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(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir):
  41         filename=gnuplot_output_dir+'mygraph'
  42         f=open(filename,'w')
  43         if len(fio_data_file) > 1:
  44                 f.write("call \'%s/graph3D.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\'\n" % (gpm_dir,title,gnuplot_output_filename,gnuplot_output_filename,mode))
  45
  46         pos=0
  47         # Let's create a temporary file for each selected fio file
  48         for file in fio_data_file:
  49                 tmp_filename = "gnuplot_temp_file.%d" % pos
  50                 png_file=file.replace('.log','')
  51                 raw_filename = "%s-2Draw" % (png_file)
  52                 smooth_filename = "%s-2Dsmooth" % (png_file)
  53                 trend_filename = "%s-2Dtrend" % (png_file)
  54                 avg  = average(disk_perf[pos])
  55                 f.write("call \'%s/graph2D.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\' \'%s\' \'%s\' \'%f\'\n" % (gpm_dir,title,tmp_filename,raw_filename,mode,smooth_filename,trend_filename,avg))
  56                 pos = pos +1
  57
  58         f.close()
  59
  60 def generate_gnuplot_math_script(title,gnuplot_output_filename,mode,average,gnuplot_output_dir,gpm_dir):
  61         filename=gnuplot_output_dir+'mymath';
  62         f=open(filename,'a')
  63         f.write("call \'%s/math.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\' %s\n" % (gpm_dir,title,gnuplot_output_filename,gnuplot_output_filename,mode,average))
  64         f.close()
  65
  66 def compute_aggregated_file(fio_data_file, gnuplot_output_filename, gnuplot_output_dir):
  67         temp_files=[]
  68         pos=0
  69
  70         # Let's create a temporary file for each selected fio file
  71         for file in fio_data_file:
  72                 tmp_filename = "%sgnuplot_temp_file.%d" % (gnuplot_output_dir, pos)
  73                 temp_files.append(open(tmp_filename,'r'))
  74                 pos = pos +1
  75
  76         f = open(gnuplot_output_dir+gnuplot_output_filename, "w")
  77         index=0
  78         # Let's add some information
  79         for tempfile in temp_files:
  80                     f.write("# Disk%d was coming from %s\n" % (index,fio_data_file[index]))
  81                     f.write(tempfile.read())
  82                     f.write("\n")
  83                     tempfile.close()
  84                     index = index + 1
  85         f.close()
  86
  87 def average(s): return sum(s) * 1.0 / len(s)
  88
  89 def compute_temp_file(fio_data_file,disk_perf,gnuplot_output_dir):
  90         files=[]
  91         temp_outfile=[]
  92         blk_size=0
  93         for file in fio_data_file:
  94                 files.append(open(file))
  95                 pos = len(files) - 1
  96                 tmp_filename = "%sgnuplot_temp_file.%d" % (gnuplot_output_dir,pos)
  97                 gnuplot_file=open(tmp_filename,'w')
  98                 temp_outfile.append(gnuplot_file)
  99                 gnuplot_file.write("#Temporary file based on file %s\n" % file)
 100                 disk_perf.append([])
 101
 102         shall_break = False
 103         while True:
 104                 current_line=[]
 105                 nb_empty_files=0
 106                 nb_files=len(files)
 107                 for file in files:
 108                         s=file.readline().replace(',',' ').split()
 109                         if not s:
 110                                 nb_empty_files+=1
 111                                 s="-1, 0, 0, 0'".replace(',',' ').split()
 112
 113                         if (nb_empty_files == nb_files):
 114                                 shall_break=True
 115                                 break;
 116
 117                         current_line.append(s);
 118
 119                 if shall_break == True:
 120                         break
 121
 122                 last_time = -1
 123                 index=0
 124                 perfs=[]
 125                 for line in current_line:
 126                         time, perf, x, block_size = line
 127                         if (blk_size == 0):
 128                                 blk_size=int(block_size)
 129
 130                         # We ignore the first 500msec as it doesn't seems to be part of the real benchmark
 131                         # Time < 500 usually reports BW=0 breaking the min computing
 132                         if (((int(time)) > 500) or (int(time)==-1)):
 133                                 disk_perf[index].append(int(perf))
 134                                 perfs.append("%s %s"% (time, perf))
 135                                 index = index + 1
 136
 137                 # If we reach this point, it means that all the traces are coherent
 138                 for p in enumerate(perfs):
 139                         perf_time,perf = p[1].split()
 140                         if (perf_time != "-1"):
 141                                 temp_outfile[p[0]].write("%s %.2f %s\n" % (p[0], float(float(perf_time)/1000), perf))
 142
 143
 144         for file in files:
 145                 file.close()
 146         for file in temp_outfile:
 147                 file.close()
 148         return blk_size
 149
 150 def compute_math(fio_data_file, title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir):
 151         global_min=[]
 152         global_max=[]
 153         average_file=open(gnuplot_output_dir+gnuplot_output_filename+'.average', 'w')
 154         min_file=open(gnuplot_output_dir+gnuplot_output_filename+'.min', 'w')
 155         max_file=open(gnuplot_output_dir+gnuplot_output_filename+'.max', 'w')
 156         stddev_file=open(gnuplot_output_dir+gnuplot_output_filename+'.stddev', 'w')
 157         global_file=open(gnuplot_output_dir+gnuplot_output_filename+'.global','w')
 158
 159         min_file.write('DiskName %s\n' % mode)
 160         max_file.write('DiskName %s\n'% mode)
 161         average_file.write('DiskName %s\n'% mode)
 162         stddev_file.write('DiskName %s\n'% mode )
 163         for disk in xrange(len(fio_data_file)):
 164 #               print disk_perf[disk]
 165                 min_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 166                 max_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 167                 average_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 168                 stddev_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 169                 avg  = average(disk_perf[disk])
 170                 variance = map(lambda x: (x - avg)**2, disk_perf[disk])
 171                 standard_deviation = math.sqrt(average(variance))
 172 #               print "Disk%d [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (disk,min(disk_perf[disk]),max(disk_perf[disk]),avg, standard_deviation)
 173                 average_file.write('%d %d\n' % (disk, avg))
 174                 stddev_file.write('%d %d\n' % (disk, standard_deviation))
 175                 local_min=min(disk_perf[disk])
 176                 local_max=max(disk_perf[disk])
 177                 min_file.write('%d %d\n' % (disk, local_min))
 178                 max_file.write('%d %d\n' % (disk, local_max))
 179                 global_min.append(int(local_min))
 180                 global_max.append(int(local_max))
 181
 182         global_disk_perf = sum(disk_perf, [])
 183         avg  = average(global_disk_perf)
 184         variance = map(lambda x: (x - avg)**2, global_disk_perf)
 185         standard_deviation = math.sqrt(average(variance))
 186
 187         global_file.write('min=%.2f\n' % min(global_disk_perf))
 188         global_file.write('max=%.2f\n' % max(global_disk_perf))
 189         global_file.write('avg=%.2f\n' % avg)
 190         global_file.write('stddev=%.2f\n' % standard_deviation)
 191         global_file.write('values_count=%d\n' % len(global_disk_perf))
 192         global_file.write('disks_count=%d\n' % len(fio_data_file))
 193         #print "Global [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (min(global_disk_perf),max(global_disk_perf),avg, standard_deviation)
 194
 195         average_file.close()
 196         min_file.close()
 197         max_file.close()
 198         stddev_file.close()
 199         global_file.close()
 200         try:
 201                 os.remove(gnuplot_output_dir+'mymath')
 202         except:
 203                 True
 204
 205         generate_gnuplot_math_script("Average values of "+title,gnuplot_output_filename+'.average',mode,int(avg),gnuplot_output_dir,gpm_dir)
 206         generate_gnuplot_math_script("Min values of "+title,gnuplot_output_filename+'.min',mode,average(global_min),gnuplot_output_dir,gpm_dir)
 207         generate_gnuplot_math_script("Max values of "+title,gnuplot_output_filename+'.max',mode,average(global_max),gnuplot_output_dir,gpm_dir)
 208         generate_gnuplot_math_script("Standard Deviation of "+title,gnuplot_output_filename+'.stddev',mode,int(standard_deviation),gnuplot_output_dir,gpm_dir)
 209
 210 def parse_global_files(fio_data_file, global_search):
 211         max_result=0
 212         max_file=''
 213         for file in fio_data_file:
 214                 f=open(file)
 215                 disk_count=0
 216                 search_value=-1
 217
 218                 # Let's read the complete file
 219                 while True:
 220                         try:
 221                                 # We do split the name from the value
 222                                 name,value=f.readline().split("=")
 223                         except:
 224                                 f.close()
 225                                 break
 226                         # If we ended the file
 227                         if not name:
 228                                 # Let's process what we have
 229                                 f.close()
 230                                 break
 231                         else:
 232                                 # disks_count is not global_search item
 233                                 # As we need it for some computation, let's save it
 234                                 if name=="disks_count":
 235                                         disks_count=int(value)
 236
 237                                 # Let's catch the searched item
 238                                 if global_search in name:
 239                                         search_value=float(value)
 240
 241                 # Let's process the avg value by estimated the global bandwidth per file
 242                 # We keep the biggest in memory for reporting
 243                 if global_search == "avg":
 244                         if (disks_count > 0) and (search_value != -1):
 245                                 result=disks_count*search_value
 246                                 if (result > max_result):
 247                                         max_result=result
 248                                         max_file=file
 249         # Let's print the avg output
 250         if global_search == "avg":
 251                 print "Biggest aggregated value of %s was %2.f in file %s\n" % (global_search, max_result, max_file)
 252         else:
 253                 print "Global search %s is not yet implemented\n" % global_search
 254
 255 def render_gnuplot(gnuplot_output_dir):
 256         print "Running gnuplot Rendering\n"
 257         try:
 258                 os.system("cd %s; gnuplot mymath" % gnuplot_output_dir)
 259                 os.system("cd %s; gnuplot mygraph" % gnuplot_output_dir)
 260         except:
 261                 print "Could not run gnuplot on mymath or mygraph !\n"
 262                 sys.exit(1);
 263
 264 def print_help():
 265     print 'fio2gnuplot.py -ghbio -t <title> -o <outputfile> -p <pattern>'
 266     print
 267     print '-h --help                           : Print this help'
 268     print '-p <pattern> or --pattern <pattern> : A pattern in regexp to select fio input files'
 269     print '-b           or --bandwidth         : A predefined pattern for selecting *_bw.log files'
 270     print '-i           or --iops              : A predefined pattern for selecting *_iops.log files'
 271     print '-g           or --gnuplot           : Render gnuplot traces before exiting'
 272     print '-o           or --outputfile <file> : The basename for gnuplot traces'
 273     print '                                       - Basename is set with the pattern if defined'
 274     print '-d           or --outputdir <dir>   : The directory where gnuplot shall render files'
 275     print '-t           or --title <title>     : The title of the gnuplot traces'
 276     print '                                       - Title is set with the block size detected in fio traces'
 277     print '-G           or --Global <type>     : Search for <type> in .global files match by a pattern'
 278     print '                                       - Available types are : min, max, avg, stddev'
 279     print '                                       - The .global extension is added automatically to the pattern'
 280
 281 def main(argv):
 282     mode='unknown'
 283     pattern=''
 284     pattern_set_by_user=False
 285     title='No title'
 286     gnuplot_output_filename='result'
 287     gnuplot_output_dir='./'
 288     gpm_dir="/usr/share/fio/"
 289     disk_perf=[]
 290     run_gnuplot=False
 291     parse_global=False
 292     global_search=''
 293
 294     if not os.path.isfile(gpm_dir+'math.gpm'):
 295             gpm_dir="/usr/local/share/fio/"
 296             if not os.path.isfile(gpm_dir+'math.gpm'):
 297                     print "Looks like fio didn't got installed properly as no gpm files found in '/usr/share/fio' or '/usr/local/share/fio'\n"
 298                     sys.exit(3)
 299
 300     try:
 301             opts, args = getopt.getopt(argv[1:],"ghbio:d:t:p:G:")
 302     except getopt.GetoptError:
 303          print_help()
 304          sys.exit(2)
 305
 306     for opt, arg in opts:
 307       if opt in ("-b", "--bandwidth"):
 308          pattern='*_bw.log'
 309       elif opt in ("-i", "--iops"):
 310          pattern='*_iops.log'
 311       elif opt in ("-p", "--pattern"):
 312          pattern_set_by_user=True
 313          pattern=arg
 314          pattern=pattern.replace('\\','')
 315       elif opt in ("-o", "--outputfile"):
 316          gnuplot_output_filename=arg
 317       elif opt in ("-d", "--outputdir"):
 318          gnuplot_output_dir=arg
 319          if not gnuplot_output_dir.endswith('/'):
 320                 gnuplot_output_dir=gnuplot_output_dir+'/'
 321          if not os.path.exists(gnuplot_output_dir):
 322                 os.makedirs(gnuplot_output_dir)
 323       elif opt in ("-t", "--title"):
 324          title=arg
 325       elif opt in ("-g", "--gnuplot"):
 326          run_gnuplot=True
 327       elif opt in ("-G", "--Global"):
 328          parse_global=True
 329          global_search=arg
 330       elif opt in ("-h", "--help"):
 331           print_help()
 332           sys.exit(1)
 333
 334     # Adding .global extension to the file
 335     if parse_global==True:
 336             if not gnuplot_output_filename.endswith('.global'):
 337                 pattern = pattern+'.global'
 338
 339     fio_data_file=find_file('.',pattern)
 340     if len(fio_data_file) == 0:
 341             print "No log file found with pattern %s!" % pattern
 342             sys.exit(1)
 343     else:
 344             print "%d files Selected with pattern '%s'" % (len(fio_data_file), pattern)
 345
 346     fio_data_file=sorted(fio_data_file, key=str.lower)
 347     for file in fio_data_file:
 348         print ' |-> %s' % file
 349         if "_bw.log" in file :
 350                 mode="Bandwidth (KB/sec)"
 351         if "_iops.log" in file :
 352                 mode="IO per Seconds (IO/sec)"
 353     if (title == 'No title') and (mode != 'unknown'):
 354             if "Bandwidth" in mode:
 355                     title='Bandwidth benchmark with %d fio results' % len(fio_data_file)
 356             if "IO" in mode:
 357                     title='IO benchmark with %d fio results' % len(fio_data_file)
 358
 359     print
 360     #We need to adjust the output filename regarding the pattern required by the user
 361     if (pattern_set_by_user == True):
 362             gnuplot_output_filename=pattern
 363             # As we do have some regexp in the pattern, let's make this simpliest
 364             # We do remove the simpliest parts of the expression to get a clear file name
 365             gnuplot_output_filename=gnuplot_output_filename.replace('-*-','-')
 366             gnuplot_output_filename=gnuplot_output_filename.replace('*','-')
 367             gnuplot_output_filename=gnuplot_output_filename.replace('--','-')
 368             gnuplot_output_filename=gnuplot_output_filename.replace('.log','')
 369             # Insure that we don't have any starting or trailing dash to the filename
 370             gnuplot_output_filename = gnuplot_output_filename[:-1] if gnuplot_output_filename.endswith('-') else gnuplot_output_filename
 371             gnuplot_output_filename = gnuplot_output_filename[1:] if gnuplot_output_filename.startswith('-') else gnuplot_output_filename
 372
 373     if parse_global==True:
 374         parse_global_files(fio_data_file, global_search)
 375     else:
 376         blk_size=compute_temp_file(fio_data_file,disk_perf,gnuplot_output_dir)
 377         title="%s @ Blocksize = %dK" % (title,blk_size/1024)
 378         compute_aggregated_file(fio_data_file, gnuplot_output_filename, gnuplot_output_dir)
 379         compute_math(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 380         generate_gnuplot_script(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 381
 382         if (run_gnuplot==True):
 383                 render_gnuplot(gnuplot_output_dir)
 384
 385     # Cleaning temporary files
 386     try:
 387         os.remove('gnuplot_temp_file.*')
 388     except:
 389         True
 390
 391 #Main
 392 if __name__ == "__main__":
 393     sys.exit(main(sys.argv))