shell和Python 两种方法分别画 iostat的监控图-Toy模板网

这篇具有很好参考价值的文章主要介绍了shell和Python 两种方法分别画 iostat的监控图。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

在服务器存储的测试中,经常需要看performance的性能曲线，这样最能直接观察HDD或者SSD的性能曲线。
如下这是一个针对HDD跑Fio读写的iostat监控log,下面介绍一下分别用shell 和Python3 写画iostat图的方法
shell和Python 两种方法分别画 iostat的监控图,python,linux,matplotlib,服务器,bash
1 shell脚本
环境:linux OS gnuplot工具
第一步 :解析iostat log。提取要作图的一行

#!/bin/bash
hdd=$1
cat ${hdd}_iostat.log  |grep -i Device |head -1  >${hdd}_iostat.txt
cat ${hdd}_iostat.log  |grep -i ${hdd}  >>${hdd}_iostat.txt

第二步:把画图的列(监控读写的IOPS)求出来

#!/bin/bash
hdd=$1
cat ${hdd}_iostat.log  |grep -i Device |head -1  >${hdd}_iostat.txt
cat ${hdd}_iostat.log  |grep -i ${hdd}  >>${hdd}_iostat.txt
num_read_iops=`head -1 ${hdd}_iostat.txt | awk  '{for (i=1;i<=NF;i++) {if ($i=="r/s") {print i}}}'`
num_write_iops=`head -1 ${hdd}_iostat.txt | awk  '{for (i=1;i<=NF;i++) {if ($i=="w/s") {print i}}}'`

第三步:直接调用gnuplot工具作图

#!/bin/bash
hdd=$1
cat ${hdd}_iostat.log  |grep -i Device |head -1  >${hdd}_iostat.txt
cat ${hdd}_iostat.log  |grep -i ${hdd}  >>${hdd}_iostat.txt
num_read_iops=`head -1 ${hdd}_iostat.txt | awk  '{for (i=1;i<=NF;i++) {if ($i=="r/s") {print i}}}'`
num_write_iops=`head -1 ${hdd}_iostat.txt | awk  '{for (i=1;i<=NF;i++) {if ($i=="w/s") {print i}}}'`
echo "
      	set terminal png 
      	set title 'Random_${hdd}_IOPS'
      	set output 'IOPS_${hdd}.png'
      	set xlabel 'count:1s'
      	set ylabel  'IOPS'
        set key right top
      	plot '${hdd}_iostat.txt' using :$num_write_iops  title 'w/s' with lines lw 1,\
       '${hdd}_iostat.txt' using :$num_read_iops title 'r/s' with lines lw 1
      	" | gnuplot

运行 :bash gnuplot.sh sde
shell和Python 两种方法分别画 iostat的监控图,python,linux,matplotlib,服务器,bash
2 python 脚本
环境:python3
python库: sys(引用参数), re(正则匹配), pandas(做表格), matplotlib(画图工具)
第一步:解析iostat log

import sys
import re
import pandas as pd
from matplotlib import pyplot as plt
hdd = sys.argv[1]
rs = []
ws = []
logfile = open(f"{hdd}_iostat.log","r")

for log in logfile:
    if hdd in log:
        data = re.sub(' +', ' ', log)
        rs.append(data.split(" ")[3])
        ws.append(data.split(" ")[4])

第二步:将需要作图的列(这次监控带宽)导入CSV

import sys
import re
import pandas as pd
from matplotlib import pyplot as plt
hdd = sys.argv[1]
rs = []
ws = []
logfile = open(f"{hdd}_iostat.log","r")

for log in logfile:
    if hdd in log:
        data = re.sub(' +', ' ', log)
        rs.append(data.split(" ")[3])
        ws.append(data.split(" ")[4])

dic = {"read": rs, "write": ws}
df = pd.DataFrame(dic)
df.to_csv(f"{hdd}_iostat.csv")
data = pd.read_csv(f"{hdd}_iostat.csv")

第三步:用matplotlib库直接作图

import sys
import re
import pandas as pd
from matplotlib import pyplot as plt
hdd = sys.argv[1]
rs = []
ws = []
logfile = open(f"{hdd}_iostat.log","r")

for log in logfile:
    if hdd in log:
        data = re.sub(' +', ' ', log)
        rs.append(data.split(" ")[3])
        ws.append(data.split(" ")[4])

dic = {"read": rs, "write": ws}
df = pd.DataFrame(dic)
df.to_csv(f"{hdd}_iostat.csv")
data = pd.read_csv(f"{hdd}_iostat.csv")

fig = plt.figure(figsize=(10, 6), dpi=300)
x = data.iloc[:, 0]
y1 = data.iloc[:, 1]
y2 = data.iloc[:, 2]
plt.xlabel(u'time (s)', size=10)
plt.ylabel("BW(kB/s)")
plt.plot(x, y1, label='rKB/s', color='orange', linestyle=':')
plt.plot(x, y2, label='wKB/s', color='cyan', linestyle='-.')
plt.legend(loc="upper right")
plt.title(f"{hdd}_BW")
plt.grid(alpha=0.4)
plt.savefig(str(hdd) + ".png")

运行 :python3 plot.py sde
shell和Python 两种方法分别画 iostat的监控图,python,linux,matplotlib,服务器,bash
这个时候我们发现matplotlib默认图像不是从0开始的,加入以下两行代码
plt.xlim(xmin=0)
plt.ylim(ymin=0)
脚本如下：

#!/bin/usr/python3
#autor xuetao 
import os
import sys
import re
import pandas as pd
from matplotlib import pyplot as plt
hdd = sys.argv[1]
rs = []
ws = []
logfile = open(f"{hdd}_iostat.log","r")

for log in logfile:
    if hdd in log:
        data = re.sub(' +', ' ', log)
        rs.append(data.split(" ")[3])
        ws.append(data.split(" ")[4])

dic = {"read": rs, "write": ws}
df = pd.DataFrame(dic)
df.to_csv(f"{hdd}_iostat.csv")

data = pd.read_csv(f"{hdd}_iostat.csv")
fig = plt.figure(figsize=(10, 6), dpi=300)
x = data.iloc[:, 0]
y1 = data.iloc[:, 1]
y2 = data.iloc[:, 2]
plt.xlabel(u'time (s)', size=10)
plt.plot(x, y1, label='rKB/s', color='orange', linestyle=':')
plt.plot(x, y2, label='wKB/s', color='cyan', linestyle='-.')
plt.legend(loc="upper right")
plt.title(f"{hdd}_BW")
plt.grid(alpha=0.4)
plt.xlim(xmin=0)
plt.ylim(ymin=0)
plt.savefig(str(hdd) + ".png")