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