专栏文章 性能测试连载 (33)-Shell 脚本分析服务器性能瓶颈

飞天小子的性能课堂 · 2019年12月11日 · 最后由 徐汪成 回复于 2019年12月11日 · 4625 次阅读

性能答疑 QQ 群:697244251

概述

我们原先在服务器上想分析硬件瓶颈,需要执行一系列的 linux 命令。对于 linux 命令不熟悉的人来说,比较困难。现在有一套集成的 shell 脚本,把常用的 linux 命令都包含在里面,一键式分析性能瓶颈

脚本功能

1、查看 CPU 利用率与负载(top、vmstat、sar)
2、查看磁盘、I/O 负载(df、iostat、iotop、sar、dstat)
3、查看内存利用率(free、vmstat)
4、查看 TCP 连接状态(netstat、ss)
5、查看 CPU 与内存占用最高的 10 个进程(top、ps)
6、查看网络流量(ifconfig、iftop、iptraf)

脚本内容

#!/bin/bash
#
os_check() {
        if [ -e /etc/redhat-release ]; then
                REDHAT=`cat /etc/redhat-release |cut -d' '  -f1`
        else
                DEBIAN=`cat /etc/issue |cut -d' ' -f1`
        fi
        if [ "$REDHAT" == "CentOS" -o "$REDHAT" == "Red" ]; then
                P_M=yum
        elif [ "$DEBIAN" == "Ubuntu" -o "$DEBIAN" == "ubutnu" ]; then
                P_M=apt-get
        else
                Operating system does not support.
                exit 1
        fi
}
if [ $LOGNAME != root ]; then
    echo "Please use the root account operation."
    exit 1
fi
if ! which vmstat &>/dev/null; then
        echo "vmstat command not found, now the install."
        sleep 1
        os_check
        $P_M install procps -y
        echo "-----------------------------------------------------------------------"
fi
if ! which iostat &>/dev/null; then
        echo "iostat command not found, now the install."
        sleep 1
        os_check
        $P_M install sysstat -y
        echo "-----------------------------------------------------------------------"
fi

while true; do
    select input in cpu_load disk_load disk_use disk_inode mem_use tcp_status cpu_top10 mem_top10 traffic quit; do
        case $input in
            cpu_load)
                #CPU利用率与负载
                echo "---------------------------------------"
                i=1
                while [[ $i -le 3 ]]; do
                    echo -e "\033[32m  参考值${i}\033[0m"
                    UTIL=`vmstat |awk '{if(NR==3)print 100-$15"%"}'`
                    USER=`vmstat |awk '{if(NR==3)print $13"%"}'`
                    SYS=`vmstat |awk '{if(NR==3)print $14"%"}'`
                    IOWAIT=`vmstat |awk '{if(NR==3)print $16"%"}'`
                    echo "Util: $UTIL"
                    echo "User use: $USER"
                    echo "System use: $SYS"
                    echo "I/O wait: $IOWAIT"
                    i=$(($i+1))
                    sleep 1
                done
                echo "---------------------------------------"
                break
                ;;
            disk_load)
                #硬盘I/O负载
                echo "---------------------------------------"
                i=1
                while [[ $i -le 3 ]]; do
                    echo -e "\033[32m  参考值${i}\033[0m"
                    UTIL=`iostat -x -k |awk '/^[v|s]/{OFS=": ";print $1,$NF"%"}'`
                    READ=`iostat -x -k |awk '/^[v|s]/{OFS=": ";print $1,$6"KB"}'`
                    WRITE=`iostat -x -k |awk '/^[v|s]/{OFS=": ";print $1,$7"KB"}'`
                    IOWAIT=`vmstat |awk '{if(NR==3)print $16"%"}'`
                    echo -e "Util:"
                    echo -e "${UTIL}"
                    echo -e "I/O Wait: $IOWAIT"
                    echo -e "Read/s:\n$READ"
                    echo -e "Write/s:\n$WRITE"
                    i=$(($i+1))
                    sleep 1
                done
                echo "---------------------------------------"
                break
                ;;
            disk_use)
                #硬盘利用率
                DISK_LOG=/tmp/disk_use.tmp
                DISK_TOTAL=`fdisk -l |awk '/^Disk.*bytes/&&/\/dev/{printf $2" ";printf "%d",$3;print "GB"}'`
                USE_RATE=`df -h |awk '/^\/dev/{print int($5)}'`
                for i in $USE_RATE; do
                    if [ $i -gt 90 ];then
                        PART=`df -h |awk '{if(int($5)=='''$i''') print $6}'`
                        echo "$PART = ${i}%" >> $DISK_LOG
                    fi
                done
                echo "---------------------------------------"
                echo -e "Disk total:\n${DISK_TOTAL}"
                if [ -f $DISK_LOG ]; then
                    echo "---------------------------------------"
                    cat $DISK_LOG
                    echo "---------------------------------------"
                    rm -f $DISK_LOG
                else
                    echo "---------------------------------------"
                    echo "Disk use rate no than 90% of the partition."
                    echo "---------------------------------------"
                fi
                break
                ;;
            disk_inode)
                #硬盘inode利用率
                INODE_LOG=/tmp/inode_use.tmp
                INODE_USE=`df -i |awk '/^\/dev/{print int($5)}'`
                for i in $INODE_USE; do
                    if [ $i -gt 90 ]; then
                        PART=`df -h |awk '{if(int($5)=='''$i''') print $6}'`
                        echo "$PART = ${i}%" >> $INODE_LOG
                    fi
                done
                if [ -f $INODE_LOG ]; then
                    echo "---------------------------------------"
                    rm -f $INODE_LOG
                else
                    echo "---------------------------------------"
                    echo "Inode use rate no than 90% of the partition."
                    echo "---------------------------------------"
                fi
                break
                ;;
            mem_use)
                #内存利用率
                echo "---------------------------------------"
                MEM_TOTAL=`free -m |awk '{if(NR==2)printf "%.1f",$2/1024}END{print "G"}'`
                USE=`free -m |awk '{if(NR==3) printf "%.1f",$3/1024}END{print "G"}'`
                FREE=`free -m |awk '{if(NR==3) printf "%.1f",$4/1024}END{print "G"}'`
                CACHE=`free -m |awk '{if(NR==2) printf "%.1f",($6+$7)/1024}END{print "G"}'`
                echo -e "Total: $MEM_TOTAL"
                echo -e "Use: $USE"
                echo -e "Free: $FREE"
                echo -e "Cache: $CACHE"
                echo "---------------------------------------"
                break
                ;;
            tcp_status)
                #网络连接状态
                echo "---------------------------------------"
                COUNT=`netstat -antp |awk '{status[$6]++}END{for(i in status) print i,status[i]}'`
                echo -e "TCP connection status:\n$COUNT"
                echo "---------------------------------------"
                ;;
            cpu_top10)
                #占用CPU高的前10个进程
                echo "---------------------------------------"
                CPU_LOG=/tmp/cpu_top.tmp
                i=1
                while [[ $i -le 3 ]]; do
                    #ps aux |awk '{if($3>0.1)print "CPU: "$3"% -->",$11,$12,$13,$14,$15,$16,"(PID:"$2")" |"sort -k2 -nr |head -n 10"}' > $CPU_LOG
                    ps aux |awk '{if($3>0.1){{printf "PID: "$2" CPU: "$3"% --> "}for(i=11;i<=NF;i++)if(i==NF)printf $i"\n";else printf $i}}' |sort -k4 -nr |head -10 > $CPU_LOG
                    #循环从11列(进程名)开始打印,如果i等于最后一行,就打印i的列并换行,否则就打印i的列
                    if [[ -n `cat $CPU_LOG` ]]; then
                       echo -e "\033[32m  参考值${i}\033[0m"
                       cat $CPU_LOG
                       > $CPU_LOG
                    else
                        echo "No process using the CPU."
                        break
                    fi
                    i=$(($i+1))
                    sleep 1
                done
                echo "---------------------------------------"
                break
                ;;
            mem_top10)
                #占用内存高的前10个进程
                echo "---------------------------------------"
                MEM_LOG=/tmp/mem_top.tmp
                i=1
                while [[ $i -le 3 ]]; do
                    #ps aux |awk '{if($4>0.1)print "Memory: "$4"% -->",$11,$12,$13,$14,$15,$16,"(PID:"$2")" |"sort -k2 -nr |head -n 10"}' > $MEM_LOG
                    ps aux |awk '{if($4>0.1){{printf "PID: "$2" Memory: "$3"% --> "}for(i=11;i<=NF;i++)if(i==NF)printf $i"\n";else printf $i}}' |sort -k4 -nr |head -10 > $MEM_LOG
                    if [[ -n `cat $MEM_LOG` ]]; then
                        echo -e "\033[32m  参考值${i}\033[0m"
                        cat $MEM_LOG
                        > $MEM_LOG
                    else
                        echo "No process using the Memory."
                        break
                    fi
                    i=$(($i+1))
                    sleep 1
                done
                echo "---------------------------------------"
                break
                ;;
            traffic)
                #查看网络流量
                while true; do
                    read -p "Please enter the network card name(eth[0-9] or em[0-9]): " eth
                    #if [[ $eth =~ ^eth[0-9]$ ]] || [[ $eth =~ ^em[0-9]$ ]] && [[ `ifconfig |grep -c "\<$eth\>"` -eq 1 ]]; then
                    if [ `ifconfig |grep -c "\<$eth\>"` -eq 1 ]; then
                            break
                    else
                        echo "Input format error or Don't have the card name, please input again."
                    fi
                done
                echo "---------------------------------------"
                echo -e " In ------ Out"
                i=1
                while [[ $i -le 3 ]]; do
                    #OLD_IN=`ifconfig $eth |awk '/RX bytes/{print $2}' |cut -d: -f2`
                    #OLD_OUT=`ifconfig $eth |awk '/RX bytes/{print $6}' |cut -d: -f2`
                    OLD_IN=`ifconfig $eth |awk -F'[: ]+' '/bytes/{if(NR==8)print $4;else if(NR==5)print $6}'`
                    #CentOS6和CentOS7 ifconfig输出进出流量信息位置不同,CentOS6中RX与TX行号等于8,CentOS7中RX行号是5,TX行号是5,所以就做了个判断.      
                    OLD_OUT=`ifconfig $eth |awk -F'[: ]+' '/bytes/{if(NR==8)print $9;else if(NR==7)print $6}'`
                    sleep 1
                    NEW_IN=`ifconfig $eth |awk -F'[: ]+' '/bytes/{if(NR==8)print $4;else if(NR==5)print $6}'`
                    NEW_OUT=`ifconfig $eth |awk -F'[: ]+' '/bytes/{if(NR==8)print $9;else if(NR==7)print $6}'`
                    IN=`awk 'BEGIN{printf "%.1f\n",'$((${NEW_IN}-${OLD_IN}))'/1024/128}'`
                    OUT=`awk 'BEGIN{printf "%.1f\n",'$((${NEW_OUT}-${OLD_OUT}))'/1024/128}'`
                    echo "${IN}MB/s ${OUT}MB/s"
                    i=$(($i+1))
                    sleep 1
                done
                echo "---------------------------------------"
                break
                ;;
                        quit)
                                exit 0
                                ;;
               *)
                    echo "---------------------------------------"
                    echo "Please enter the number."
                    echo "---------------------------------------"
                    break
                    ;;
        esac
    done
done

脚本运行

sh show_sys_info.sh

1)cpu_load:分析 cpu

util:cpu 已使用百分比

user use:用户使用 cpu 百分比

system use:系统使用 cpu 百分比

I/O wait:I/O 等待百分比

三次刷新,间隔时间 1s

2)disk_load 分析磁盘

Util:每秒磁盘处理的百分比
vda:
I/O Wait:IO 等待时间
Read/s:每秒磁盘读数据量
vda:
Write/s:每秒磁盘写数据量
vda:

3)disk_use 磁盘使用情况

disk total:打印磁盘数量和磁盘分区使用情况

4)disk_inode 如果有磁盘 inodo 超过 90%,这里会打印出来

5)mem_use 内存占用情况

Total:总内存

use:已使用内存

free:剩余内存

cache:缓存占用内存

6)tcp_status tcp 状态

7)cpu_top10 占用 cpu 最大的十个进程
8)mem_top10 占用内存最大的十个进程
9)traffic 查看网卡进出流量

如果觉得我的文章对您有用,请随意打赏。您的支持将鼓励我继续创作!
共收到 2 条回复 时间 点赞

一条 SAR 走天下

SunOS/HU-UX 这些系统可用吗?

需要 登录 后方可回复, 如果你还没有账号请点击这里 注册