Tomcat日志分析

#!/bin/bash
#Shell      statistics.sh
#Author 　　 lipc
#Date     　20200818
#Ps    　　 tomcat接入日志分析,需要tomcat日志开启访问时间记录

time1=$(date "+%Y-%m-%d %H:%M:%S")
time2=$(date "+%Y-%m-%d-%H")
echo "请输入日志路径,例如:/tmp/localhost_access_log.2020-12-19.txt"
read name
echo "请输入日志存放路径,例如:/tmp/test/"
read dir
echo "请输入模块名称,例如:pbs"
read module
log="$dir""$module"$time2.log
statistics="$dir""$module"statistics$time2.log
IFS=$'
'

function check() {
    if [ -f $name ]; then
        echo $name'文件目录检测成功'
        echo ""
        echo "程序开始时间："$time1 >>$statistics
        echo "正在收集URL接口信息，预计一分钟，请等待......"
        meta=$(cat $name | awk '{print $7}' | awk -F 'sig' '{print $1}' | sort | uniq | grep -v -w '/' | grep -v 400)
        amount=$(echo "$meta" | wc -l)
        echo "URL接口信息收集完成,即将开始下一步操作......"
        echo ""
    else
        echo $name'文件目录检测失败，请重新检查文件路径'
        exit 1
    fi
}

function metadata() {
    echo "正在收集日志元数据，用于后续统计分析操作,日志存放于$log......"
    i=0
    for url in $meta; do
        w=$(expr $i * 100 / $amount)
        echo "当前进度$w%"
        for time in $(seq -w 07 16); do
            num=$(cat $name | grep 2020:$time | grep $url | wc -l)
            echo $num $url 2020:$time >>$log
        done
        i=$(expr $i + 1)
    done
    echo "日志元数据收集完毕，存放于$log"
    echo ""
}

function concurrent() {
    echo "正在分析日志元数据，用于获取接口并发信息,日志存放于$statistics......"
    echo "每个接口的5个最高并发时段" >>$statistics
    for url in $meta; do
        cat $log | grep $url | sort -nr | head -n 5 >>$statistics
        echo "" >>$statistics
    done

    echo "全天最高并发的5个接口" >>$statistics
    for url in $meta; do
        a=$(cat $name | grep $url | wc -l)
        echo $a $url >>"$dir"a.log
    done
    cat "$dir"a.log | sort -nr | head -n 5 >>$statistics
}

function speed() {
    echo "正在分析接口耗时信息......"
    echo "" >>$statistics
    echo "全天10个最高耗时请求:" >>$statistics
    echo "" >>$statistics
    for url in $meta; do
        cat $name | grep $url | sort -nr -k 11 | head -n 1 >>"$dir"b.log
    done

    cat "$dir"b.log | sort -nr -k 11 | head -n 10 >>$statistics
    echo " " >>$statistics
    meta1=$(cat "$dir"b.log | sort -nr -k 11 | head -n 10 | awk '{print $7}' | awk -F 'sig' '{print $1}' | grep -v -w '/' | grep -v 400)

    echo "分析每个接口3个最高耗请求和3个最低耗请求，及其时间:" >>$statistics
    echo "" >>$statistics
    for url in $meta1; do
        echo "$url接口3个最高耗请求，及其时间:" >>$statistics
        cat $name | grep $url | sort -nr -k 11 | head -n 3 >>$statistics
        echo "$url接口3个最低耗请求，及其时间:" >>$statistics
        cat $name | grep $url | sort -nr -k 11 | tail -n 3 >>$statistics
        num2=$(cat $name | grep $url | awk '{print $11}' | awk '{sum+=$1} END {print "Average = ", sum/NR}')
        echo "$url接口平均耗时:"$num2 ms >>$statistics
        echo " " >>$statistics
    done
    echo "接口耗时信息分析已结束......"
}

function peak() {
    echo "正在分析接口高峰时段信息......"
    echo "本次日志分析，接口高峰时段时间" >>$statistics
    cat $statistics | awk '{print $3}' | grep 2020 | sort | uniq -c | sort -nr | head -n 6 >>$statistics
}

check
metadata
concurrent
speed

time3=$(date "+%Y-%m-%d %H:%M:%S")
echo "程序结束时间："$time3 >>$statistics
echo "本次分析结束"