

[root@server1 ~]# mkdir ansible[root@server1 ~]# cd ansible/# 编写hosts的模板文件[root@server1 ansible]# vim hosts.j2127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4::1 localhost localhost.localdomain localhost6 localhost6.localdomain6{% for host in groups.all_ip %}{{hostvars[host].ansible_ens33.ipv4.address }} {{hostvars[host].ansible_hostname }}{% endfor %}# 编写剧本yml文件[root@server1 ansible]# vim hosts.yml- name: update hosts filehosts: all_iptasks:- name: copy hosts.j2 to other serverstemplate:src: hosts.j2dest: /etc/hosts# 执行剧本yml文件[root@server1 ansible]# ansible-playbook hosts.yml

创建存放角色的目录roles

[root@server1 ansible]# mkdir roles

此时我们的角色目录roles不是ansible默认的角色目录，我们需要到/etc/ansible/ansible.cfg文件中修改，否则ansible找不到我们创建的角色

至此环境的准备工作结束了，让我们开始对具体的服务进行配置吧！

搭建数据库

创建数据库msql角色

[root@server1 ansible]# ansible-galaxy init role/mysql- Role role/mysql was created successfully

因为数据库我们选择搭建一个主从架构，设计的参数较多，并且有些是重复的，有些是运行ansible的py脚本的节点判断是主节点还是从节点的依据，所以我们可以预先定义一些变量在了主机清单文件中，后续的keepalived高可用搭建在主机清单文件中定义了一些变量，为了主备节点的判断。那么还有一些通用的变量，为了偷偷懒少写重复的值，可以在角色的vars/main.yml下定义：

[root@server1 ansible]# vim roles/mysql/vars/main.yml---# vars file for role/mysqlmysql_sock: /var/lib/mysql/mysql.sockmysql_port: 3306repl_user: replrepl_passwd: "123456"

编写模板文件，用于修改mysql的配置文件f，此时预先定义的变量就派上用场了，主从节点的配置server-id的值需要不一样的。

[mysqld]datadir=/var/lib/mysqlsocket=/var/lib/mysql/mysql.socksymbolic-links=0{% if master is defined %}server-id=1innodb_file_per_table=on{% else %}server-id=2{% endif %}log-bin=master-binbinlog-format=ROW# 这些是开启gtid的参数log-slave-updates=truegtid-mode=onenforce-gtid-consistency=truemaster-info-repository=TABLErelay-log-info-repository=TABLEsync-master-info=1binlog-rows-query-log_events=1[mysqld_safe]log-error=/var/log/mysqld.logpid-file=/var/run/mysqld/mysqld.pid

编写mysql/tasks/main.yml文件，完成一键搭建主从架构的ansible脚本。

注：

一键搭建MySQL主从复制架构的ansible脚本这里采用了基于GTID的主从复制模式，比较方便，不需要我们去找主节点bin-log的position

并且我这里使用的是yum安装mysql而不是MariaDB，因此需要在安装之前先配置下载mysql源，否则因为yum源列表里没有mysql会默认安装MariaDB替换

[root@server1 ansible]# vim roles/mysql/tasks/main.yml ---# tasks file for role/mysql- name: yum install wgetyum:name: wgetstate: present- name: wget mysql reposhell: wget /mysql-community-release-el7-5.noarch.rpm- name: rpm mysql reposhell: rpm -ivh mysql-community-release-el7-5.noarch.rpmignore_errors: True- name: yum install mysqlyum:name: mysql-serverstate: present- name: config ftemplate:src: f.j2dest: /etc/f- name: start mysqlservice:name: mysqldstate: restartedenabled: yes- name: install MYSQL-pythonyum:name: MySQL-pythonstate: present- name: update mysql root passwordshell: mysql -e "update mysql.user set password=password('123456') where user='root' and host='localhost';flush privileges;"ignore_errors: True- name: create repl usermysql_user:name: "{{ repl_user }}"host: '192.168.146.%'password: "{{ repl_passwd }}"priv: "*.*:REPLICATION SLAVE"state: presentlogin_user: 'root'login_password: '123456'login_host: localhostlogin_unix_socket: "{{ mysql_sock }}"login_port: "{{ mysql_port }}"when: master is defined- name: change master tomysql_replication:mode: changemastermaster_host: "{{ master_ip }}"master_user: "{{ repl_user }}"master_password: "{{ repl_passwd }}"login_password: '123456'login_host: localhostlogin_unix_socket: "{{ mysql_sock }}"login_port: "{{ mysql_port }}"when: slave is defined- name: start_slavemysql_replication:mode: startslavelogin_user: 'root'login_password: '123456'login_host: localhostlogin_unix_socket: "{{ mysql_sock }}"login_port: "{{ mysql_port }}"when: slave is defined- name: get_slave_infomysql_replication:login_host: localhostlogin_user: rootlogin_port: "{{ mysql_port }}"login_password: '123456'login_unix_socket: "{{ mysql_sock }}"mode: getslavewhen: slave is definedregister: info- name: dispaly_slavedebug:msg: "Slave_IO_Running={{ info.Slave_IO_Running }} Slave_SQL_Running={{ info.Slave_SQL_Running }}"when: slave is defined

搭建NAS存储节点

本架构通过NFS服务共享本地的资源给web节点使用，先创建nfs角色

[root@server1 ansible]# ansible-galaxy init role/nfs- Role role/nfs was created successfully

创建好角色后，编写nfs/tasks/main.yml文件

[root@server1 ansible]# vim role/nfs/tasks/main.yml ---# tasks file for role/nfs- name: install nfs,expectyum:name: "{{ item }}"state: presentloop:- nfs-utils- rpcbind- expect*- name: config dir for dynamicshell: mkdir /data | chmod -Rf 777 /data | echo "/data 192.168.146.0/24(rw,sync,no_root_squash)" >> /etc/exports- name: make dir for exp and shfile:path: /shstate: directory- name: make dir for backupfile:path: /backupstate: directory- name: config expect.shtemplate:src: rsync.exp.j2dest: /sh/rsync.exp- name: config beifen.shtemplate:src: beifen.sh.j2dest: /sh/beifen.sh- name: chmod beifen.shfile:path: /sh/beifen.shmode: '0755'- name: cron tabshell: echo "0 1 * * * root /sh/beifen.sh" >> /etc/crontab - name: start nfsservice:name: "{{ item }}"state: restartedenabled: yesloop:- rpcbind- nfs-server

准备expect脚本的模板文件rsync.exp.j2

[root@manager ansible]# vim roles/nfs/templates/rsync.exp.j2#!/usr/bin/expectset mulu [lindex $argv 0]set timeout 10spawn rsync -avzr /backup/$mulu root@{{rsync_server }}::backup_serverexpect Passwordsend "123456\r"expect eof

准备备份脚本的模板文件beifen.sh.j2

[root@manager ansible]# vim roles/nfs/templates/beifen.sh.j2#!/bin/bash# 准备压缩文件的目录mulu=`ip a | grep global|awk -F'[ /]+' '{print $3}'`_`date +%F`echo $mulumkdir -pv /backup/$mulu &> /dev/null# 打包待发送的数据tar zcf /backup/$mulu/conf.tar.gz /data/* &> /dev/nulltouch /backup/$mulu# 发送数据# 这一句就是执行expect脚本expect /sh/rsync.exp $mulu# 保留七天以内的数据find /backup -mtime +7 -delete

搭建备份节点

备份选择rsync服务端，定时接收NAS存储节点推送的备份文件。因此先创建rsync角色

[root@manager ansible]# ansible-galaxy init roles/rsync- Role roles/rsync was created successfully

编写文件rsyncd.conf

[root@manager ~]# vim ansible/roles/rsync/files/rsyncd.conf [backup_server]path = /backupuid = rootgid = rootmax connections = 2timeout = 300read only = falseauth users = rootsecrets file = /etc/rsync.passwdstrict modes = yesuse chroot = yes

编写rsync/tasks/main.yml文件

[root@manager ansible]# vim roles/rsync/tasks/main.yml ---# tasks file for roles/rsync- name: yum install rsyncyum:name: rsyncstate: present- name: config rsyncd.confcopy:src: rsyncd.confdest: /etc/rsyncd.conf- name: make dir for backupfile:path: /backupstate: directory- name: prepare rsync.passwdshell: echo "root:123456" >> /etc/rsync.passwd | chmod 600 /etc/rsync.passwd- name: start rsyncservice:name: rsyncdstate: startedenabled: yes

搭建web节点

因为本架构是LAMP环境，web服务器选用apache，因此先创建apache角色

[root@server1 ansible]# ansible-galaxy init role/apache- Role role/apache was created successfully[root@server1 ansible]# ll role/总用量 0drwxr-xr-x. 10 root root 154 4月 9 14:30 apachedrwxr-xr-x. 10 root root 154 4月 9 14:16 nginx

编写apache/tasks/main.yml文件

[root@server1 ansible]# vim role/apache/tasks/main.yml ---# tasks file for role/apache- name: yum install lamp environmentyum:name: httpd,php-fpm,php-mysql,mod_phpstate: present- name: start httpdservice:name: httpdstate: restartedenabled: yes- name: start php-fpmservice:name: php-fpmstate: restartedenabled: yes- name: Client Install nfs Serveryum:name: nfs-utilsstate: present- name: mount nfs resourcesmount:src: nas:/datapath: /var/www/htmlfstype: nfsopts: defaultsstate: mounted

搭建负载均衡节点

因为本架构的负载均衡实现选择的是nginx，那就先创建nginx角色

# 创建nginx角色[root@server1 ansible]# ansible-galaxy init role/nginx- Role role/nginx was created successfully

编写模板文件，为前端的两台nginx负载均衡节点创建负载均衡的配置文件lb.conf，为了实现会话保持，在这顺便使用最简单的方法：使用基于ip_hash算法的负载均衡

[root@server1 ansible]# vim role/nginx/templates/lb.conf.j2upstream webservers{server apache1;server apache2;ip_hash;}server{location / {proxy_pass http://webservers;proxy_next_upstream error timeout invalid_header http_500 http_502 http_504;proxy_set_header Host $host;proxy_set_header X-Real-IP $remote_addr;proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;}}

创建好nginx角色和模板文件之后，那就来编写nginx/tasks/main.yml文件吧

[root@server1 ansible]# vim role/nginx/tasks/main.yml ---# tasks file for nginx- name: yum install epelyum:name: epel-release.noarchstate: present- name: yum install nginxyum:name: nginxstate: present- name: config lb.conftemplate:src: lb.conf.j2dest: /etc/nginx/conf.d/lb.conf- name: start nginxservice:name: nginxstate: restartedenabled: yes

配置keepalived高可用

本架构前端负载均衡节点需要做高可用，因此创建keepalived角色，VIP：192.168.146.200

[root@server1 ansible]# ansible-galaxy init role/keepalived- Role role/keepalived was created successfully

接着编写模板文件，用于在负载均衡节点上创建keepalived的配置文件

[root@server1 ansible]# vim role/keepalived/templates/keepalived.conf.j2!Configuration File for keepalivedglobal_defs {router_id {{ansible_hostname }}}vrrp_script check_nginx {script "/usr/local/src/check_nginx_pid.sh"interval 1weight -10}vrrp_instance VI_1 {state {{mb }}interface ens33virtual_router_id 10priority {{priority }}advert_int 1authentication {auth_type PASSauth_pass 1111}track_script {check_nginx}virtual_ipaddress {192.168.146.200}}

编写用于检测nginx服务状态的脚本，以便于keepalived实现VIP飘移

[root@manager ansible]# vim roles/keepalived/files/check_nginx_pid.sh#!/bin/bashnginx_process_num=`ps -C nginx --no-header | wc -l`if [ $nginx_process_num -eq 0 ];thenexit 1elseexit 0fi

创建好keepalived角色和模板文件后，开始编写keepalived/tasks/main.yml文件

[root@server1 ansible]# vim role/keepalived/tasks/main.yml ---# tasks file for role/keepalived- name: yum install keepalivedyum:name: keepalivedstate: present- name: copy check_nginx_pid.shcopy:src: check_nginx_pid.shdest: /usr/local/src/check_nginx_pid.sh- name: chmod shfile:path: /usr/local/src/check_nginx_pid.shmode: '0755'- name: config keepalived.conftemplate:src: keepalived.conf.j2dest: /etc/keepalived/keepalived.conf- name: start keepalivedservice:name: keepalivedstate: restartedenabled: yes

配置Prometheus监控

这里的为服务集群配置Prometheus监控服务，需要安装Prometheus-server负责拉取数据并展示，再在各个节点上安装node-exporter用于监控服务器的CPU、内存、机器开启状态，还需要安装alertmanager用于告警。

因此下面创建了三个角色，负责上述的三项工作，来完成Prometheus监控服务的配置

准备二进制软件包

这里的Prometheus监控服务的组件都使用二进制包来安装，但是从官网上下载二进制包非常慢，并且用tar解压时会有归档异常，如果用ansible工具的脚本直接编写语句，在受控节点下载二进制包并且解压，首先会非常慢，其次解压会因为报错而不执行（本人亲测，若大佬们有更好的方法请指教！！！）

这里作为小白的我没有找到更好的解决办法，因此我只能选择将本地事先准备好的Prometheus组件的压缩包上传到ansible管理节点，并且将每个组件的压缩包先解压到创建的角色如：roles/prometheus/files目录下，然后在编写角色的任务时，用copy模块直接到受控节点上

# 解压压缩包[root@manager ansible]# tar -zxvf prometheus-2.25.0.linux-amd64.tar.gz -C roles/prometheus/files/prometheus-2.25.0.linux-amd64/prometheus-2.25.0.linux-amd64/consoles/prometheus-2.25.0.linux-amd64/consoles/index.html.exampleprometheus-2.25.0.linux-amd64/consoles/node-cpu.htmlprometheus-2.25.0.linux-amd64/consoles/node-disk.htmlprometheus-2.25.0.linux-amd64/consoles/node-overview.htmlprometheus-2.25.0.linux-amd64/consoles/node.htmlprometheus-2.25.0.linux-amd64/consoles/prometheus-overview.htmlprometheus-2.25.0.linux-amd64/consoles/prometheus.htmlprometheus-2.25.0.linux-amd64/console_libraries/prometheus-2.25.0.linux-amd64/console_libraries/menu.libprometheus-2.25.0.linux-amd64/console_libraries/prom.libprometheus-2.25.0.linux-amd64/prometheus.ymlprometheus-2.25.0.linux-amd64/LICENSEprometheus-2.25.0.linux-amd64/NOTICEprometheus-2.25.0.linux-amd64/prometheusprometheus-2.25.0.linux-amd64/promtool# 报错信息不用管！！！gzip: stdin: unexpected end of filetar: 归档文件中异常的 EOFtar: 归档文件中异常的 EOFtar: Error is not recoverable: exiting now# 解压好了！[root@manager ansible]# ll roles/prometheus/files/总用量 55608-rw-r--r--. 1 root root1606 4月 11 20:20 node.ymldrwxr-xr-x 4 3434 3434132 2月 18 prometheus-2.25.0.linux-amd64

安装其他组件同理

tar -zxvf alertmanager-0.21.0.linux-amd64.tar.gz -C roles/alertmanager/files/tar -zxvf node_exporter-1.3.1.linux-amd64.tar.gz -C roles/node-exporter/files/

配置Prometheus-server节点

老样子，第一步创建角色

ansible-galaxy init roles/prometheus

为Prometheus服务配置文件编写模板文件prometheus.yml.j2

[root@manager ansible]# vim roles/prometheus/templates/prometheus.yml.j2 global:scrape_interval:30sevaluation_interval: 30squery_log_file: ./promql.logalerting:alertmanagers:- static_configs:- targets:{% for alertmanager in groups['alertmanagers'] %}- {{alertmanager }}:9093{% endfor %}rule_files:- "rules/*.yml"scrape_configs:- job_name: 'prometheus'static_configs:- targets:{% for prometheu in groups['prometheus'] %}- "{{ prometheu }}:9090"{% endfor %}- job_name: "node"static_configs:- targets:{% for node in groups['node-exporter'] %}- "{{ node }}:9100"{% endfor %}

为Prometheus配置成service项编写模板文件prometheus.service.j2

[root@manager ansible]# vim roles/prometheus/templates/prometheus.service.j2 [Unit]Description=PrometheusAfter=network.target[Service]WorkingDirectory=/usr/local/prometheusExecStart=/usr/local/prometheus/prometheusExecReload=/bin/kill -HUP $MAINPIDExecStop=/bin/kill -KILL $MAINPIDType=simpleKillMode=control-groupRestart=on-failureRestartSec=3s[Install]WantedBy=multi-user.target

编写告警规则文件node.yml

因为node.yml文件中需要用到jinjia2语法写监控项，所以就不用template模块来在受控节点上编写该文件，而是直接使用copy模块将编写好的文件复制过去即可

[root@manager ansible]# vim roles/prometheus/files/node.yml groups:- name: node.rules # 报警规则组名称expr: up == 0for: 30s #持续时间,表示持续30秒获取不到信息，则触发报警annotations:summary: "Instance {{ $labels.instance }} down" # 自定义摘要for: 2mlabels:severity: warningannotations:summary: "{{$labels.instance}}: {{$labels.mountpoint }} 分区使>用过高"description: "{{$labels.instance}}: {{$labels.mountpoint }} 分>区使用大于 80% (当前值: {{ $value }})"- alert: node Memoryexpr: 100 - (node_memory_MemFree_bytes+node_memory_Cached_bytes+node_memory_Buffers_bytes) / node_memory_MemTotal_bytes * 100 > 80for: 2mlabels:severity: warningannotations:summary: "{{$labels.instance}}: 内存使用过高"description: "{{$labels.instance}}: 内存使用大于 80% (当前值: {{ $value }})"- alert: node CPUexpr: 100 - (avg(irate(node_cpu_seconds_total{mode="idle"}[5m])) by (instance) * 100) > 80for: 2mlabels:severity: warningannotations:summary: "{{$labels.instance}}: CPU使用过高"description: "{{$labels.instance}}: CPU使用大于 80% (当前值: {{ $value }})"

编写prometheus/tasks/main.yml

[root@manager ansible]# vim roles/prometheus/tasks/main.yml ---# tasks file for roles/promethues- name: copy prometheus.tar.gzcopy:src: prometheus-2.25.0.linux-amd64dest: /usr/local/- name: create soft linkfile:src: /usr/local/prometheus-2.25.0.linux-amd64dest: /usr/local/prometheusstate: link- name: chmod filefile:path: /usr/local/prometheus/prometheusmode: '0755'- name: copy service filetemplate:src: prometheus.service.j2dest: /etc/systemd/system/prometheus.service- name: copy config prometheus yamltemplate:src: prometheus.yml.j2dest: /usr/local/prometheus/prometheus.yml- name: create rules dirfile:path: /usr/local/prometheus/rulesstate: directory- name: copy rules yamlcopy:src: node.ymldest: /usr/local/prometheus/rules/node.yml- name: start prometheusservice:name: prometheusstate: startedenabled: yes

配置node-exporter探针

创建角色

ansible-galaxy init roles/node-exporter

编写模板文件node_exporter.service.j2

[root@manager ansible]# vim roles/node-exporter/templates/node_exporter.service.j2 [Unit]Description=Node ExporterAfter=network.target[Service]WorkingDirectory=/prometheus_exporter/node_exporter/ExecStart=/prometheus_exporter/node_exporter/node_exporterExecStop=/bin/kill -KILL $MAINPIDType=simpleKillMode=control-groupRestart=on-failureRestartSec=3s[Install]WantedBy=multi-user.target

编写node-exporter/tasks/main.yml

[root@manager ansible]# vim roles/node-exporter/tasks/main.yml ---# tasks file for roles/node-exporter- name: create dirfile:path: /prometheus_exporterstate: directory- name: copy filecopy:src: node_exporter-1.3.1.linux-amd64dest: /prometheus_exporter- name: create linkfile:src: /prometheus_exporter/node_exporter-1.3.1.linux-amd64dest: /prometheus_exporter/node_exporterstate: link- name: chmod filefile:path: /prometheus_exporter/node_exporter/node_exportermode: '0755'- name: copy service filetemplate:src: node_exporter.service.j2dest: /etc/systemd/system/node_exporter.service- name: start node_exporterservice:name: node_exporterstate: restartedenabled: yes

配置alertmanager组件

创建角色

ansible-galaxy init roles/alertmanager

编写模板文件alertmanager.service.j2

[root@manager ansible]# vim roles/alertmanager/templates/alertmanager.service.j2 [Unit]Description=AlertManagerAfter=network.target[Service]WorkingDirectory=/usr/local/alertmanager/ExecStart=/usr/local/alertmanager/alertmanagerExecReload=/bin/kill -HUP $MAINPIDExecStop=/bin/kill -KILL $MAINPIDType=simpleKillMode=control-groupRestart=on-failureRestartSec=3s[Install]WantedBy=multi-user.target

编写模板文件alertmanager.yml.j2

smtp_auth_password字段的密码不是QQ邮箱的密码，到QQ邮箱的设置 ——> 账户里，找到下图的生成授权码，填写在该字段上

[root@manager ansible]# vim roles/alertmanager/templates/alertmanager.yml.j2 global:resolve_timeout: 5msmtp_from: "2249807270@"smtp_smarthost: ':465'smtp_auth_username: "2249807270@"smtp_auth_password: "vwsorrnckxwjdhgf"smtp_require_tls: falseroute:group_by: ['alertname']group_wait: 10sgroup_interval: 10srepeat_interval: 24hreceiver: 'default-receiver'receivers:- name: 'default-receiver'email_configs:- to: '2249807270@'inhibit_rules:- source_match:severity: 'critical'target_match:severity: 'warning'equal: ['alertname', 'dev', 'instance']

编写alertmanager/tasks/main.yml

[root@manager ansible]# vim roles/alertmanager/tasks/main.yml ---# tasks file for roles/alertmanager- name: copy filecopy:src: alertmanager-0.21.0.linux-amd64dest: /usr/local/- name: create linkfile:src: /usr/local/alertmanager-0.21.0.linux-amd64dest: /usr/local/alertmanagerstate: link- name: chmod filefile:path: /usr/local/alertmanager/alertmanagermode: '0755'- name: copy service filetemplate:src: alertmanager.service.j2dest: /etc/systemd/system/alertmanager.service- name: copy config yamltemplate:src: alertmanager.yml.j2dest: /usr/local/alertmanager/alertmanager.yml- name: start serverservice:name: alertmanagerstate: restartedenabled: yes

总剧本

到此，整个架构的所有服务，都已经用角色编辑好了！现在我们最后一步就剩编写一个总剧本，也是最简单的，把所有的编写好的角色都调用一遍就好了

[root@server1 ansible]# vim all.yml- name: config mysql replicationhosts: mysqlroles:- mysql- name: config nfshosts: nfsroles:- nfs- name: config rsynchosts: rsyncroles:- rsync- name: config lamphosts: webroles:- apache- name: config lbhosts: balancersroles:- nginx- keepalived- name: install prometheushosts: prometheusroles:- prometheus- alertmanager- name: install node-exporterhosts: node-exporterroles:- node-exporter

一键安装！

ansible-playbook all.yml

搭建博客

这里的博客使用typecho，现在到我们搭建好的NFS服务器上，解压typecho压缩包到共享的资源目录/data/下

[root@nas ~]# yum install -y wget[root@nas ~]# wget /downloads/1.1-17.10.30-release.tar.gz[root@nas ~]# ll总用量 11156-rw-r--r--. 1 root root 487445 10月 30 1.1-17.10.30-release.tar.gz[root@nas ~]# tar -zxvf 1.1-17.10.30-release.tar.gz[root@nas ~]# ll总用量 11156-rw-r--r--. 1 root root 487445 10月 30 1.1-17.10.30-release.tar.gzdrwxr-xr-x. 6 501 games111 10月 30 build[root@nas ~]# mv build/ typecho[root@nas ~]# cp -r typecho/ /data/[root@nas ~]# ll /data/总用量 0drwxr-xr-x. 6 root root 111 4月 9 20:42 typecho

到MySQL主库服务器上，创建好用于博客软件登录和博客软件使用的数据库，并授予好权限

[root@master ~]# mysql -uroot -p123456mysql> create database bbs;mysql> grant all on bbs.* to 'bbs'@'192.168.146.%' identified by '123456';mysql> flush privileges;

现在，直接打开浏览器，输入192.168.146.200/typecho，就会跳转到开启typecho博客界面了

让我们点击下一步，输入连接的数据库基本信息

再把管理员信息输入完成，即可开始安装博客，后续跟着提示信息一步步安装即可。