shell脚本监控VPS系统资源

几个月前开始使用VPS，每月限制300GB流量，流量方便基本够用了，但是有时候由于受到一些恶意访问，导致CPU、Memory等资源消耗较大，导致VPS上的博客网站响应时间太慢甚至有时根本无法打开网页。所以，我简单做了个脚本来进行监控和发邮件报警。
由于不是做很专业的运维，暂时不想上很专业的监控工具，如：Nagios、Puppet、Cacti、Zabbix等，所以自己手动了个Shell脚本来监控我关心的CPU、Load、Memory、网络传输、博客网站能否打开等内容。将该监控脚本分享如下：
https://github.com/smilejay/shell/blob/master/sh2013/vps_monitor.sh

#!/bin/bash
#set -x
# the script to monitor my VPS
# It will alert when memory, load, CPU%, networking, httpd/mysqld or home-page
# is in an abnormal state.
# author: Jay
# date: 2013-10-16

EMAIL="[email protected]"
WARN_MSG=""

# alert when free memory is less than 50 MB
function mem_free()
{
threshold=50 # 50 MB free memory
free_mem=$(free -m | grep "buffers/cache" | awk '{print $4}')
if [ $free_mem -lt $threshold ]; then
WARN_MSG=$WARN_MSG"Free memeory is less than $threshold MB.\n"
return 1
fi
return 0
}

# alert when load (5min) is larger than 4
function load_avg()
{
threshold=4 # load is 4
load_5min=$(cat /proc/loadavg | awk '{print $2}')
if [ $(echo "$load_5min > $threshold" | bc) -eq 1 ]; then
WARN_MSG=$WARN_MSG"5min average load is larger than $threshold.\n"
return 1
fi
return 0
}

# alert when CPU idle% is less than 20%
function cpu_idle()
{
threshold=20 # CPU idle% 20%
cpu_idle=$(sar 1 5 | grep -i 'Average' | awk '{print $NF}')
if [ $(echo "$cpu_idle < $threshold" | bc) -eq 1 ]; then
# in printf cmd, %% represents a single %
WARN_MSG=$WARN_MSG"CPU idle%% is less than $threshold%%.\n"
return 1
fi
return 0
}

# alert when networking tx speed is larger than 80 kB/s
function network_tx()
{
threshold=80 # TX speed 80 kB/s
interface=eth0
tx_speed=$(sar -n DEV 10 5 | grep "Average" | grep "$interface" | awk '{print $6}')
if [ $(echo "$tx_speed > $threshold" | bc) -eq 1 ]; then
WARN_MSG=$WARN_MSG"networking TX speed is larger than $threshold kB/s.\n"
return 1
fi
return 0
}

# alert when httpd or mysqld process doesn't exist
function httpd_mysqld()
{
ps -ef | grep 'httpd' | grep -v 'grep'
if [ $? -ne 0 ]; then
WARN_MSG=$WARN_MSG"httpd process doesn't exist.\n"
return 1
else
ps -ef | grep 'mysqld' | grep -v 'grep'
if [ $? -ne 0 ]; then
WARN_MSG=$WARN_MSG"mysqld process doesn't exist.\n"
return 1
fi
fi
return 0
}

# alert when 'Stay hungry' doesn't exist in the home page http://smilejay.com/
function home_page()
{
url=http://smilejay.com/
keywords="Stay hungry"
curl -sL $url | grep -i "$keywords"
if [ $? -ne 0 ]; then
WARN_MSG=$WARN_MSG"keywords '$keywords' doesn't exist on link $url.\n"
return 1
fi
return 0
}

# use an array to store the return value of each monitoring function
mem_free
chk[1]=$?
load_avg
chk[2]=$?
cpu_idle
chk[3]=$?
network_tx
chk[4]=$?
httpd_mysqld
chk[5]=$?
home_page
chk[6]=$?

# send warning email to the web master if any of the check fails.
for i in $(seq 1 6)
do
if [ ${chk[i]} -ne 0 ]; then
printf "$WARN_MSG" | mailx -s "Warning from smilejay.com" $EMAIL
exit 1
fi
done