Nagios+pnp4nagios+rrdtool 安装配置为nagios添加自定义插件(三)

nagios博大精深，可以以shell、perl等语句为nagios写插件，来满足自己监控的需要。本文写mysql中tps、qps的插件，并把收集到的结果以图形形式展现出来，这样输出的结果就有一定的要求了。

编写插件tps  qps

check_qps 插件如下内容

#!/bin/sh

mytool="/usr/local/mysql/bin/mysql-umy_perfor -pmy_perfor"

state_ok=0

state_warning=1

state_critical=2

state_unknown=3

Uptime=`echo "show  /*50000 global */ status like'uptime'"|$mytool -N`

for i in $Uptime

do

uptime_new=$i

done

$mytool -e "select total_numfrom test.monitor_status where statu_item='Uptime' order by id desc limit1">/home/zhaohp/monitor.txt

sed -i -e '1d'/home/zhaohp/monitor.txt

uptime_old=`cat/home/zhaohp/monitor.txt`

$mytool -e "insert intotest.monitor_status values (0,'Uptime',$uptime_new,now())"

if [ -n "$uptime_old" ]

then

uptime=$[$uptime_new-$uptime_old]

#  echo $uptime

fi

# ***********************check_qps*******************************

Question=`echo "show  /*50000 global */ status like'Queries'"|$mytool -N`

for i in $Question

do

queries_new=$i

done

$mytool -e "select total_numfrom test.monitor_status where statu_item='Question' order by id desc limit1" >/home/zhaohp/monitor.txt

sed -i -e '1d'/home/zhaohp/monitor.txt

queries_old=`cat/home/zhaohp/monitor.txt`

$mytool -e "insert intotest.monitor_status values (0,'Question',$queries_new,now())"

if [ -n  "$uptime_old" ]

then

qps=$((($queries_new-$queries_old)/$uptime))

# echo $queries_new

# echo $queries_old

# echo $uptime

if [ $qps -le 10000 ]; then

echo "qps is ok - qps is $qps |qps=$qps;15000;20000"

exit $state_ok

elif [ $qps -le 15000 ]; then

echo "warning - qps is $qps |qps=$qps;15000;20000"

exit $state_warning

elif [ $qps -le 20000 ];  then

echo " critical - qps is $qps |qps=$qps;15000;20000"

exit $state_critical

else

echo "unkown"

exit $state_unknown

fi

fi

check_tps 插件如下内容

#!/bin/sh

mytool="/usr/local/mysql/bin/mysql-umy_perfor -pmy_perfor"

state_ok=0

state_warning=1

state_critical=2

state_unknown=3

Uptime=`echo "show  /*50000 global */ status like'uptime'"|$mytool -N`

for i in $Uptime

do

uptime_new=$i

done

$mytool -e "select total_numfrom test.monitor_tps where statu_item='Uptime' order by id desc limit1">/home/zhaohp/monitor_uptime.txt

sed -i -e '1d'/home/zhaohp/monitor_uptime.txt

uptime_old=`cat/home/zhaohp/monitor_uptime.txt`

$mytool -e "insert intotest.monitor_tps values (0,'Uptime',$uptime_new,now())"

if [ -n "$uptime_old" ]

then

uptime=$[$uptime_new-$uptime_old]

#  echo $uptime

fi

# *********************** TPS monitor********************************

commit=`echo "show globalstatus like 'Com_commit'"|$mytool -N`

for i in $commit

do

commit_new=$i

done

rollback=`echo "show globalstatus like 'Com_rollback'"|$mytool -N`

for i in $rollback

do

rollback_new=$i

done

$mytool -e "select total_numfrom test.monitor_tps where statu_item='com_commit' order by id desc limit1" >/home/zhaohp/monitor_commit.txt

sed -i -e '1d'/home/zhaohp/monitor_commit.txt

commit_old=`cat/home/zhaohp/monitor_commit.txt`

#   echo "commit_old is$commit_old "

$mytool -e "select total_numfrom test.monitor_tps where statu_item='com_rollback' order by id desc limit1" >/home/zhaohp/monitor_rollback.txt

sed -i -e '1d'/home/zhaohp/monitor_rollback.txt

rollback_old=`cat/home/zhaohp/monitor_rollback.txt`

# echo " rollback is $rollback_old"

$mytool -e "insert intotest.monitor_tps values (0,'com_commit',$commit_new,now())"

$mytool -e "insert intotest.monitor_tps values (0,'com_rollback',$rollback_new,now())"

if [ -n "$uptime_old" ]

then

#tps=$((($commit_new+$rollback_new-$commit_old-$rollback_old)/$uptime))

total_new=$(($commit_new+$rollback_new))

#   echo "total_new is $total_new"

total_old=$(($commit_old+$rollback_old))

#   echo "total_old is  $total_old"

tps=$((($total_new-$total_old)/$uptime))

if [ $tps -le 500 ]; then

echo "tps is ok - tps is$tps  | tps=$tps;600;700"

exit $state_ok

elif [ $tps -le 600 ]; then

echo "warning - tps is $tps |tps=$tps;600;700"

exit $state_warning

elif [ $tps -le 700 ];  then

echo " critical - tps is $tps |tps=$tps;600;700"

exit $state_critical

else

echo "unkown"

exit $state_unknown

fi

fi

需要两张表 放在test库

CREATETABLE `monitor_status` (

`id` bigint(20) NOT NULL AUTO_INCREMENT,

`statu_item` varchar(50) DEFAULT '',

`total_num` bigint(20) DEFAULT '0',

`CreateDate` timestamp NOT NULL DEFAULT'0000-00-00 00:00:00',

PRIMARY KEY (`id`)

)ENGINE=MyISAM  DEFAULT CHARSET=utf8;

CREATETABLE `monitor_tps` (

`id` bigint(20) NOT NULL AUTO_INCREMENT,

`statu_item` varchar(50) DEFAULT '',

`total_num` bigint(20) DEFAULT '0',

`CreateDate` timestamp NOT NULL DEFAULT'0000-00-00 00:00:00',

PRIMARY KEY (`id`)

)ENGINE=MyISAM DEFAULT CHARSET=utf8;

创建用户，创建的用户既用于此脚本又用于check_mysql_health插件

grantselect,update,insert,delete on test.* to 'my_perfor'@'localhost' identified by'my_perfor';

给脚本 check_qps、check_tps 赋权

chmod 755 /usr/local/nagios/libexec/check_qps

chmod 755  /usr/local/nagios/libexec/check_tps

创建存放文件目录

mkdir  /home/zhaohp/

改变属主和权限

chown-R nagios.nagios /home/zhaohp/

chmod-R 777 /home/zhaohp/

注意脚本插件输出格式，不然pnp 脚本抓取不到数据

为什么不能出图

FAQ :

、为什么/usr/local/nagios/share/perfdata/目录中生成不出数据

确认是否是权限问题。因为当时装好PNP时太性急了，看到PERFDATA目录没有生成数据就自己手动创建了两个主机名称的目录，（因为RRDTOOL需要过一会才会创建数据）,这样导致权限不对，无法生成数据
因为创建这两目录所属用户和组成了root， nagios用户没有权限写入到root权限的目录中。
所以才创建不出数据。

、为什么/usr/local/nagios/share/perfdata目录中还是没有数据生成，

解决方法：安装pnp的时候是否make install-config安装了模板安装了这些模板后进入/usr/local/nagios/etc/pnp目录中去掉后面的扩展。

是否在services.cfg文件中为服务添加了process_perf_data 1

这一项。否则不会生成数据的

我们在做完上面的这些后发现还是没有数据产生，这时你就要等五分到十分钟。因为pnp需要这么久才能产生数据，才能出图。