Nagios监控搭建和配置(笔记)

   废话不多说,本博文纯属于个人笔记,可能会出现杂乱无章的感觉,只是把遇到的问题一一的记录下来,方便日后查看,也能帮助遇到类型问题的还在纠结的人。

十载的吉林网站建设经验,针对设计、前端、开发、售后、文案、推广等六对一服务,响应快,48小时及时工作处理。成都营销网站建设的优势是能够根据用户设备显示端的尺寸不同,自动调整吉林建站的显示方式,使网站能够适用不同显示终端,在浏览器中调整网站的宽度,无论在任何一种浏览器上浏览网站,都能展现优雅布局与设计,从而大程度地提升浏览体验。创新互联从事“吉林网站设计”,“吉林网站推广”以来,每个客户项目都认真落实执行。


系统版本及信息

cat /etc/redhat-release 
CentOS release 6.2 (Final)

uname -a
Linux  2.6.32-220.el6.x86_64  x86_64 x86_64 x86_64 GNU/Linux

ifconfig |sed -n 1,2p
eth0      Link encap:Ethernet  HWaddr 40:F2:E9:29:5F:EA  
          inet addr:192.168.0.2  Bcast:192.168.69.255  Mask:255.255.255.0
          
关闭 Iptables  selinux

 软件版本信息

LAMP/LNMP 忽略,任何一个环境都可以,我这里是yum 安装的LNMP环境
nagios-4.0.5.tar.gz
nagios-plugins-1.4.16.tar.gz
nrpe-2.15.tar.gz
pnp4nagios-0.6.19.tar.gz

安装Nagios软件准备工作

确保 yum 能正常使用,建议是配置网络 yum ,安装系统所需库文件
yum groupinstall "Compatibility libraries" "Base" "Development tools"

安装lamp及所需包

yum -y install http* php* MySQL* perl* net-snmp* openssl* glibc rrdtoolrrdtool-devel rrdtool-perl rrdtool-php

chkconfig mysqld   on                     
chkconfig httpd    on
chkconfig snmpd   on
service httpd  start
service mysqld start
service snmpd start

测试ok 继续下一步
ps -ef | grep -v grep | grep http mysql snmp #分别查看,web页面访问测试

安装Nagios

1、创建nagios程序用户、组
[root@nagios ~]# useradd -s /sbin/nologin nagios
[root@nagios ~]# mkdir /usr/local/nagios
[root@nagios ~]# chown -R nagios.nagios /usr/local/nagios/

2、编译安装nagios

[root@nagios tools]# tar zxf nagios-4.0.5.tar.gz
[root@nagios tools]# cd nagios-4.0.5
[root@nagios nagios-4.0.5]# ./configure --prefix=/usr/local/nagios
[root@nagios nagios-4.0.5]# make all &&make install && make install-init && make install-commandmode&& make install-config && make install-webconf
[root@nagios nagios-4.0.5]# echo $?
0
3、加入开机启动
chkconfig --add nagios
chkconfig nagios on
chkconfig--list nagios

安装nagios-plugins 插件

[root@nagios tools]# tar zxf nagios-plugins-1.4.16.tar.gz
[root@nagios tools]# cd nagios-plugins-1.4.16
[root@nagios tools nagios-plugins-1.4.16]# ./configure --prefix=/usr/local/nagios/
[root@nagios tools nagios-plugins-1.4.16]# make
[root@nagios tools nagios-plugins-1.4.16]# make install
[root@nagios tools nagios-plugins-1.4.16]# echo $?
0

编辑http.conf配置文件

cd /etc/httpd/conf
cp -a httpd.conf httpd.conf.bak

vim httpd.conf # 添加在最后面即可

####### setting for nagios #######
ScriptAlias /nagios/cgi-bin "/usr/local/nagios/sbin"

AuthType Basic
Options ExecCGI
AllowOverride None
Order allow,deny
Allow from all
AuthName "nagios access"
AuthUserFile /usr/local/nagios/etc/htpasswd
Require valid-user

Alias /nagios "/usr/local/nagios/share"

AuthType Basic
Options ExecCGI
AllowOverride None
Order allow,deny
Allow from all
AuthName "nagios access"
AuthUserFile /usr/local/nagios/etc/htpasswd
Require valid-user


修改
DirectoryIndex index.html index.html.var
为
DirectoryIndex index.php index.html index.html.var

修改
Options Indexes FollowSymLinks
为
Options FollowSymLinks    #防止网站列目录

service httpd restart 

增加nagios登陆认证文件,一定要用默认的nagiosadmin作为用户,否则需要修改其他文件,修改之前备份,这里就不备份了

[root@nagios etc]# cd /usr/local/nagios/etc
[root@nagios etc]# sed -i s@nagiosadmin@nagiosadmin\,admin@g cgi.cfg
[root@nagios etc]# sed -i s@\#default_user_name=guest@default_user_name=admin@g cgi.cfg
[root@nagios nagios]# htpasswd -c /usr/local/nagios/etc/htpasswd admin
New password: ******
Re-type new password:******

安装 Nrpe 插件

[root@nagios tools]# tar zxf nrpe-2.15.tar.gz
[root@nagios tools]# cd nrpe-2.15
[root@nagios nrpe-2.15]# ./configure;make all;make install-plugin;make install-daemon;make install-daemon-config

启动Nrpe

[root@nagios nrpe-2.15]# /usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
[root@nagios nrpe-2.15]# netstat -antl |grep 5666
tcp        0      0 0.0.0.0:5666                0.0.0.0:*                   LISTEN  
[root@nagios libexec]#/usr/local/nagios/libexec/check_nrpe -H 127.0.0.1
NRPE v2.15

关闭Nrpe

[root@nagios libexec]# ps -ef | grep -v grep | grep nrpe
[root@nagios libexec]# kill -9 进程号

检测nagios

[root@nagios etc]# /usr/local/nagios/bin/nagios-v /usr/local/nagios/etc/nagios.cfg
Total Warnings: 0
Total Errors:   0 表示OK

启动nagios

[root@nagios etc]# service nagios start  stop  restart 开启  停止  重启

http://IP/nagios

安装 pnp4nagios 插件

[root@nagios tools]# tar zxf pnp4nagios-0.6.19.tar.gz
[root@nagios tools]# cd pnp4nagios-0.6.19
[root@nagios tools pnp4nagios-0.6.19]#./configure
make all
make install
make install-config
make install-init
make install-webconf

创建默认配置文件
cd /usr/local/pnp4nagios/etc
cp misccommands.cfg-sample misccommands.cfg
cp nagios.cfg-sample nagios.cfg
cp rra.cfg-sample rra.cfg
cd pages
cp web_traffic.cfg-sample web_traffic.cfg
cd ../check_commands/
cp check_all_local_disks.cfg-samplecheck_all_local_disks.cfg
cp check_nrpe.cfg-sample check_nrpe.cfg
cp check_nwstat.cfg-sample check_nwstat.cfg
cp /usr/local/pnp4nagios/libexec/* /usr/local/nagios/libexec/

vim /usr/local/nagios/etc/nagios.cfg
检查
enable_environment_macros=1
process_performance_data=1          
host_perfdata_command=process-host-perfdata
service_perfdata_command=process-service-perfdata

提示:如果nagios版本是4.X,上面配置会导致后面,生成不了流量图,报如下错误

PNP4Nagios Version 0.6.19
Please check the documentation for information about the following error.
perfdata directory "/usr/local/pnp4nagios/var/perfdata/localhost" for host "localhost" does not exist.Read FAQ online
file [line]:
application/models/data.php [148]:
back

Nagios监控搭建和配置(笔记)

出现这个错误的原因是参照
解决方案是使用 Bulk Mode方式

vim /usr/local/nagios/etc/nagios.cfg

检查
enable_environment_macros=1
process_performance_data=1 

添加到最后即可
# service performance data

service_perfdata_file=/usr/local/pnp4nagios/var/service-perfdata
service_perfdata_file_template=DATATYPE::SERVICEPERFDATA\tTIMET::$TIMET$\tHOSTNAME::$HOSTNAME$\tSERVICEDESC::$SERVICEDESC$\tSERVICEPERFDATA::$SERVICEPERFDATA$\tSERVICECHECKCOMMAND::$SERVICECHECKCOMMAND$\tHOSTSTATE::$HOSTSTATE$\tHOSTSTATETYPE::$HOSTSTATETYPE$\tSERVICESTATE::$SERVICESTATE$\tSERVICESTATETYPE::$SERVICESTATETYPE$
service_perfdata_file_mode=a
service_perfdata_file_processing_interval=15
service_perfdata_file_processing_command=process-service-perfdata-file

# host performance data starting with Nagios 3.0

host_perfdata_file=/usr/local/pnp4nagios/var/host-perfdata
host_perfdata_file_template=DATATYPE::HOSTPERFDATA\tTIMET::$TIMET$\tHOSTNAME::$HOSTNAME$\tHOSTPERFDATA::$HOSTPERFDATA$\tHOSTCHECKCOMMAND::$HOSTCHECKCOMMAND$\tHOSTSTATE::$HOSTSTATE$\tHOSTSTATETYPE::$HOSTSTATETYPE$
host_perfdata_file_mode=a
host_perfdata_file_processing_interval=15
host_perfdata_file_processing_command=process-host-perfdata-file
保存

vim  /usr/local/nagios/etc/objects/commands.cfg
define command{
       command_name    check_nrpe
       command_line    $USER1$/check_nrpe-H $HOSTADDRESS$ -c $ARG1$
} #这一段放在上面即可

如下:同步模式设定方法添加到末尾就可以,记住在这个配置文件里面, 默认有这个配置,需要找到注释掉,然后将下面的配置添加,如果不注释掉,在你检查nagios的配置文件的时候会报错

define command{
       command_name    process-service-perfdata-file
       command_line    /usr/local/pnp4nagios/libexec/process_perfdata.pl --bulk=/usr/local/pnp4nagios/var/service-perfdata
}

define command{
       command_name    process-host-perfdata-file
       command_line    /usr/local/pnp4nagios/libexec/process_perfdata.pl --bulk=/usr/local/pnp4nagios/var/host-perfdata
}

定义pnp的主机和服务两个模版添加在最后面
 
vim /usr/local/nagios/etc/objects/templates.cfg
define host {
  name       host-pnp
  action_url /pnp4nagios/index.php/graph?host=$HOSTNAME$&srv=_HOST_
  register  0
}
define service {
  name       service-pnp
  action_url /pnp4nagios/index.php/graph?host=$HOSTNAME$&srv=$SERVICEDESC$
  register  0
}

也可以添加在,其他参数下面省略了,下面这个方法可以减少很多配置主机启用pnp时的时间
vim /usr/local/nagios/etc/objects/templates.cfg
define host {

        action_url /pnp4nagios/index.php/graph?host=$HOSTNAME$&srv=_HOST_
}
define service {
        
       action_url /pnp4nagios/index.php/graph?host=$HOSTNAME$&srv=$SERVICEDESC$  
}

先做一下pnp4nagios环境测试添加在httpd.conf最后面

vim /etc/httpd/conf/httpd.conf

Alias /pnp4nagios "/usr/local/pnp4nagios/share"

AllowOverride None
Order allow,deny
Allow from all
AuthName "Nagios Access"
AuthType Basic
AuthUserFile /usr/local/nagios/etc/htpasswd
Require valid-user

 

RewirteEngine On
Options FollowSymLinks
RewirteBase /pnp4nagios
 RewirteRule ^(application|modules|system) -[F,L]
RewirteCond %{REQUEST_FILENAME} !-f
RewirteCond %{REQUEST_FILENAME} !-d
RewirteRule .* index.php/$0 [PT,L]


service httpd restart

访问 http://IP/pnp4nagios

Nagios监控搭建和配置(笔记)

cd /usr/local/pnp4nagios/share/

mv install.php install.php.bak

编辑nagios.cfg文件

vim /usr/local/nagios/etc/nagios.cfg

cfg_file=/usr/local/nagios/etc/objects/commands.cfg
cfg_file=/usr/local/nagios/etc/objects/contacts.cfg
cfg_file=/usr/local/nagios/etc/objects/timeperiods.cfg
cfg_file=/usr/local/nagios/etc/objects/templates.cfg
cfg_file=/usr/local/nagios/etc/objects/localhost.cfg
cfg_file=/usr/local/nagios/etc/objects/hosts.cfg
cfg_file=/usr/local/nagios/etc/objects/hostgroup.cfg
cfg_file=/usr/local/nagios/etc/objects/services.cfg

或者
cfg_file=/usr/local/nagios/etc/objects/commands.cfg
cfg_file=/usr/local/nagios/etc/objects/contacts.cfg
cfg_file=/usr/local/nagios/etc/objects/timeperiods.cfg
cfg_file=/usr/local/nagios/etc/objects/templates.cfg
cfg_file=/usr/local/nagios/etc/objects/localhost.cfg

cfg_dir=/usr/local/nagios/etc/objects/apps


提示:此操作只是启用了linux主机监控,没有启用windows和switch,如果需要把注释去掉即可,第一种和第二种都可以
区别是:第一种共同使用一个配置文件,第二种独立使用配置文件,这里我都会演示,下面以第一种和第二种进行区分

添加主机配置,第一种方法

默认nagios/etc/objects/ 下面没有 service.cfg host.cfg hostgroup.cfg 这几个配置文件,需要手动添加

vim hosts.cfg
 
define host{
         use                      linux-server,host-pnp #这个是根据templates.cfg信息定义,如果上面定义的模板host-pnp添加在define host和define sevice里面,这儿host-pnp可以不用加,因为linux-server已经包含了
         host_name                cacti                 #必须是 被监控的主机名
         alias                    cacti-web             #别名随便定义
         address                  192.168.0.3           #主机ip地址
         contact_groups           admins                #邮件组,下面会演示
}
define host{
         use                      linux-server,host-pnp 
         host_name                nginx                 
         alias                    nginx-web             
         address                  192.168.0.4           
         contact_groups           admins                
}
有多少机器就这样添加多少台

vim hostgroup.cfg
define hostgroup{
         hostgroup_name      servers                   #组名
         alias               servers_group             #别名
         members             cacti,nginx               #主机名 多个 逗号 隔开
}

vim service.cfg          #所有主机在同一配置文件,很乱

#### set cacti host  
define service{
         use                  local-service,services-pnp
         host_name            cacti
         service_description  http
         check_command        check_http
         contact_groups       admins
         flap_detection_enabled          0
}
 
define service{
         use                   local-service,services-pnp
         host_name             cacti
         service_description   SSH_port
         check_command         check_tcp!22 
         contact_groups        admins
         flap_detection_enabled          0
}
define service{
       use                      local-service,services-pnp
       host_name                cacti
       service_description      check_/
       check_command            check_nrpe!check_/  #使用nrpe检测,客户端需要定义
       contact_groups           admins
       flap_detection_enabled   0
}
    
#### set nginx host

define service{
        use                      local-service,service-pnp
        host_name                nginx
        service_description      Check_free_mem
        check_command            check_nrpe!check_free_mem
        contact_groups           admins
        flap_detection_enabled   0
}
 define service{
       use                      local-service,services-pnp
       host_name                nginx
       service_description      check_/
       check_command            check_nrpe!check_/  #使用nrpe检测,客户端需要定义
       contact_groups           admins
       flap_detection_enabled   0            
}
有多少就需要添加多少,第一种方法 end

添加主机配置,第二种方法

cd nagios/etc/objects/
mkdir app
cd app

vim 192.168.0.2.cfg  #在一个独立的文件定义所有监控对象,这个没有定义组,意义不大

###定义host

define host{
         use                      linux-server,host-pnp #这个是根据templates.cfg信息定义,如果上面定义的模板host-pnp添加在define host和define sevice里面,这儿host-pnp可以不用加,因为linux-server已经包含了
         host_name                nginx                 #必须是 被监控的主机名
         alias                    nginx-web             #别名随便定义
         address                  192.168.0.4           #主机ip地址
         contact_groups           admins                #邮件组,下面会演示
}

###定义service

define service{
        use                      local-service,service-pnp
        host_name                nginx
        service_description      Check_free_mem
        check_command            check_nrpe!check_free_mem
        contact_groups           admins
        flap_detection_enabled   0
}

 define service{
       use                      local-service,services-pnp
       host_name                nginx
       service_description      check_/
       check_command            check_nrpe!check_/  #使用nrpe检测,客户端需要定义
       contact_groups           admins
       flap_detection_enabled   0            
}
vim 192.168.0.3.cfg

###定义host

define host{
         use                      linux-server,host-pnp #这个是根据templates.cfg信息定义,如果上面定义的模板host-pnp添加在define host和define sevice里面,这儿host-pnp可以不用加,因为linux-server已经包含了
         host_name                cacti                 #必须是 被监控的主机名
         alias                    cacti-web             #别名随便定义
         address                  192.168.0.3           #主机ip地址
         contact_groups           admins                #邮件组,下面会演示
}

###定义service

define service{
        use                      local-service,service-pnp
        host_name                cacti
        service_description      Check_free_mem
        check_command            check_nrpe!check_free_mem
        contact_groups           admins
        flap_detection_enabled   0
}

define service{
        use                      local-service,service-pnp
        host_name                cacti
        service_description      Check_free_mem
        check_command            check_nrpe!check_free_mem
        contact_groups           admins
        flap_detection_enabled   0
}

这种办法比第一种方便许多,添加主机2种方法 END

nagios邮件报警设置

[root@nagios objects]# vim contacts.cfg  #参数详解,请百度
define contact{
       contact_name                   nagiosadmin    
       use                            generic-contact       
       alias                          Nagios Admin                                              service_notification_period        24x7
       host_notification_period           24x7
       service_notification_options       w,u,c,r
       host_notification_options          d,u,r
       service_notification_commands      notify-service-by-email        
       host_notification_commands         notify-host-by-email
       email                              xxxx@163.com  
       }
define contactgroup{
       contactgroup_name          admins  #这个就是上面那个admins
       alias                      Nagios Administrators
       members                    nagiosadmin
       }

检查配置文件是否有错

/usr/local/nagios/bin/nagios -v /usr/local/nagios/etc/nagios.cfg


Total Warnings: 0
Total Errors:   0

Things look okay - No serious problems were detected during the pre-flight check

service nagios restart

服务端配置 end

 客户端安装配置

需要安装net-snmp,如果有其他错误根据提示进行解决
yum -y install net-snmp*

1、创建nagios程序用户、组
[root@nagios ~]# useradd -s /sbin/nologin nagios
[root@nagios ~]# mkdir /usr/local/nagios
[root@nagios ~]# chown -R nagios.nagios /usr/local/nagios/

2、安装nagios-plugins 插件
[root@nagios tools]# tar zxf nagios-plugins-1.4.16.tar.gz
[root@nagios tools]# cd nagios-plugins-1.4.16
[root@nagios tools nagios-plugins-1.4.16]# ./configure --prefix=/usr/local/nagios/
[root@nagios tools nagios-plugins-1.4.16]# make
[root@nagios tools nagios-plugins-1.4.16]# make install
[root@nagios tools nagios-plugins-1.4.16]# echo $?
0

3、安装 Nrpe 插件
[root@nagios tools]# tar zxf nrpe-2.15.tar.gz
[root@nagios tools]# cd nrpe-2.15
[root@nagios nrpe-2.15]# ./configure;make all;make install-plugin;make install-daemon;make install-daemon-config

编辑nrpe.cfg

sed -I 's/allowed_hosts=127.0.0.1/allowed_hosts=127.0.0.1,192.168.0.2/g' /usr/local/nagios/etc/nrpe.cfg

vim /usr/local/nagios/etc/nrpe.cfg

command[check_swap]=/usr/local/nagios/libexec/check_swap -w 20% -c 10%
command[check_data]=/usr/local/nagios/libexec/check_disk -w 20% -c 10% -p /data
command[check_/]=/usr/local/nagios/libexec/check_disk -w 20% -c 10% -p /
command[check_zombie_procs]=/usr/local/nagios/libexec/check_procs -w 5 -c 10 -s Z
command[check_total_procs]=/usr/local/nagios/libexec/check_procs -w 150 -c 200 
保存

echo "/usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg-d" >> /etc/rc.local

启动Nrpe

[root@nagios nrpe-2.15]# /usr/local/nagios/bin/nrpe -c /usr/local/nagios/etc/nrpe.cfg -d
[root@nagios nrpe-2.15]# netstat -antl |grep 5666
tcp        0      0 0.0.0.0:5666                0.0.0.0:*                   LISTEN  

这个在服务端操作,确保ok,如果不能请检查客户端防火墙和网络是否允许通信

[root@nagios libexec]#/usr/local/nagios/libexec/check_nrpe -H 192.168.0.3
NRPE v2.15

关闭Nrpe

[root@nagios libexec]# ps -ef | grep -v grep | grep nrpe
[root@nagios libexec]# kill -9 进程号


pnp不出图时候,查看日志

vim /usr/local/pnp4nagios/etc/process_perfdata.cfg

修改

LOG_LEVEL = 0

LOG_LEVEL = 2

more /usr/local/pnp4nagios/var/perfdata.log

提示:nagios 监控进程时候,即便pnp配置ok,也不会出图,例如下面的

Total Processes
Nagios监控搭建和配置(笔记)
OK10-20-2014 16:44:4583d 1h 9m 16s1/3PROCS OK: 503 processes 
zombie_procs
Nagios监控搭建和配置(笔记)
OK10-20-2014 16:46:0083d 1h 7m 58s1/3PROCS OK: 0 processes with STATE = Z

PNP4Nagios Version 0.6.19

Please check the documentation for information about the following error.

XML file "/usr/local/pnp4nagios/var/perfdata/app-11/Total_Processes.xml" not found. Read FAQ online

file [line]:

application/models/data.php [312]:

back

Nagios监控搭建和配置(笔记)

至于原因可以参考,非常详细

http://storysky.blog.51cto.com/628458/583787/

Nagios如果系统监控插件满足不了需求,可以自行开发插件


例如下面是一个内存监控插件,插件是百度找的还是不错的,我这里借用一下

vim /usr/local/nagios/libexec/check_mem

#!/bin/bash
STAT_OK=0
STAT_WARNING=1
STAT_CRITICAL=2
STAT_UNKNOWN=3

total_mem=`free -m |awk 'NR==2{print $2}'`
used_mem=`free -m |awk 'NR==3{print $3}'` #取的是系统真正用掉的内存
free_mem=`free -m |awk 'NR==3{print $4}'` #取的是free+cache的内存
use_per=`echo "scale=2;$used_mem/$total_mem"|bc|sed 's/^.//g'`

help() {
        echo "USAGE:`basename $0` [-w]  [-c]  [-h]"
        exit -1
}

while getopts ":w:c:h" opt
do
        case $opt in
                w)      warning=$OPTARG
                        ;;
                c)      critical=$OPTARG
                        ;;
                h)      help
                        ;;
                ?)      unkown=$OPTARG
                        echo "error,plase check for help,USAGE:./`basename $0` -h"
                        exit $STAT_UNKNOWN
                        ;;
        esac
done


if [[ $use_per -lt $warning ]];
then
        echo "OK - total:$total_mem MB,used:$used_mem MB,free:$free_mem MB | total_mem=$total_mem used_mem=$used_mem free_mem=$free_mem"
        exit $STAT_OK

elif [[ $use_per -ge $warning ]] && [[ $use_per -lt $critical ]];
then
                 echo "WARNING - total:$total_mem MB,used:$used_mem MB,free:$free_mem MB | total_mem=$total_mem used_mem=$used_mem free_mem=$free_mem"
                 exit $STAT_WARNING
else
                 echo "CRITICAL - total:$total_mem MB,used:$used_mem MB,free:$free_mem MB | total_mem=$total_mem used_mem=$used_mem free_mem=$free_mem"
                 exit $STAT_CRITICAL
        fi
fi
保存

chown nagios.nagios check_mem
chmod +x check_mem
./check_mem -w 80 -c 90
OK - total:15926 MB,used:1839 MB,free:14086 MB | total_mem=15926 used_mem=1839 free_mem=14086

vim /usr/local/nagios/etc/nrpe.cfg

添加
command[check_free_mem]=/usr/local/nagios/libexec/check_mem -w 80 -c 90

重启nrpe

在编辑/usr/local/nagios/etc/objects/app/的文件
添加
define service{
        use                      local-service,service-pnp
        host_name                cacti
        service_description      Check_free_mem
        check_command            check_nrpe!check_free_mem
        contact_groups           admins
        flap_detection_enabled   0
}

检查nagios 重启nagios

   Windows和交换机监控配置不难,只要思路清晰,肯定能弄出来,nagios配置其实不难,就是有点麻烦而已,只要把配置文件的关系弄明白,一切都很简单

到此全部结束


当前文章:Nagios监控搭建和配置(笔记)
标题URL:http://myzitong.com/article/pdhioo.html