1 | #!/usr/bin/perl
|
---|
2 |
|
---|
3 | use strict;
|
---|
4 | use warnings;
|
---|
5 |
|
---|
6 | use Sort::Naturally;
|
---|
7 | use POSIX qw(floor strftime);
|
---|
8 |
|
---|
9 | print "\n===== Generate Timing (GANTT) =====\n";
|
---|
10 |
|
---|
11 | # 0. Configuration
|
---|
12 | my $debug = 0;
|
---|
13 | my $import_dir;
|
---|
14 |
|
---|
15 | # 1. Initialization
|
---|
16 | if (!defined $ARGV[0] || !-d $ARGV[0])
|
---|
17 | {
|
---|
18 | &printUsage('Directory not provided or doesn\'t exist');
|
---|
19 | }
|
---|
20 | my $dir = $ARGV[0];
|
---|
21 | my $timing_csv_path = &filenameCat($dir, 'timing.csv');
|
---|
22 | if (!-e $timing_csv_path)
|
---|
23 | {
|
---|
24 | &printUsage('Directory doesn\'t contain timing.csv: ' . $dir);
|
---|
25 | }
|
---|
26 | print 'Timing File: ' . $timing_csv_path . "\n";
|
---|
27 | my $chart_width = 1024;
|
---|
28 | if (defined $ARGV[1])
|
---|
29 | {
|
---|
30 | if ($ARGV[1] !~ /^\d+$/)
|
---|
31 | {
|
---|
32 | &printUsage('Chart width not a number');
|
---|
33 | }
|
---|
34 | $chart_width = $ARGV[1];
|
---|
35 | }
|
---|
36 | print "Chart Width: " . $chart_width . "px\n";
|
---|
37 | print "===================================\n\n";
|
---|
38 |
|
---|
39 | # Read in timing.csv and parse information into data structure
|
---|
40 | my $timing_data = {};
|
---|
41 | my $id_2_worker_id = {};
|
---|
42 | if (open(TIN, '<:utf8', $timing_csv_path))
|
---|
43 | {
|
---|
44 | my $line;
|
---|
45 | while ($line = <TIN>)
|
---|
46 | {
|
---|
47 | my @parts = split(/,/, $line);
|
---|
48 | if ($parts[1] eq 'M0')
|
---|
49 | {
|
---|
50 | $timing_data->{'M'} = {'N'=>$parts[2], 'S'=>$parts[3], 'E'=>$parts[4]};
|
---|
51 | }
|
---|
52 | elsif ($parts[1] =~ /W\d+/)
|
---|
53 | {
|
---|
54 | $timing_data->{$parts[1]} = {'N'=>$parts[2], 'S'=>$parts[3], 'E'=>$parts[4], 'F'=>{}};
|
---|
55 | $id_2_worker_id->{$parts[0]} = $parts[1];
|
---|
56 | }
|
---|
57 | elsif ($parts[1] =~ /T\d+/)
|
---|
58 | {
|
---|
59 | my $worker_id = $id_2_worker_id->{$parts[7]};
|
---|
60 | my $stop = $parts[4];
|
---|
61 | my $filepath = $parts[8];
|
---|
62 | $import_dir = &longestCommonPath($filepath, $import_dir);
|
---|
63 | $timing_data->{$worker_id}->{'F'}->{$parts[3]} = {'FN'=>$filepath, 'PS'=>($stop - $parts[5]), 'PE'=>$stop, 'E'=>$stop};
|
---|
64 | }
|
---|
65 | }
|
---|
66 | close(TIN);
|
---|
67 | }
|
---|
68 | else
|
---|
69 | {
|
---|
70 | die('Error! Failed to open file for reading: ' . $timing_csv_path);
|
---|
71 | }
|
---|
72 | my $number_of_workers = scalar(keys(%{$id_2_worker_id}));;
|
---|
73 |
|
---|
74 | # 3. Produce pretty HTML chart of timing information including jobs
|
---|
75 | print " * Generating timing information as HTML... ";
|
---|
76 | open(HTMLOUT, '>:utf8', $dir . '/gantt.html') or die('Error! Failed to open file for writing: gantt.html');
|
---|
77 | print HTMLOUT "<html>\n";
|
---|
78 | print HTMLOUT '<head>' . "\n";
|
---|
79 | print HTMLOUT '<style type="text/css">' . "\n";
|
---|
80 | print HTMLOUT 'div.thread {position:relative}' . "\n";
|
---|
81 | print HTMLOUT 'div.master {border:1px solid gray;color:white;font-weight:bold}' . "\n";
|
---|
82 | print HTMLOUT 'div.worker {border:1px solid black;background-color:green;color:white;font-weight:bold}' . "\n";
|
---|
83 | print HTMLOUT 'div.time {font-size:smaller;font-weight:normal}' . "\n";
|
---|
84 | print HTMLOUT 'div.job {background-color:transparent;color:black;border:1px solid black;display:block;font-size:smaller;position:relative;text-align:center}' . "\n";
|
---|
85 | print HTMLOUT 'span.process {z-index:-1;background-color:#C7C7C7;position:absolute}' . "\n";
|
---|
86 | print HTMLOUT "th {text-align:left}\n";
|
---|
87 | print HTMLOUT '</style>' . "\n";
|
---|
88 | print HTMLOUT '</head>' . "\n";
|
---|
89 | print HTMLOUT "<body>\n";
|
---|
90 | print HTMLOUT "<h2>Statistics</h2>\n";
|
---|
91 | print HTMLOUT "<table>\n";
|
---|
92 |
|
---|
93 | my $total_duration = $timing_data->{'M'}->{'E'} - $timing_data->{'M'}->{'S'};
|
---|
94 | my $file_count = 0;
|
---|
95 | my $total_io_time = 0;
|
---|
96 | my $total_process_time = 0;
|
---|
97 | my $fastest_file = 0;
|
---|
98 | my $slowest_file = 0;
|
---|
99 | my $problem_files = 0;
|
---|
100 | foreach my $worker_id (keys %{$timing_data})
|
---|
101 | {
|
---|
102 | if ($worker_id ne 'M')
|
---|
103 | {
|
---|
104 | foreach my $job_start ( keys %{$timing_data->{$worker_id}->{'F'}} )
|
---|
105 | {
|
---|
106 | my $process_start = $timing_data->{$worker_id}->{'F'}->{$job_start}->{'PS'};
|
---|
107 | my $process_end = $timing_data->{$worker_id}->{'F'}->{$job_start}->{'PE'};
|
---|
108 | my $job_end = $timing_data->{$worker_id}->{'F'}->{$job_start}->{'E'};
|
---|
109 | if ($process_start == 0 || $process_end == 0 || $job_end == 0)
|
---|
110 | {
|
---|
111 | $problem_files++;
|
---|
112 | }
|
---|
113 | else
|
---|
114 | {
|
---|
115 | my $io_duration = ($process_start - $job_start) + ($job_end - $process_end);
|
---|
116 | my $process_duration = $process_end - $process_start;
|
---|
117 | my $total_duration = $io_duration + $process_duration;
|
---|
118 | &debugPrint("filename: " . $timing_data->{$worker_id}->{'F'}->{$job_start}->{'FN'});
|
---|
119 | &debugPrint("start: $job_start ps: $process_start pe: $process_end end: $job_end");
|
---|
120 | &debugPrint("io: $io_duration process: $process_duration duration: $total_duration");
|
---|
121 | # Running stats
|
---|
122 | $total_io_time += $io_duration;
|
---|
123 | $total_process_time += $process_duration;
|
---|
124 | if ($fastest_file == 0 || $total_duration < $fastest_file)
|
---|
125 | {
|
---|
126 | $fastest_file = $total_duration;
|
---|
127 | }
|
---|
128 | if ($slowest_file == 0 || $total_duration > $slowest_file)
|
---|
129 | {
|
---|
130 | $slowest_file = $total_duration;
|
---|
131 | }
|
---|
132 | }
|
---|
133 | # Shorten filename
|
---|
134 | $timing_data->{$worker_id}->{'F'}->{$job_start}->{'FN'} = substr($timing_data->{$worker_id}->{'F'}->{$job_start}->{'FN'}, length($import_dir) + 1);
|
---|
135 | $file_count++;
|
---|
136 | }
|
---|
137 | }
|
---|
138 | }
|
---|
139 | my $avg_processing_time = floor(($total_io_time + $total_process_time) / $file_count);
|
---|
140 |
|
---|
141 | print HTMLOUT "<tr><th>Import Directory:</th><td>" . $import_dir . "</td></tr>\n";
|
---|
142 | print HTMLOUT "<tr><th>Processing Time:</th><td>" . &renderTime($total_duration) . "</td></tr>\n";
|
---|
143 | print HTMLOUT "<tr><th>Processing Threads:</th><td>" . $number_of_workers . "</td></tr>\n";
|
---|
144 | print HTMLOUT "<tr><th>Files Processed:</th><td>" . $file_count . "</td></tr>\n";
|
---|
145 | print HTMLOUT "<tr><th>Problem Files:</th><td>" . $problem_files . "</td></tr>\n";
|
---|
146 | print HTMLOUT "<tr><th>Serial Processing Time:</th><td>" . &renderTime($total_process_time) . "</td></tr>\n";
|
---|
147 | print HTMLOUT "<tr><th>Serial IO Time:</th><td>" . &renderTime($total_io_time) . "</td></tr>\n";
|
---|
148 | print HTMLOUT "<tr><th>Average File Processing Time:</th><td>" . &renderTime($avg_processing_time) . "</td></tr>\n";
|
---|
149 | print HTMLOUT "<tr><th>Fastest File:</th><td>" . &renderTime($fastest_file) . "</td></tr>\n";
|
---|
150 | print HTMLOUT "<tr><th>Slowest File:</th><td>" . &renderTime($slowest_file) . "</td></tr>\n";
|
---|
151 |
|
---|
152 | print HTMLOUT "</table>\n";
|
---|
153 | print HTMLOUT "<hr />\n";
|
---|
154 | print HTMLOUT "<h2>Timing Chart (Gannt)</h2>\n";
|
---|
155 | print HTMLOUT renderLine($chart_width, $timing_data->{'M'}->{'S'}, $timing_data->{'M'}->{'E'}, 'master', $timing_data->{'M'}->{'N'}, $timing_data->{'M'}->{'S'}, $timing_data->{'M'}->{'E'}, {});
|
---|
156 | foreach my $worker_id (nsort keys %{$timing_data})
|
---|
157 | {
|
---|
158 | if ($worker_id ne 'M')
|
---|
159 | {
|
---|
160 | my $data = $timing_data->{$worker_id};
|
---|
161 | print HTMLOUT renderLine($chart_width, $timing_data->{'M'}->{'S'}, $timing_data->{'M'}->{'E'}, 'worker', $worker_id . ' [' . $data->{'N'} . ']', $data->{'S'}, $data->{'E'}, $data->{'F'});
|
---|
162 | }
|
---|
163 | }
|
---|
164 | print HTMLOUT '<div>' . "\n";
|
---|
165 | print HTMLOUT "</body>\n";
|
---|
166 | print HTMLOUT "</html>";
|
---|
167 | close(HTMLOUT);
|
---|
168 |
|
---|
169 | print "Done!\n";
|
---|
170 | print "Complete!\n\n";
|
---|
171 | exit;
|
---|
172 |
|
---|
173 |
|
---|
174 | ## @function debugPrint()
|
---|
175 | #
|
---|
176 | sub debugPrint
|
---|
177 | {
|
---|
178 | my $msg = shift(@_);
|
---|
179 | if ($debug)
|
---|
180 | {
|
---|
181 | print STDERR '[DEBUG] ' . $msg . "\n";
|
---|
182 | }
|
---|
183 | }
|
---|
184 | ## debugPrint() ##
|
---|
185 |
|
---|
186 |
|
---|
187 | ## @function filenameCat
|
---|
188 | #
|
---|
189 | sub filenameCat
|
---|
190 | {
|
---|
191 | my $path = join('/', @_);
|
---|
192 | $path =~ s/[\/\\]+/\//g;
|
---|
193 | # protocols
|
---|
194 | $path =~ s/^(HDFS|HDFSShell|HDThriftFS):\//$1:\/\//;
|
---|
195 | return $path;
|
---|
196 | }
|
---|
197 | ## filenameCat() ##
|
---|
198 |
|
---|
199 | ## @function printUsage()
|
---|
200 | #
|
---|
201 | sub printUsage
|
---|
202 | {
|
---|
203 | my $msg = shift(@_);
|
---|
204 | if (defined $msg)
|
---|
205 | {
|
---|
206 | print 'Error! ' . $msg . "\n";
|
---|
207 | }
|
---|
208 | die("Usage: generate_gantt.pl <results dir> [<width in pixels>]\n\n");
|
---|
209 | }
|
---|
210 | ## printUsage() ##
|
---|
211 |
|
---|
212 |
|
---|
213 | ## @function longestCommonPath
|
---|
214 | #
|
---|
215 | sub longestCommonPath
|
---|
216 | {
|
---|
217 | my ($path_new, $path_current) = @_;
|
---|
218 | my $result = '';
|
---|
219 | if (defined $path_current)
|
---|
220 | {
|
---|
221 | my @path_new_parts = split(/\//, $path_new);
|
---|
222 | my @path_current_parts = split(/\//, $path_current);
|
---|
223 | my @path_parts;
|
---|
224 | for (my $i = 0; $i < scalar(@path_current_parts); $i++)
|
---|
225 | {
|
---|
226 | if ($path_current_parts[$i] eq $path_new_parts[$i])
|
---|
227 | {
|
---|
228 | push(@path_parts, $path_new_parts[$i]);
|
---|
229 | }
|
---|
230 | else
|
---|
231 | {
|
---|
232 | last;
|
---|
233 | }
|
---|
234 | }
|
---|
235 | $result = &filenameCat(@path_parts);
|
---|
236 | }
|
---|
237 | else
|
---|
238 | {
|
---|
239 | $result = $path_new;
|
---|
240 | }
|
---|
241 | return $result;
|
---|
242 | }
|
---|
243 | ## longestCommonPath() ##
|
---|
244 |
|
---|
245 |
|
---|
246 | ## @function renderLine()
|
---|
247 | #
|
---|
248 | sub renderLine
|
---|
249 | {
|
---|
250 | my ($table_width, $start, $end, $class, $tname, $tstart, $tend, $jobs) = @_;
|
---|
251 | &debugPrint("renderLine($table_width, $start, $end, $class, $tname, $tstart, $tend, <jobs>)");
|
---|
252 | # All timings need to be relative to 0 (relative start)
|
---|
253 | my $duration = $end - $start;
|
---|
254 | my $rtstart = $tstart - $start;
|
---|
255 | my $rtend = $tend - $start;
|
---|
256 | # We need to scale these depending on the timing of this thread relative to
|
---|
257 | # the master thread
|
---|
258 | my $width = $chart_width;
|
---|
259 | my $left = 0;
|
---|
260 | if ($start != $tstart)
|
---|
261 | {
|
---|
262 | my $left_offset_percent = $rtstart / $duration;
|
---|
263 | $left = $left_offset_percent * $table_width;
|
---|
264 | }
|
---|
265 | # - subtract any left offset from width
|
---|
266 | $width = $width - $left;
|
---|
267 | # - right offset directly subtracted from width
|
---|
268 | if ($end != $tend)
|
---|
269 | {
|
---|
270 | my $right_offset_percent = ($duration - $rtend) / $duration;
|
---|
271 | my $right = $right_offset_percent * $table_width;
|
---|
272 | $width = $width - $right;
|
---|
273 | }
|
---|
274 | my $html = '<div class="thread ' . $class . '" style="left:' . $left . 'px;width:' . $width . 'px;">';
|
---|
275 | if ($class eq 'master')
|
---|
276 | {
|
---|
277 | $html .= '<div style="background-color:blue">';
|
---|
278 | }
|
---|
279 | $html .= '<div class="time" style="display:table-cell">' . &renderTime($rtstart) . '</div><div style="display:table-cell;padding-left:20px;width:100%;">' . ucfirst($class) . ': ' . $tname . '</div><div class="time" style="display:table-cell">' . renderTime($rtend) . '</div></div>';
|
---|
280 | my $previous_jright = 0;
|
---|
281 | foreach my $jstart (sort keys %{$jobs})
|
---|
282 | {
|
---|
283 | my $rjstart = $jstart - $start;
|
---|
284 | my $rpstart = $jobs->{$jstart}->{'PS'} - $start;
|
---|
285 | my $rpend = $jobs->{$jstart}->{'PE'} - $start;
|
---|
286 | my $rjend = $jobs->{$jstart}->{'E'} - $start;
|
---|
287 | my $jduration = $jobs->{$jstart}->{'E'} - $jstart;
|
---|
288 | # Scale Job co-ordinates
|
---|
289 | my $jleft_percent = $rjstart / $duration;
|
---|
290 | my $jleft = floor($jleft_percent * $table_width);
|
---|
291 | my $jwidth_percent = $jduration / $duration;
|
---|
292 | my $jwidth = floor($jwidth_percent * $table_width);
|
---|
293 | if ($jleft < $previous_jright)
|
---|
294 | {
|
---|
295 | $jleft = $previous_jright;
|
---|
296 | }
|
---|
297 | if ($jleft + $jwidth > $left + $width)
|
---|
298 | {
|
---|
299 | $jwidth = ($left + $width) - $jleft;
|
---|
300 | }
|
---|
301 | # Then scale process timings within that!
|
---|
302 | my $rpleft_percent = ($rpstart - $rjstart) / $duration;
|
---|
303 | my $rpleft = floor($rpleft_percent * $table_width);
|
---|
304 | my $rpwidth_percent = ($rpend - $rpstart) / $duration;
|
---|
305 | my $rpwidth = floor($rpwidth_percent * $table_width);
|
---|
306 | $html .= '<div class="job" style="left:' . $jleft . 'px;width:' . $jwidth . 'px" title="S:' . renderTime($rjstart) . ', PS:' . renderTime($rpstart) . ', PE:' . renderTime($rpend) . ', E:' . renderTime($rjend) . '"><span class="process" style="left:' . $rpleft . 'px;width:' . $rpwidth . 'px"> </span><span style="background-color:transparent;z-index:1">' . $jobs->{$jstart}->{'FN'} . '</span></div>';
|
---|
307 | $previous_jright = $jleft + $jwidth;
|
---|
308 | }
|
---|
309 | return $html;
|
---|
310 | }
|
---|
311 | ## renderLine() ##
|
---|
312 |
|
---|
313 |
|
---|
314 | ## @function renderTime()
|
---|
315 | #
|
---|
316 | sub renderTime
|
---|
317 | {
|
---|
318 | my ($seconds) = @_;
|
---|
319 | my $time_str = '';
|
---|
320 | # determine how many hours
|
---|
321 | my $an_hour = 60 * 60;
|
---|
322 | my $hours = floor($seconds / $an_hour);
|
---|
323 | $seconds = $seconds - ($hours * $an_hour);
|
---|
324 | my $a_minute = 60;
|
---|
325 | my $minutes = floor($seconds / $a_minute);
|
---|
326 | $seconds = $seconds - ($minutes * $a_minute);
|
---|
327 | if ($hours > 0)
|
---|
328 | {
|
---|
329 | $time_str = sprintf('%dh%02dm%02ds', $hours, $minutes, $seconds);
|
---|
330 | }
|
---|
331 | elsif ($minutes > 0)
|
---|
332 | {
|
---|
333 | $time_str = sprintf('%dm%02ds', $minutes, $seconds);
|
---|
334 | }
|
---|
335 | else
|
---|
336 | {
|
---|
337 | $time_str = $seconds . 's';
|
---|
338 | }
|
---|
339 | return $time_str;
|
---|
340 | }
|
---|