tools/plot/fio2gnuplot.py

   1 #!/usr/bin/python
   2 #
   3 #  Copyright (C) 2013 eNovance SAS <licensing@enovance.com>
   4 #  Author: Erwan Velu  <erwan@enovance.com>
   5 #
   6 #  The license below covers all files distributed with fio unless otherwise
   7 #  noted in the file itself.
   8 #
   9 #  This program is free software; you can redistribute it and/or modify
  10 #  it under the terms of the GNU General Public License version 2 as
  11 #  published by the Free Software Foundation.
  12 #
  13 #  This program is distributed in the hope that it will be useful,
  14 #  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 #  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 #  GNU General Public License for more details.
  17 #
  18 #  You should have received a copy of the GNU General Public License
  19 #  along with this program; if not, write to the Free Software
  20 #  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21
  22 import os
  23 import fnmatch
  24 import sys
  25 import getopt
  26 import re
  27 import math
  28
  29 def find_file(path, pattern):
  30         fio_data_file=[]
  31         # For all the local files
  32         for file in os.listdir(path):
  33             # If the file math the regexp
  34             if fnmatch.fnmatch(file, pattern):
  35                 # Let's consider this file
  36                 fio_data_file.append(file)
  37
  38         return fio_data_file
  39
  40 def generate_gnuplot_script(title,gnuplot_output_filename,mode):
  41         f=open("mygraph",'w')
  42         f.write("call \'graph3D.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\'\n" % (title,gnuplot_output_filename,gnuplot_output_filename,mode))
  43         f.close()
  44
  45 def generate_gnuplot_math_script(title,gnuplot_output_filename,mode,average):
  46         f=open("mymath",'a')
  47         f.write("call \'math.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\' %s\n" % (title,gnuplot_output_filename,gnuplot_output_filename,mode,average))
  48         f.close()
  49
  50 def compute_aggregated_file(fio_data_file, gnuplot_output_filename):
  51         temp_files=[]
  52         pos=0
  53         # Let's create a temporary file for each selected fio file
  54         for file in fio_data_file:
  55                 tmp_filename = "gnuplot_temp_file.%d" % pos
  56                 temp_files.append(open(tmp_filename,'r'))
  57                 pos = pos +1
  58
  59         f = open(gnuplot_output_filename, "w")
  60         index=0
  61         # Let's add some information
  62         for tempfile in temp_files:
  63                     f.write("# Disk%d was coming from %s\n" % (index,fio_data_file[index]))
  64                     f.write(tempfile.read())
  65                     f.write("\n")
  66                     tempfile.close()
  67                     index = index + 1
  68         f.close()
  69
  70 def average(s): return sum(s) * 1.0 / len(s)
  71
  72 def compute_temp_file(fio_data_file,disk_perf):
  73         files=[]
  74         temp_outfile=[]
  75         blk_size=0
  76         for file in fio_data_file:
  77                 files.append(open(file))
  78                 pos = len(files) - 1
  79                 tmp_filename = "gnuplot_temp_file.%d" % pos
  80                 temp_outfile.append(open(tmp_filename,'w'))
  81                 disk_perf.append([])
  82
  83         shall_break = False
  84         while True:
  85                 current_line=[]
  86                 for file in files:
  87                         s=file.readline().replace(',',' ').split()
  88                         if not s:
  89                                 shall_break=True
  90                                 break;
  91                         current_line.append(s);
  92
  93                 if shall_break == True:
  94                         break
  95
  96                 last_time = -1
  97                 index=0
  98                 perfs=[]
  99                 for line in current_line:
 100                         time, perf, x, block_size = line
 101                         if (blk_size == 0):
 102                                 blk_size=int(block_size)
 103
 104                         # We ignore the first 500msec as it doesn't seems to be part of the real benchmark
 105                         # Time < 500 usually reports BW=0 breaking the min computing
 106                         if ((int(time)) > 500):
 107                                 disk_perf[index].append(int(perf))
 108                                 perfs.append(perf)
 109                                 index = index + 1
 110
 111                 # If we reach this point, it means that all the traces are coherent
 112                 for p in enumerate(perfs):
 113                         temp_outfile[p[0]].write("%s %.2f %s\n" % (p[0], float(float(time)/1000), p[1]))
 114
 115         for file in files:
 116                 file.close()
 117         for file in temp_outfile:
 118                 file.close()
 119         return blk_size
 120
 121 def compute_math(fio_data_file, title,gnuplot_output_filename,mode,disk_perf):
 122         global_min=[]
 123         global_max=[]
 124         average_file=open(gnuplot_output_filename+'.average', 'w')
 125         min_file=open(gnuplot_output_filename+'.min', 'w')
 126         max_file=open(gnuplot_output_filename+'.max', 'w')
 127         stddev_file=open(gnuplot_output_filename+'.stddev', 'w')
 128         global_file=open(gnuplot_output_filename+'.global','w')
 129
 130         min_file.write('DiskName %s\n' % mode)
 131         max_file.write('DiskName %s\n'% mode)
 132         average_file.write('DiskName %s\n'% mode)
 133         stddev_file.write('DiskName %s\n'% mode )
 134         for disk in xrange(len(fio_data_file)):
 135 #               print disk_perf[disk]
 136                 min_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 137                 max_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 138                 average_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 139                 stddev_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 140                 avg  = average(disk_perf[disk])
 141                 variance = map(lambda x: (x - avg)**2, disk_perf[disk])
 142                 standard_deviation = math.sqrt(average(variance))
 143 #               print "Disk%d [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (disk,min(disk_perf[disk]),max(disk_perf[disk]),avg, standard_deviation)
 144                 average_file.write('%d %d\n' % (disk, avg))
 145                 stddev_file.write('%d %d\n' % (disk, standard_deviation))
 146                 local_min=min(disk_perf[disk])
 147                 local_max=max(disk_perf[disk])
 148                 min_file.write('%d %d\n' % (disk, local_min))
 149                 max_file.write('%d %d\n' % (disk, local_max))
 150                 global_min.append(int(local_min))
 151                 global_max.append(int(local_max))
 152
 153         global_disk_perf = sum(disk_perf, [])
 154         avg  = average(global_disk_perf)
 155         variance = map(lambda x: (x - avg)**2, global_disk_perf)
 156         standard_deviation = math.sqrt(average(variance))
 157
 158         global_file.write('min=%.2f\n' % min(global_disk_perf))
 159         global_file.write('max=%.2f\n' % max(global_disk_perf))
 160         global_file.write('avg=%.2f\n' % avg)
 161         global_file.write('stddev=%.2f\n' % standard_deviation)
 162         global_file.write('values_count=%d\n' % len(global_disk_perf))
 163         global_file.write('disks_count=%d\n' % len(fio_data_file))
 164         #print "Global [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (min(global_disk_perf),max(global_disk_perf),avg, standard_deviation)
 165
 166         average_file.close()
 167         min_file.close()
 168         max_file.close()
 169         stddev_file.close()
 170         global_file.close()
 171         try:
 172                 os.remove('mymath')
 173         except:
 174                 True
 175
 176         generate_gnuplot_math_script("Average values of "+title,gnuplot_output_filename+'.average',mode,int(avg))
 177         generate_gnuplot_math_script("Min values of "+title,gnuplot_output_filename+'.min',mode,average(global_min))
 178         generate_gnuplot_math_script("Max values of "+title,gnuplot_output_filename+'.max',mode,average(global_max))
 179         generate_gnuplot_math_script("Standard Deviation of "+title,gnuplot_output_filename+'.stddev',mode,int(standard_deviation))
 180
 181 def parse_global_files(fio_data_file, global_search):
 182         max_result=0
 183         max_file=''
 184         for file in fio_data_file:
 185                 f=open(file)
 186                 disk_count=0
 187                 search_value=-1
 188
 189                 # Let's read the complete file
 190                 while True:
 191                         try:
 192                                 # We do split the name from the value
 193                                 name,value=f.readline().split("=")
 194                         except:
 195                                 f.close()
 196                                 break
 197                         # If we ended the file
 198                         if not name:
 199                                 # Let's process what we have
 200                                 f.close()
 201                                 break
 202                         else:
 203                                 # disks_count is not global_search item
 204                                 # As we need it for some computation, let's save it
 205                                 if name=="disks_count":
 206                                         disks_count=int(value)
 207
 208                                 # Let's catch the searched item
 209                                 if global_search in name:
 210                                         search_value=float(value)
 211
 212                 # Let's process the avg value by estimated the global bandwidth per file
 213                 # We keep the biggest in memory for reporting
 214                 if global_search == "avg":
 215                         if (disks_count > 0) and (search_value != -1):
 216                                 result=disks_count*search_value
 217                                 if (result > max_result):
 218                                         max_result=result
 219                                         max_file=file
 220         # Let's print the avg output
 221         if global_search == "avg":
 222                 print "Biggest aggregated value of %s was %2.f in file %s\n" % (global_search, max_result, max_file)
 223         else:
 224                 print "Global search %s is not yet implemented\n" % global_search
 225
 226 def render_gnuplot():
 227         print "Running gnuplot Rendering\n"
 228         try:
 229                 os.system("gnuplot mymath")
 230                 os.system("gnuplot mygraph")
 231         except:
 232                 print "Could not run gnuplot on mymath or mygraph !\n"
 233                 sys.exit(1);
 234
 235 def print_help():
 236     print 'fio2gnuplot.py -ghbio -t <title> -o <outputfile> -p <pattern>'
 237     print
 238     print '-h --help                           : Print this help'
 239     print '-p <pattern> or --pattern <pattern> : A pattern in regexp to select fio input files'
 240     print '-b           or --bandwidth         : A predefined pattern for selecting *_bw.log files'
 241     print '-i           or --iops              : A predefined pattern for selecting *_iops.log files'
 242     print '-g           or --gnuplot           : Render gnuplot traces before exiting'
 243     print '-o           or --outputfile <file> : The basename for gnuplot traces'
 244     print '                                       - Basename is set with the pattern if defined'
 245     print '-t           or --title <title>     : The title of the gnuplot traces'
 246     print '                                       - Title is set with the block size detected in fio traces'
 247     print '-G           or --Global <type>     : Search for <type> in .global files match by a pattern'
 248     print '                                       - Available types are : min, max, avg, stddev'
 249     print '                                       - The .global extension is added automatically to the pattern'
 250
 251 def main(argv):
 252     mode='unknown'
 253     pattern=''
 254     pattern_set_by_user=False
 255     title='No title'
 256     gnuplot_output_filename='result'
 257     disk_perf=[]
 258     run_gnuplot=False
 259     parse_global=False
 260     global_search=''
 261
 262     try:
 263             opts, args = getopt.getopt(argv[1:],"ghbio:t:p:G:")
 264     except getopt.GetoptError:
 265          print_help()
 266          sys.exit(2)
 267
 268     for opt, arg in opts:
 269       if opt in ("-b", "--bandwidth"):
 270          pattern='*_bw.log'
 271       elif opt in ("-i", "--iops"):
 272          pattern='*_iops.log'
 273       elif opt in ("-p", "--pattern"):
 274          pattern_set_by_user=True
 275          pattern=arg
 276          pattern=pattern.replace('\\','')
 277       elif opt in ("-o", "--outputfile"):
 278          gnuplot_output_filename=arg
 279       elif opt in ("-t", "--title"):
 280          title=arg
 281       elif opt in ("-g", "--gnuplot"):
 282          run_gnuplot=True
 283       elif opt in ("-G", "--Global"):
 284          parse_global=True
 285          global_search=arg
 286       elif opt in ("-h", "--help"):
 287           print_help()
 288           sys.exit(1)
 289
 290     # Adding .global extension to the file
 291     if parse_global==True:
 292             if not gnuplot_output_filename.endswith('.global'):
 293                 pattern = pattern+'.global'
 294
 295     fio_data_file=find_file('.',pattern)
 296     if len(fio_data_file) == 0:
 297             print "No log file found with pattern %s!" % pattern
 298             sys.exit(1)
 299
 300     fio_data_file=sorted(fio_data_file, key=str.lower)
 301     for file in fio_data_file:
 302         print 'Selected %s' % file
 303         if "_bw.log" in file :
 304                 mode="Bandwidth (KB/sec)"
 305         if "_iops.log" in file :
 306                 mode="IO per Seconds (IO/sec)"
 307     if (title == 'No title') and (mode != 'unknown'):
 308             if "Bandwidth" in mode:
 309                     title='Bandwidth benchmark with %d fio results' % len(fio_data_file)
 310             if "IO" in mode:
 311                     title='IO benchmark with %d fio results' % len(fio_data_file)
 312
 313     #We need to adjust the output filename regarding the pattern required by the user
 314     if (pattern_set_by_user == True):
 315             gnuplot_output_filename=pattern
 316             # As we do have some regexp in the pattern, let's make this simpliest
 317             # We do remove the simpliest parts of the expression to get a clear file name
 318             gnuplot_output_filename=gnuplot_output_filename.replace('-*-','-')
 319             gnuplot_output_filename=gnuplot_output_filename.replace('*','-')
 320             gnuplot_output_filename=gnuplot_output_filename.replace('--','-')
 321             gnuplot_output_filename=gnuplot_output_filename.replace('.log','')
 322             # Insure that we don't have any starting or trailing dash to the filename
 323             gnuplot_output_filename = gnuplot_output_filename[:-1] if gnuplot_output_filename.endswith('-') else gnuplot_output_filename
 324             gnuplot_output_filename = gnuplot_output_filename[1:] if gnuplot_output_filename.startswith('-') else gnuplot_output_filename
 325
 326     if parse_global==True:
 327         parse_global_files(fio_data_file, global_search)
 328     else:
 329         blk_size=compute_temp_file(fio_data_file,disk_perf)
 330         title="%s @ Blocksize = %dK" % (title,blk_size/1024)
 331         compute_aggregated_file(fio_data_file, gnuplot_output_filename)
 332         compute_math(fio_data_file,title,gnuplot_output_filename,mode,disk_perf)
 333         generate_gnuplot_script(title,gnuplot_output_filename,mode)
 334
 335         if (run_gnuplot==True):
 336                 render_gnuplot()
 337
 338     # Cleaning temporary files
 339     try:
 340         os.remove('gnuplot_temp_file.*')
 341     except:
 342         True
 343
 344 #Main
 345 if __name__ == "__main__":
 346     sys.exit(main(sys.argv))