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=0
 180                 perfs=[]
 181                 for line in current_line:
 182                         time, perf, x, block_size = line
 183                         if (blk_size == 0):
 184                                 try:
 185                                         blk_size=int(block_size)
 186                                 except:
 187                                         print "Error while reading the following line :"
 188                                         print line
 189                                         sys.exit(1);
 190
 191                         # We ignore the first 500msec as it doesn't seems to be part of the real benchmark
 192                         # Time < 500 usually reports BW=0 breaking the min computing
 193                         if (((int(time)) > 500) or (int(time)==-1)):
 194                                 # Now it's time to estimate if the data we got is part of the time range we want to plot
 195                                 if ((int(time)>(int(min_time)*1000)) and ((int(time) < (int(max_time)*1000)) or max_time=="-1")):
 196                                         disk_perf[index].append(int(perf))
 197                                         perfs.append("%s %s"% (time, perf))
 198                                         index = index + 1
 199
 200                 # If we reach this point, it means that all the traces are coherent
 201                 for p in enumerate(perfs):
 202                         perf_time,perf = p[1].split()
 203                         if (perf_time != "-1"):
 204                                 temp_outfile[p[0]].write("%s %.2f %s\n" % (p[0], float(float(perf_time)/1000), perf))
 205
 206
 207         for file in files:
 208                 file.close()
 209         for file in temp_outfile:
 210                 file.close()
 211         return blk_size
 212
 213 def compute_math(fio_data_file, title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir):
 214         print "Computing Maths"
 215         global_min=[]
 216         global_max=[]
 217         average_file=open(gnuplot_output_dir+gnuplot_output_filename+'.average', 'w')
 218         min_file=open(gnuplot_output_dir+gnuplot_output_filename+'.min', 'w')
 219         max_file=open(gnuplot_output_dir+gnuplot_output_filename+'.max', 'w')
 220         stddev_file=open(gnuplot_output_dir+gnuplot_output_filename+'.stddev', 'w')
 221         global_file=open(gnuplot_output_dir+gnuplot_output_filename+'.global','w')
 222
 223         min_file.write('DiskName %s\n' % mode)
 224         max_file.write('DiskName %s\n'% mode)
 225         average_file.write('DiskName %s\n'% mode)
 226         stddev_file.write('DiskName %s\n'% mode )
 227         for disk in xrange(len(fio_data_file)):
 228 #               print disk_perf[disk]
 229                 min_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 230                 max_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 231                 average_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 232                 stddev_file.write("# Disk%d was coming from %s\n" % (disk,fio_data_file[disk]))
 233                 avg  = average(disk_perf[disk])
 234                 variance = map(lambda x: (x - avg)**2, disk_perf[disk])
 235                 standard_deviation = math.sqrt(average(variance))
 236 #               print "Disk%d [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (disk,min(disk_perf[disk]),max(disk_perf[disk]),avg, standard_deviation)
 237                 average_file.write('%d %d\n' % (disk, avg))
 238                 stddev_file.write('%d %d\n' % (disk, standard_deviation))
 239                 local_min=min(disk_perf[disk])
 240                 local_max=max(disk_perf[disk])
 241                 min_file.write('%d %d\n' % (disk, local_min))
 242                 max_file.write('%d %d\n' % (disk, local_max))
 243                 global_min.append(int(local_min))
 244                 global_max.append(int(local_max))
 245
 246         global_disk_perf = sum(disk_perf, [])
 247         avg  = average(global_disk_perf)
 248         variance = map(lambda x: (x - avg)**2, global_disk_perf)
 249         standard_deviation = math.sqrt(average(variance))
 250
 251         global_file.write('min=%.2f\n' % min(global_disk_perf))
 252         global_file.write('max=%.2f\n' % max(global_disk_perf))
 253         global_file.write('avg=%.2f\n' % avg)
 254         global_file.write('stddev=%.2f\n' % standard_deviation)
 255         global_file.write('values_count=%d\n' % len(global_disk_perf))
 256         global_file.write('disks_count=%d\n' % len(fio_data_file))
 257         #print "Global [ min=%.2f max=%.2f avg=%.2f stddev=%.2f \n" % (min(global_disk_perf),max(global_disk_perf),avg, standard_deviation)
 258
 259         average_file.close()
 260         min_file.close()
 261         max_file.close()
 262         stddev_file.close()
 263         global_file.close()
 264         try:
 265                 os.remove(gnuplot_output_dir+'mymath')
 266         except:
 267                 True
 268
 269         generate_gnuplot_math_script("Average values of "+title,gnuplot_output_filename+'.average',mode,int(avg),gnuplot_output_dir,gpm_dir)
 270         generate_gnuplot_math_script("Min values of "+title,gnuplot_output_filename+'.min',mode,average(global_min),gnuplot_output_dir,gpm_dir)
 271         generate_gnuplot_math_script("Max values of "+title,gnuplot_output_filename+'.max',mode,average(global_max),gnuplot_output_dir,gpm_dir)
 272         generate_gnuplot_math_script("Standard Deviation of "+title,gnuplot_output_filename+'.stddev',mode,int(standard_deviation),gnuplot_output_dir,gpm_dir)
 273
 274 def parse_global_files(fio_data_file, global_search):
 275         max_result=0
 276         max_file=''
 277         for file in fio_data_file:
 278                 f=open(file)
 279                 disk_count=0
 280                 search_value=-1
 281
 282                 # Let's read the complete file
 283                 while True:
 284                         try:
 285                                 # We do split the name from the value
 286                                 name,value=f.readline().split("=")
 287                         except:
 288                                 f.close()
 289                                 break
 290                         # If we ended the file
 291                         if not name:
 292                                 # Let's process what we have
 293                                 f.close()
 294                                 break
 295                         else:
 296                                 # disks_count is not global_search item
 297                                 # As we need it for some computation, let's save it
 298                                 if name=="disks_count":
 299                                         disks_count=int(value)
 300
 301                                 # Let's catch the searched item
 302                                 if global_search in name:
 303                                         search_value=float(value)
 304
 305                 # Let's process the avg value by estimated the global bandwidth per file
 306                 # We keep the biggest in memory for reporting
 307                 if global_search == "avg":
 308                         if (disks_count > 0) and (search_value != -1):
 309                                 result=disks_count*search_value
 310                                 if (result > max_result):
 311                                         max_result=result
 312                                         max_file=file
 313         # Let's print the avg output
 314         if global_search == "avg":
 315                 print "Biggest aggregated value of %s was %2.f in file %s\n" % (global_search, max_result, max_file)
 316         else:
 317                 print "Global search %s is not yet implemented\n" % global_search
 318
 319 def render_gnuplot(fio_data_file, gnuplot_output_dir):
 320         print "Running gnuplot Rendering"
 321         try:
 322                 # Let's render all the compared files if some
 323                 if len(fio_data_file) > 1:
 324                         print " |-> Rendering comparing traces"
 325                         os.system("cd %s; for i in *.gnuplot; do gnuplot $i; done" % gnuplot_output_dir)
 326                 print " |-> Rendering math traces"
 327                 os.system("cd %s; gnuplot mymath" % gnuplot_output_dir)
 328                 print " |-> Rendering 2D & 3D traces"
 329                 os.system("cd %s; gnuplot mygraph" % gnuplot_output_dir)
 330
 331                 name_of_directory="the current"
 332                 if gnuplot_output_dir != "./":
 333                         name_of_directory=gnuplot_output_dir
 334                 print "\nRendering traces are available in %s directory" % name_of_directory
 335         except:
 336                 print "Could not run gnuplot on mymath or mygraph !\n"
 337                 sys.exit(1);
 338
 339 def print_help():
 340     print 'fio2gnuplot.py -ghbiod -t <title> -o <outputfile> -p <pattern> -G <type> -m <time> -M <time>'
 341     print
 342     print '-h --help                           : Print this help'
 343     print '-p <pattern> or --pattern <pattern> : A pattern in regexp to select fio input files'
 344     print '-b           or --bandwidth         : A predefined pattern for selecting *_bw.log files'
 345     print '-i           or --iops              : A predefined pattern for selecting *_iops.log files'
 346     print '-g           or --gnuplot           : Render gnuplot traces before exiting'
 347     print '-o           or --outputfile <file> : The basename for gnuplot traces'
 348     print '                                       - Basename is set with the pattern if defined'
 349     print '-d           or --outputdir <dir>   : The directory where gnuplot shall render files'
 350     print '-t           or --title <title>     : The title of the gnuplot traces'
 351     print '                                       - Title is set with the block size detected in fio traces'
 352     print '-G           or --Global <type>     : Search for <type> in .global files match by a pattern'
 353     print '                                       - Available types are : min, max, avg, stddev'
 354     print '                                       - The .global extension is added automatically to the pattern'
 355     print '-m           or --min_time <time>   : Only consider data starting from <time> seconds (default is 0)'
 356     print '-M           or --max_time <time>   : Only consider data ending before <time> seconds (default is -1 aka nolimit)'
 357
 358 def main(argv):
 359     mode='unknown'
 360     pattern=''
 361     pattern_set_by_user=False
 362     title='No title'
 363     gnuplot_output_filename='result'
 364     gnuplot_output_dir='./'
 365     gpm_dir="/usr/share/fio/"
 366     disk_perf=[]
 367     run_gnuplot=False
 368     parse_global=False
 369     global_search=''
 370     min_time=0
 371     max_time=-1
 372
 373     if not os.path.isfile(gpm_dir+'math.gpm'):
 374             gpm_dir="/usr/local/share/fio/"
 375             if not os.path.isfile(gpm_dir+'math.gpm'):
 376                     print "Looks like fio didn't got installed properly as no gpm files found in '/usr/share/fio' or '/usr/local/share/fio'\n"
 377                     sys.exit(3)
 378
 379     try:
 380             opts, args = getopt.getopt(argv[1:],"ghbio:d:t:p:G:m:M:")
 381     except getopt.GetoptError:
 382          print_help()
 383          sys.exit(2)
 384
 385     for opt, arg in opts:
 386       if opt in ("-b", "--bandwidth"):
 387          pattern='*_bw.log'
 388       elif opt in ("-i", "--iops"):
 389          pattern='*_iops.log'
 390       elif opt in ("-p", "--pattern"):
 391          pattern_set_by_user=True
 392          pattern=arg
 393          pattern=pattern.replace('\\','')
 394       elif opt in ("-o", "--outputfile"):
 395          gnuplot_output_filename=arg
 396       elif opt in ("-d", "--outputdir"):
 397          gnuplot_output_dir=arg
 398          if not gnuplot_output_dir.endswith('/'):
 399                 gnuplot_output_dir=gnuplot_output_dir+'/'
 400          if not os.path.exists(gnuplot_output_dir):
 401                 os.makedirs(gnuplot_output_dir)
 402       elif opt in ("-t", "--title"):
 403          title=arg
 404       elif opt in ("-m", "--min_time"):
 405          min_time=arg
 406       elif opt in ("-M", "--max_time"):
 407          max_time=arg
 408       elif opt in ("-g", "--gnuplot"):
 409          run_gnuplot=True
 410       elif opt in ("-G", "--Global"):
 411          parse_global=True
 412          global_search=arg
 413       elif opt in ("-h", "--help"):
 414           print_help()
 415           sys.exit(1)
 416
 417     # Adding .global extension to the file
 418     if parse_global==True:
 419             if not gnuplot_output_filename.endswith('.global'):
 420                 pattern = pattern+'.global'
 421
 422     fio_data_file=find_file('.',pattern)
 423     if len(fio_data_file) == 0:
 424             print "No log file found with pattern %s!" % pattern
 425             sys.exit(1)
 426     else:
 427             print "%d files Selected with pattern '%s'" % (len(fio_data_file), pattern)
 428
 429     fio_data_file=sorted(fio_data_file, key=str.lower)
 430     for file in fio_data_file:
 431         print ' |-> %s' % file
 432         if "_bw.log" in file :
 433                 mode="Bandwidth (KB/sec)"
 434         if "_iops.log" in file :
 435                 mode="IO per Seconds (IO/sec)"
 436     if (title == 'No title') and (mode != 'unknown'):
 437             if "Bandwidth" in mode:
 438                     title='Bandwidth benchmark with %d fio results' % len(fio_data_file)
 439             if "IO" in mode:
 440                     title='IO benchmark with %d fio results' % len(fio_data_file)
 441
 442     print
 443     #We need to adjust the output filename regarding the pattern required by the user
 444     if (pattern_set_by_user == True):
 445             gnuplot_output_filename=pattern
 446             # As we do have some regexp in the pattern, let's make this simpliest
 447             # We do remove the simpliest parts of the expression to get a clear file name
 448             gnuplot_output_filename=gnuplot_output_filename.replace('-*-','-')
 449             gnuplot_output_filename=gnuplot_output_filename.replace('*','-')
 450             gnuplot_output_filename=gnuplot_output_filename.replace('--','-')
 451             gnuplot_output_filename=gnuplot_output_filename.replace('.log','')
 452             # Insure that we don't have any starting or trailing dash to the filename
 453             gnuplot_output_filename = gnuplot_output_filename[:-1] if gnuplot_output_filename.endswith('-') else gnuplot_output_filename
 454             gnuplot_output_filename = gnuplot_output_filename[1:] if gnuplot_output_filename.startswith('-') else gnuplot_output_filename
 455
 456     if parse_global==True:
 457         parse_global_files(fio_data_file, global_search)
 458     else:
 459         blk_size=compute_temp_file(fio_data_file,disk_perf,gnuplot_output_dir,min_time,max_time)
 460         title="%s @ Blocksize = %dK" % (title,blk_size/1024)
 461         compute_aggregated_file(fio_data_file, gnuplot_output_filename, gnuplot_output_dir)
 462         compute_math(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 463         generate_gnuplot_script(fio_data_file,title,gnuplot_output_filename,gnuplot_output_dir,mode,disk_perf,gpm_dir)
 464
 465         if (run_gnuplot==True):
 466                 render_gnuplot(fio_data_file, gnuplot_output_dir)
 467
 468     # Cleaning temporary files
 469     try:
 470         os.remove('gnuplot_temp_file.*')
 471     except:
 472         True
 473
 474 #Main
 475 if __name__ == "__main__":
 476     sys.exit(main(sys.argv))