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