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 import shutil
  29
  30 def find_file(path, pattern):
  31         fio_data_file=[]
  32         # For all the local files
  33         for file in os.listdir(path):
  34             # If the file math the regexp
  35             if fnmatch.fnmatch(file, pattern):
  36                 # Let's consider this file
  37                 fio_data_file.append(file)
  38
  39         return fio_data_file
  40
  41 def generate_gnuplot_script(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir):
  42         print "Generating rendering scripts"
  43         filename=gnuplot_output_dir+'mygraph'
  44         f=open(filename,'w')
  45
  46         # Plotting 3D or comparing graphs doesn't have a meaning unless if there is at least 2 traces
  47         if len(fio_data_file) > 1:
  48                 f.write("call \'%s/graph3D.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\'\n" % (gpm_dir,title,gnuplot_output_filename,gnuplot_output_filename,mode))
  49
  50                 # Setting up the compare files that will be plot later
  51                 compare=open(gnuplot_output_dir + 'compare.gnuplot','w')
  52                 compare.write('''
  53 set title '%s'
  54 set terminal png size 1280,1024
  55 set ytics axis out auto
  56 set key top left reverse
  57 set xlabel "Time (Seconds)"
  58 set ylabel '%s'
  59 set yrange [0:]
  60 '''% (title,mode))
  61                 compare.close()
  62                 #Copying the common file for all kind of graph (raw/smooth/trend)
  63                 compare_raw_filename="compare-%s-2Draw" % (gnuplot_output_filename)
  64                 compare_smooth_filename="compare-%s-2Dsmooth" % (gnuplot_output_filename)
  65                 compare_trend_filename="compare-%s-2Dtrend" % (gnuplot_output_filename)
  66                 shutil.copy(gnuplot_output_dir+'compare.gnuplot',gnuplot_output_dir+compare_raw_filename+".gnuplot")
  67                 shutil.copy(gnuplot_output_dir+'compare.gnuplot',gnuplot_output_dir+compare_smooth_filename+".gnuplot")
  68                 shutil.copy(gnuplot_output_dir+'compare.gnuplot',gnuplot_output_dir+compare_trend_filename+".gnuplot")
  69
  70                 #Setting up a different output filename for each kind of graph
  71                 compare_raw=open(gnuplot_output_dir+compare_raw_filename + ".gnuplot",'a')
  72                 compare_raw.write("set output '%s.png'\n" % compare_raw_filename)
  73                 compare_smooth=open(gnuplot_output_dir+compare_smooth_filename+".gnuplot",'a')
  74                 compare_smooth.write("set output '%s.png'\n" % compare_smooth_filename)
  75                 compare_trend=open(gnuplot_output_dir+compare_trend_filename+".gnuplot",'a')
  76                 compare_trend.write("set output '%s.png'\n" % compare_trend_filename)
  77
  78         pos=0
  79         # Let's create a temporary file for each selected fio file
  80         for file in fio_data_file:
  81                 tmp_filename = "gnuplot_temp_file.%d" % pos
  82
  83                 # Plotting comparing graphs doesn't have a meaning unless if there is at least 2 traces
  84                 if len(fio_data_file) > 1:
  85                         # Adding the plot instruction for each kind of comparing graphs
  86                         if pos ==0 :
  87                                 compare_raw.write("plot '%s' using 2:3 with linespoints title '%s'" % (tmp_filename,fio_data_file[pos]))
  88                                 compare_smooth.write("plot '%s' using 2:3 smooth csplines title '%s'" % (tmp_filename,fio_data_file[pos]))
  89                                 compare_trend.write("plot '%s' using 2:3 smooth bezier title '%s'" % (tmp_filename,fio_data_file[pos]))
  90                         else:
  91                                 compare_raw.write(",\\\n'%s' using 2:3 with linespoints title '%s'" % (tmp_filename,fio_data_file[pos]))
  92                                 compare_smooth.write(",\\\n'%s' using 2:3 smooth csplines title '%s'" % (tmp_filename,fio_data_file[pos]))
  93                                 compare_trend.write(",\\\n'%s' using 2:3 smooth bezier title '%s'" % (tmp_filename,fio_data_file[pos]))
  94
  95                 png_file=file.replace('.log','')
  96                 raw_filename = "%s-2Draw" % (png_file)
  97                 smooth_filename = "%s-2Dsmooth" % (png_file)
  98                 trend_filename = "%s-2Dtrend" % (png_file)
  99                 avg  = average(disk_perf[pos])
 100                 f.write("call \'%s/graph2D.gpm\' \'%s' \'%s\' \'%s\' \'%s\' \'%s\' \'%s\' \'%s\' \'%f\'\n" % (gpm_dir,title,tmp_filename,fio_data_file[pos],raw_filename,mode,smooth_filename,trend_filename,avg))
 101                 pos = pos +1
 102
 103         # Plotting comparing graphs doesn't have a meaning unless if there is at least 2 traces
 104         if len(fio_data_file) > 1:
 105                 os.remove(gnuplot_output_dir+"compare.gnuplot")
 106                 compare_raw.close()
 107                 compare_smooth.close()
 108                 compare_trend.close()
 109         f.close()
 110
 111 def generate_gnuplot_math_script(title,gnuplot_output_filename,mode,average,gnuplot_output_dir,gpm_dir):
 112         filename=gnuplot_output_dir+'mymath';
 113         f=open(filename,'a')
 114         f.write("call \'%s/math.gpm\' \'%s' \'%s\' \'\' \'%s\' \'%s\' %s\n" % (gpm_dir,title,gnuplot_output_filename,gnuplot_output_filename,mode,average))
 115         f.close()
 116
 117 def compute_aggregated_file(fio_data_file, gnuplot_output_filename, gnuplot_output_dir):
 118         print "Processing data file 2/2"
 119         temp_files=[]
 120         pos=0
 121
 122         # Let's create a temporary file for each selected fio file
 123         for file in fio_data_file:
 124                 tmp_filename = "%sgnuplot_temp_file.%d" % (gnuplot_output_dir, pos)
 125                 temp_files.append(open(tmp_filename,'r'))
 126                 pos = pos +1
 127
 128         f = open(gnuplot_output_dir+gnuplot_output_filename, "w")
 129         index=0
 130         # Let's add some information
 131         for tempfile in temp_files:
 132                     f.write("# Disk%d was coming from %s\n" % (index,fio_data_file[index]))
 133                     f.write(tempfile.read())
 134                     f.write("\n")
 135                     tempfile.close()
 136                     index = index + 1
 137         f.close()
 138
 139 def average(s): return sum(s) * 1.0 / len(s)
 140
 141 def compute_temp_file(fio_data_file,disk_perf,gnuplot_output_dir, min_time, max_time):
 142         end_time=max_time
 143         if end_time == -1:
 144                 end_time="infinite"
 145         print "Processing data file 1/2 with %s<time<%s" % (min_time,end_time)
 146         files=[]
 147         temp_outfile=[]
 148         blk_size=0
 149         for file in fio_data_file:
 150                 files.append(open(file))
 151                 pos = len(files) - 1
 152                 tmp_filename = "%sgnuplot_temp_file.%d" % (gnuplot_output_dir,pos)
 153                 gnuplot_file=open(tmp_filename,'w')
 154                 temp_outfile.append(gnuplot_file)
 155                 gnuplot_file.write("#Temporary file based on file %s\n" % file)
 156                 disk_perf.append([])
 157
 158         shall_break = False
 159         while True:
 160                 current_line=[]
 161                 nb_empty_files=0
 162                 nb_files=len(files)
 163                 for myfile in files:
 164                         s=myfile.readline().replace(',',' ').split()
 165                         if not s:
 166                                 nb_empty_files+=1
 167                                 s="-1, 0, 0, 0".replace(',',' ').split()
 168
 169                         if (nb_empty_files == nb_files):
 170                                 shall_break=True
 171                                 break;
 172
 173                         current_line.append(s);
 174
 175                 if shall_break == True:
 176                         break
 177
 178                 last_time = -1
 179                 index=-1
 180                 perfs=[]
 181                 for line in enumerate(current_line):
 182                         # Index will be used to remember what file was featuring what value
 183                         index=index+1
 184
 185                         time, perf, x, block_size = line[1]
 186                         if (blk_size == 0):
 187                                 try:
 188                                         blk_size=int(block_size)
 189                                 except:
 190                                         print "Error while reading the following line :"
 191                                         print line
 192                                         sys.exit(1);
 193
 194                         # We ignore the first 500msec as it doesn't seems to be part of the real benchmark
 195                         # Time < 500 usually reports BW=0 breaking the min computing
 196                         if (min_time == 0):
 197                                 min_time==0.5
 198
 199                         # Then we estimate if the data we got is part of the time range we want to plot
 200                         if ((float(time)>(float(min_time)*1000)) and ((int(time) < (int(max_time)*1000)) or max_time==-1)):
 201                                         disk_perf[index].append(int(perf))
 202                                         perfs.append("%d %s %s"% (index, time, perf))
 203
 204                 # If we reach this point, it means that all the traces are coherent
 205                 for p in enumerate(perfs):
 206                         index, perf_time,perf = p[1].split()
 207                         temp_outfile[int(index)].write("%s %.2f %s\n" % (index, float(float(perf_time)/1000), perf))
 208
 209
 210         for file in files:
 211                 file.close()
 212         for file in temp_outfile:
 213                 file.close()
 214         return blk_size
 215
 216 def compute_math(fio_data_file, title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir):
 217         print "Computing Maths"
 218         global_min=[]
 219         global_max=[]
 220         average_file=open(gnuplot_output_dir+gnuplot_output_filename+'.average', 'w')
 221         min_file=open(gnuplot_output_dir+gnuplot_output_filename+'.min', 'w')
 222         max_file=open(gnuplot_output_dir+gnuplot_output_filename+'.max', 'w')
 223         stddev_file=open(gnuplot_output_dir+gnuplot_output_filename+'.stddev', 'w')
 224         global_file=open(gnuplot_output_dir+gnuplot_output_filename+'.global','w')
 225
 226         min_file.write('DiskName %s\n' % mode)
 227         max_file.write('DiskName %s\n'% mode)
 228         average_file.write('DiskName %s\n'% mode)
 229         stddev_file.write('DiskName %s\n'% mode )
 230         for disk in xrange(len(fio_data_file)):
 231 #               print disk_perf[disk]
 232                 min_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 233                 max_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 234                 average_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 235                 stddev_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 236                 avg  = average(disk_perf[disk])
 237                 variance = map(lambda x: (x - avg)**2, disk_perf[disk])
 238                 standard_deviation = math.sqrt(average(variance))
 239 #               print "Disk%d [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (disk,min(disk_perf[disk]),max(disk_perf[disk]),avg, standard_deviation)
 240                 average_file.write('%d %d\n' % (disk, avg))
 241                 stddev_file.write('%d %d\n' % (disk, standard_deviation))
 242                 local_min=min(disk_perf[disk])
 243                 local_max=max(disk_perf[disk])
 244                 min_file.write('%d %d\n' % (disk, local_min))
 245                 max_file.write('%d %d\n' % (disk, local_max))
 246                 global_min.append(int(local_min))
 247                 global_max.append(int(local_max))
 248
 249         global_disk_perf = sum(disk_perf, [])
 250         avg  = average(global_disk_perf)
 251         variance = map(lambda x: (x - avg)**2, global_disk_perf)
 252         standard_deviation = math.sqrt(average(variance))
 253
 254         global_file.write('min=%.2f\n' % min(global_disk_perf))
 255         global_file.write('max=%.2f\n' % max(global_disk_perf))
 256         global_file.write('avg=%.2f\n' % avg)
 257         global_file.write('stddev=%.2f\n' % standard_deviation)
 258         global_file.write('values_count=%d\n' % len(global_disk_perf))
 259         global_file.write('disks_count=%d\n' % len(fio_data_file))
 260         #print "Global [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (min(global_disk_perf),max(global_disk_perf),avg, standard_deviation)
 261
 262         average_file.close()
 263         min_file.close()
 264         max_file.close()
 265         stddev_file.close()
 266         global_file.close()
 267         try:
 268                 os.remove(gnuplot_output_dir+'mymath')
 269         except:
 270                 True
 271
 272         generate_gnuplot_math_script("Average values of "+title,gnuplot_output_filename+'.average',mode,int(avg),gnuplot_output_dir,gpm_dir)
 273         generate_gnuplot_math_script("Min values of "+title,gnuplot_output_filename+'.min',mode,average(global_min),gnuplot_output_dir,gpm_dir)
 274         generate_gnuplot_math_script("Max values of "+title,gnuplot_output_filename+'.max',mode,average(global_max),gnuplot_output_dir,gpm_dir)
 275         generate_gnuplot_math_script("Standard Deviation of "+title,gnuplot_output_filename+'.stddev',mode,int(standard_deviation),gnuplot_output_dir,gpm_dir)
 276
 277 def parse_global_files(fio_data_file, global_search):
 278         max_result=0
 279         max_file=''
 280         for file in fio_data_file:
 281                 f=open(file)
 282                 disk_count=0
 283                 search_value=-1
 284
 285                 # Let's read the complete file
 286                 while True:
 287                         try:
 288                                 # We do split the name from the value
 289                                 name,value=f.readline().split("=")
 290                         except:
 291                                 f.close()
 292                                 break
 293                         # If we ended the file
 294                         if not name:
 295                                 # Let's process what we have
 296                                 f.close()
 297                                 break
 298                         else:
 299                                 # disks_count is not global_search item
 300                                 # As we need it for some computation, let's save it
 301                                 if name=="disks_count":
 302                                         disks_count=int(value)
 303
 304                                 # Let's catch the searched item
 305                                 if global_search in name:
 306                                         search_value=float(value)
 307
 308                 # Let's process the avg value by estimated the global bandwidth per file
 309                 # We keep the biggest in memory for reporting
 310                 if global_search == "avg":
 311                         if (disks_count > 0) and (search_value != -1):
 312                                 result=disks_count*search_value
 313                                 if (result > max_result):
 314                                         max_result=result
 315                                         max_file=file
 316         # Let's print the avg output
 317         if global_search == "avg":
 318                 print "Biggest aggregated value of %s was %2.f in file %s\n" % (global_search, max_result, max_file)
 319         else:
 320                 print "Global search %s is not yet implemented\n" % global_search
 321
 322 def render_gnuplot(fio_data_file, gnuplot_output_dir):
 323         print "Running gnuplot Rendering"
 324         try:
 325                 # Let's render all the compared files if some
 326                 if len(fio_data_file) > 1:
 327                         print " |-> Rendering comparing traces"
 328                         os.system("cd %s; for i in *.gnuplot; do gnuplot $i; done" % gnuplot_output_dir)
 329                 print " |-> Rendering math traces"
 330                 os.system("cd %s; gnuplot mymath" % gnuplot_output_dir)
 331                 print " |-> Rendering 2D & 3D traces"
 332                 os.system("cd %s; gnuplot mygraph" % gnuplot_output_dir)
 333
 334                 name_of_directory="the current"
 335                 if gnuplot_output_dir != "./":
 336                         name_of_directory=gnuplot_output_dir
 337                 print "\nRendering traces are available in %s directory" % name_of_directory
 338         except:
 339                 print "Could not run gnuplot on mymath or mygraph !\n"
 340                 sys.exit(1);
 341
 342 def print_help():
 343     print 'fio2gnuplot.py -ghbiod -t <title> -o <outputfile> -p <pattern> -G <type> -m <time> -M <time>'
 344     print
 345     print '-h --help                           : Print this help'
 346     print '-p <pattern> or --pattern <pattern> : A pattern in regexp to select fio input files'
 347     print '-b           or --bandwidth         : A predefined pattern for selecting *_bw.log files'
 348     print '-i           or --iops              : A predefined pattern for selecting *_iops.log files'
 349     print '-g           or --gnuplot           : Render gnuplot traces before exiting'
 350     print '-o           or --outputfile <file> : The basename for gnuplot traces'
 351     print '                                       - Basename is set with the pattern if defined'
 352     print '-d           or --outputdir <dir>   : The directory where gnuplot shall render files'
 353     print '-t           or --title <title>     : The title of the gnuplot traces'
 354     print '                                       - Title is set with the block size detected in fio traces'
 355     print '-G           or --Global <type>     : Search for <type> in .global files match by a pattern'
 356     print '                                       - Available types are : min, max, avg, stddev'
 357     print '                                       - The .global extension is added automatically to the pattern'
 358     print '-m           or --min_time <time>   : Only consider data starting from <time> seconds (default is 0)'
 359     print '-M           or --max_time <time>   : Only consider data ending before <time> seconds (default is -1 aka nolimit)'
 360
 361 def main(argv):
 362     mode='unknown'
 363     pattern=''
 364     pattern_set_by_user=False
 365     title='No title'
 366     gnuplot_output_filename='result'
 367     gnuplot_output_dir='./'
 368     gpm_dir="/usr/share/fio/"
 369     disk_perf=[]
 370     run_gnuplot=False
 371     parse_global=False
 372     global_search=''
 373     min_time=0
 374     max_time=-1
 375
 376     if not os.path.isfile(gpm_dir+'math.gpm'):
 377             gpm_dir="/usr/local/share/fio/"
 378             if not os.path.isfile(gpm_dir+'math.gpm'):
 379                     print "Looks like fio didn't got installed properly as no gpm files found in '/usr/share/fio' or '/usr/local/share/fio'\n"
 380                     sys.exit(3)
 381
 382     try:
 383             opts, args = getopt.getopt(argv[1:],"ghbio:d:t:p:G:m:M:")
 384     except getopt.GetoptError:
 385          print_help()
 386          sys.exit(2)
 387
 388     for opt, arg in opts:
 389       if opt in ("-b", "--bandwidth"):
 390          pattern='*_bw.log'
 391       elif opt in ("-i", "--iops"):
 392          pattern='*_iops.log'
 393       elif opt in ("-p", "--pattern"):
 394          pattern_set_by_user=True
 395          pattern=arg
 396          pattern=pattern.replace('\\','')
 397       elif opt in ("-o", "--outputfile"):
 398          gnuplot_output_filename=arg
 399       elif opt in ("-d", "--outputdir"):
 400          gnuplot_output_dir=arg
 401          if not gnuplot_output_dir.endswith('/'):
 402                 gnuplot_output_dir=gnuplot_output_dir+'/'
 403          if not os.path.exists(gnuplot_output_dir):
 404                 os.makedirs(gnuplot_output_dir)
 405       elif opt in ("-t", "--title"):
 406          title=arg
 407       elif opt in ("-m", "--min_time"):
 408          min_time=arg
 409       elif opt in ("-M", "--max_time"):
 410          max_time=arg
 411       elif opt in ("-g", "--gnuplot"):
 412          run_gnuplot=True
 413       elif opt in ("-G", "--Global"):
 414          parse_global=True
 415          global_search=arg
 416       elif opt in ("-h", "--help"):
 417           print_help()
 418           sys.exit(1)
 419
 420     # Adding .global extension to the file
 421     if parse_global==True:
 422             if not gnuplot_output_filename.endswith('.global'):
 423                 pattern = pattern+'.global'
 424
 425     fio_data_file=find_file('.',pattern)
 426     if len(fio_data_file) == 0:
 427             print "No log file found with pattern %s!" % pattern
 428             sys.exit(1)
 429     else:
 430             print "%d files Selected with pattern '%s'" % (len(fio_data_file), pattern)
 431
 432     fio_data_file=sorted(fio_data_file, key=str.lower)
 433     for file in fio_data_file:
 434         print ' |-> %s' % file
 435         if "_bw.log" in file :
 436                 mode="Bandwidth (KB/sec)"
 437         if "_iops.log" in file :
 438                 mode="IO per Seconds (IO/sec)"
 439     if (title == 'No title') and (mode != 'unknown'):
 440             if "Bandwidth" in mode:
 441                     title='Bandwidth benchmark with %d fio results' % len(fio_data_file)
 442             if "IO" in mode:
 443                     title='IO benchmark with %d fio results' % len(fio_data_file)
 444
 445     print
 446     #We need to adjust the output filename regarding the pattern required by the user
 447     if (pattern_set_by_user == True):
 448             gnuplot_output_filename=pattern
 449             # As we do have some regexp in the pattern, let's make this simpliest
 450             # We do remove the simpliest parts of the expression to get a clear file name
 451             gnuplot_output_filename=gnuplot_output_filename.replace('-*-','-')
 452             gnuplot_output_filename=gnuplot_output_filename.replace('*','-')
 453             gnuplot_output_filename=gnuplot_output_filename.replace('--','-')
 454             gnuplot_output_filename=gnuplot_output_filename.replace('.log','')
 455             # Insure that we don't have any starting or trailing dash to the filename
 456             gnuplot_output_filename = gnuplot_output_filename[:-1] if gnuplot_output_filename.endswith('-') else gnuplot_output_filename
 457             gnuplot_output_filename = gnuplot_output_filename[1:] if gnuplot_output_filename.startswith('-') else gnuplot_output_filename
 458
 459     if parse_global==True:
 460         parse_global_files(fio_data_file, global_search)
 461     else:
 462         blk_size=compute_temp_file(fio_data_file,disk_perf,gnuplot_output_dir,min_time,max_time)
 463         title="%s @ Blocksize = %dK" % (title,blk_size/1024)
 464         compute_aggregated_file(fio_data_file, gnuplot_output_filename, gnuplot_output_dir)
 465         compute_math(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 466         generate_gnuplot_script(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 467
 468         if (run_gnuplot==True):
 469                 render_gnuplot(fio_data_file, gnuplot_output_dir)
 470
 471     # Cleaning temporary files
 472     try:
 473         os.remove('gnuplot_temp_file.*')
 474     except:
 475         True
 476
 477 #Main
 478 if __name__ == "__main__":
 479     sys.exit(main(sys.argv))