failover 故障检测、报警、自动切换 程序 monitor_nginx.sh 用于对web server (Nginx) 监测与故障处理

 

前言

前几天，群里有人说用shell程序实现failover,所以今天自己写一个shell程序 monitor_nginx.sh 

failover,fault  detection and  switching   ,故障转移与切换程序(shell)

   

一.程序功能

用于检测web server (Nginx)是否宕掉，如果是用短信报警(sendmail139邮箱报警)并自动故障处理，浮动改热备机IP，实现故障转移与自动切换

 

二.系统架构

如图所示

172.29.141.112    上部署  Nginx  与 fastcgi(对外开放)

172.29.141.113    上部署  Nginx  与 fastcgi (作为热备)

172.29.141.115    上部署  MySQL

 

三.  程序源码

cat monitor_nginx.sh

#!/bin/bash #author longxibendi #blog http://blog.youkuaiyun.com/longxibendi #function failover the web server and send message to SA # send email to longxibendi@139.com function_sendEmail () { /usr/local/bin/sendEmail -f monitor_sys@163.com -t longxibendi@139.com -s smtp.163.com -u "web sever down" -xu monitor_sys -xp 123456789 -m "` date ` " } # change ip address function_change_ip () { /sbin/ifconfig eth0 172.29.141.112 broadcast 172.29.141.255 netmask 255.255.255.0 up /sbin/route add -host 172.29.141.112 dev eth0 /sbin/arping -I eth0 -c 3 -s 172.29.141.112 172.29.141.115 ### /sbin/arping is to add arp list in 172.29.141.115 } # to start local nginx web server function_start_nginx_server () { ulimit -SHn 65535 /usr/local/webserver/php/sbin/php-fpm start /usr/local/webserver/nginx/sbin/nginx } # main function function_main () { while true do http_status_code=`curl -o /dev/null -s -w %{http_code} http://172.29.141.112/wordpress/` # echo "$http_status_code"; if [ ${http_status_code} != 200 ] ;then function_sendEmail ; function_change_ip ; function_start_nginx_server ; fi # execute curl per 3 seconds sleep 3 done } function_main ;  

 

四.程序说明

function_sendEmail                     发送email到139邮箱，报警

function_change_ip                     浮动更改ip，添加路由，修改db的arp表

function_start_nginx_server         启动本机nginx服务器，这一步也可以调整，如果本机的nginx已启动的话

function_main                            主函数，通过curl 网址得到状态码

 

基本思想有两条

1． 通过curl检测故障

2． 通过浮动修改(ifconfig) ip  实现故障切换(failover)

 

 

五.如何部署

在172.29.141.113上部署。如果在112上部署，不需要function_change_ip这个函数。这是一种简单的解决方案，也可以用keepalive，如果PV很大，可能需要用lvs+keepalive或者硬件负载均衡器。

 

 

六.参考资料

 

1. man   curl  这里给出重要的参数

       -w/--write-out <format>

              Defines what to display on stdout after a completed and success-

              ful operation. The format is a string  that  may  contain  plain

              text mixed with any number of variables. The string can be spec-

              ified as "string", to get read from a particular file you  spec-

              ify  it  "@filename"  and  to  tell curl to read the format from

              stdin you write "@-".

 

              The variables present in the output format will  be  substituted

              by  the  value or text that curl thinks fit, as described below.

              All variables are specified like %{variable_name} and to  output

              a normal % you just write them like %%. You can output a newline

              by using /n, a carriage return with /r and a tab space with  /t.

 

              NOTE: The %-letter is a special letter in the win32-environment,

              where all occurrences of %  must  be  doubled  when  using  this

              option.

 

              Available variables are at this point:

 

              url_effective  The  URL  that  was  fetched last. This is mostly

                             meaningful if you’ve told curl  to  follow  loca-

                             tion: headers.

 

              http_code      The  numerical  code  that  was found in the last

                             retrieved HTTP(S) page.

 

              http_connect   The numerical code that was  found  in  the  last

                             response   (from  a  proxy)  to  a  curl  CONNECT

                             request. (Added in 7.12.4)

 

              time_total     The total time, in seconds, that the full  opera-

                             tion lasted. The time will be displayed with mil-

                             lisecond resolution.

 

              time_namelookup

                             The time, in seconds,  it  took  from  the  start

                             until the name resolving was completed.

 

              time_connect   The  time,  in  seconds,  it  took from the start

                             until the connect to the remote host  (or  proxy)

                             was completed.

 

              time_pretransfer

                             The  time,  in  seconds,  it  took from the start

                             until the file transfer is just about  to  begin.

                             This includes all pre-transfer commands and nego-

                             tiations that are specific to the particular pro-

                             tocol(s) involved.

 

              time_redirect  The time, in seconds, it took for all redirection

                             steps include name lookup,  connect,  pretransfer

                             and   transfer   before   final  transaction  was

                             started. time_redirect shows the complete  execu-

                             tion  time  for  multiple redirections. (Added in

                             7.12.3)

 

              time_starttransfer

                             The time, in seconds,  it  took  from  the  start

                             until  the  first byte is just about to be trans-

                             ferred. This includes time_pretransfer  and  also

                             the  time  the  server  needs  to  calculate  the

                             result.

 

              size_download  The total amount of bytes that were downloaded.

 

              size_upload    The total amount of bytes that were uploaded.

 

              size_header    The total amount of bytes of the downloaded head-

                             ers.

 

              size_request   The  total  amount of bytes that were sent in the

                             HTTP request.

 

              speed_download The average download speed that curl measured for

                             the complete download.

 

              speed_upload   The  average  upload speed that curl measured for

                             the complete upload.

 

              content_type   The Content-Type of the  requested  document,  if

                             there was any.

 

              num_connects   Number  of new connects made in the recent trans-

                             fer. (Added in 7.12.3)

 

              num_redirects  Number of redirects that  were  followed  in  the

                             request. (Added in 7.12.3)

 

              ftp_entry_path The initial path libcurl ended up in when logging

                             on to the remote FTP server. (Added in 7.15.4)

2.HTTP状态码参考

http://baike.baidu.com/view/1790469.htm

 

 

 

声明：本文档可以随意更改，但必须署名原作者

作者：凤凰舞者 qq:578989855