运维实例：批量替换配置文件+批量重启服务

工作中遇到的一个小问题，记录一下。假设有一千台机器运行着nginx但每台机器上的nginx目录可能不太一样，比如
/export/servers/openresty-local/nginx/
/export/servers/openresty/nginx/
/export/servers/openresty-ngx-ump/nginx/
现在要替换掉nginx目录下的某个脚本，并且重启这个nginx服务。需要用到的工具有Ansbile和自己写的批量替换脚本batch_op.py。

#准备远程机器列表
$ vim ip_list
10.191.172.201
10.190.143.38
10.187.110.4
10.190.49.237
10.190.198.192
10.190.163.211

#准备替换脚本
$ vim /tmp/b.sh
#!/bin/bash

if [ $(id -u) != "0" ]; then
    echo "Error: need root to run this script."
    exit 1
fi

a=`ps -ef | grep openresty | grep -v grep | awk '{ print $11 }' | cut -d\/ -f1,2,3,4`

if [ ! $a ]; then
    echo 'Error! Not found $a directory.'
elif [ -f $a/lualib/resty/nginx_diviner.lua ]; then
   cp $a/lualib/resty/nginx_diviner.lua $a/lualib/resty/nginx_diviner.lua.bak.`date +%Y%m%d%H%M%S` && \
   wget -q -O $a/lualib/resty/nginx_diviner.lua http://172.22.193.65:8003/nginx_diviner.lua && \
   chown admin:admin $a/lualib/resty/nginx_diviner.lua* && \
   $a/nginx/sbin/nginx -s reload && \
   echo "Found and replaced $a/lualib/resty/nginx_diviner.lua"
else
   echo "Not Found $a/lualib/resty/nginx_diviner.lua"
   exit 1
fi

if [ "$a" != "/export/servers/openresty" ] && [ -f /export/servers/openresty/lualib/resty/nginx_diviner.lua ]; then
   cp /export/servers/openresty/lualib/resty/nginx_diviner.lua /export/servers/openresty/lualib/resty/nginx_diviner.lua.bak.`date +%Y%m%d%H%M%S` && \
   wget -q -O /export/servers/openresty/lualib/resty/nginx_diviner.lua http://172.22.193.65:8003/nginx_diviner.lua && \
   chown admin:admin /export/servers/openresty/lualib/resty/nginx_diviner.lua && \
   echo "Extra Operation! also found and replaced /export/servers/openresty/lualib/resty/nginx_diviner.lua"
fi

把本地userA用户的key推送到远程root用户下, 实现本地userA用户无密码登陆到远程的root用户

$ vim deploy_key_local_admin_to_remote_root.yml   #准备push ssh-key
- hosts: all
  remote_user: root
  gather_facts: no
 
  tasks:
  - name: install ssh key
    authorized_key: user=root
                    key="{{ lookup('file', '/home/userA/.ssh/id_rsa.pub') }}" 
                    state=present

$ vim push_file.yml    #将本地的操作脚本Push到列表中的机器上
- hosts: all
  user: root
  tasks:
  - name: push ump_agent_code to remote
    copy: 
      src: /tmp/b.sh
      dest: /tmp/
      owner: admin
      group: admin
      mode: 0755
      backup: no
      follow: yes

开始操作

ansible-playbook -i ip_list deploy_ssh_key.yml -k  #推送ssh key, 需要输入密码

ansible-playbook -i ip_list  push_file.yml -u root    #将/tmp/b.sh推送到ip_list所有机器的/tmp目录

./batch_op.py -f ip_list -u root "/tmp/b.sh"    #批量在机器上执行b.sh并返回执行结果