[27549] | 1 | #!/usr/bin/perl
|
---|
| 2 |
|
---|
| 3 | use strict;
|
---|
| 4 | use warnings;
|
---|
| 5 |
|
---|
| 6 | use Devel::Peek;
|
---|
| 7 | use Time::Local;
|
---|
| 8 |
|
---|
| 9 | print "\n===== GS Hadoop Report =====\n";
|
---|
| 10 | print "Generate a report (CSV) from the output of a parallel processing Green-\n";
|
---|
| 11 | print "stone import, suitable for feeding into the Gantt chart generator.\n";
|
---|
| 12 | print "============================\n\n";
|
---|
| 13 |
|
---|
| 14 | # Configuration
|
---|
| 15 | if (!defined $ARGV[0] || !-d $ARGV[0])
|
---|
| 16 | {
|
---|
| 17 | die("usage: hadoop_report.pl <path to results>\n");
|
---|
| 18 | }
|
---|
| 19 | my $results_path = $ARGV[0];
|
---|
| 20 |
|
---|
| 21 | # Read in hadoop.log and parse top level record
|
---|
| 22 | print ' * Reading and parsing "hadoop.log"... ';
|
---|
| 23 | my $job_record = {'host'=>'', 'job'=>'', 'start'=>0, 'end'=>0, 'cpu_time'=>0};
|
---|
| 24 | my $hadoop_log_path = $results_path . '/hadoop.log';
|
---|
| 25 | if (open(HLIN, '<:utf8', $hadoop_log_path))
|
---|
| 26 | {
|
---|
| 27 | while (my $line = <HLIN>)
|
---|
| 28 | {
|
---|
| 29 | if ($line =~ /host:(.+)/)
|
---|
| 30 | {
|
---|
| 31 | $job_record->{'host'} = $1;
|
---|
| 32 | }
|
---|
| 33 | elsif ($line =~ /Running job: job_(\d+_\d+)/)
|
---|
| 34 | {
|
---|
| 35 | $job_record->{'job'} = $1;
|
---|
| 36 | }
|
---|
| 37 | elsif ($line =~ /CPU time spent \(ms\)=(\d+)/)
|
---|
| 38 | {
|
---|
| 39 | $job_record->{'cpu_time'} = $1;
|
---|
| 40 | }
|
---|
| 41 | elsif ($job_record->{'start'} == 0 && $line =~ /(\d\d)\/(\d\d)\/(\d\d) (\d\d):(\d\d):(\d\d)/)
|
---|
| 42 | {
|
---|
| 43 | $job_record->{'start'} = timelocal($6, $5, $4, $3, ($2 - 1), $1);
|
---|
| 44 | }
|
---|
| 45 | elsif ($line =~ /(\d\d)\/(\d\d)\/(\d\d) (\d\d):(\d\d):(\d\d)/)
|
---|
| 46 | {
|
---|
| 47 | my $end = timelocal($6, $5, $4, $3, ($2 - 1), $1);
|
---|
| 48 | if ($end > $job_record->{'end'})
|
---|
| 49 | {
|
---|
| 50 | $job_record->{'end'} = $end;
|
---|
| 51 | }
|
---|
| 52 | }
|
---|
| 53 | }
|
---|
| 54 | close(HLIN);
|
---|
| 55 | if ($job_record->{'start'} == 0 || $job_record->{'end'} == 0)
|
---|
| 56 | {
|
---|
| 57 | die('Error! Failed to parse timing information from log: ' . $hadoop_log_path);
|
---|
| 58 | }
|
---|
| 59 | }
|
---|
| 60 | else
|
---|
| 61 | {
|
---|
| 62 | die('Error! Failed to open file for reading: ' . $hadoop_log_path);
|
---|
| 63 | }
|
---|
| 64 | print "Done!\n";
|
---|
| 65 |
|
---|
| 66 | # Read in data_locality.csv (will be matched to task logs)
|
---|
| 67 | print ' * Reading and parsing "data_locality.csv"... ';
|
---|
| 68 | my $data_was_local = {};
|
---|
| 69 | my $data_locality_csv_path = $results_path . '/data_locality.csv';
|
---|
| 70 | if (open(DLIN, '<:utf8', $data_locality_csv_path))
|
---|
| 71 | {
|
---|
| 72 | while (my $line = <DLIN>)
|
---|
| 73 | {
|
---|
| 74 | if ($line =~ /(\d+),\d,(\d)/)
|
---|
| 75 | {
|
---|
| 76 | $data_was_local->{$1} = $2;
|
---|
| 77 | }
|
---|
| 78 | }
|
---|
| 79 | close(DLIN);
|
---|
| 80 | }
|
---|
| 81 | else
|
---|
| 82 | {
|
---|
| 83 | die('Error! Failed to open file for reading: ' . $data_locality_csv_path);
|
---|
| 84 | }
|
---|
| 85 | print "Done!\n";
|
---|
| 86 |
|
---|
| 87 | # Read in all task logs and parse task records
|
---|
| 88 | my $task_records;
|
---|
| 89 | print " * Locating task logs...\n";
|
---|
| 90 | if (opendir(DH, $results_path))
|
---|
| 91 | {
|
---|
| 92 | my @files = readdir(DH);
|
---|
| 93 | foreach my $file (sort @files)
|
---|
| 94 | {
|
---|
| 95 | if ($file =~ /import-hadoop-(\d+_\d+)_m_(\d+)_\d+\.log/)
|
---|
| 96 | {
|
---|
| 97 | my $job_no = $1;
|
---|
| 98 | my $task_no = $2;
|
---|
| 99 | my $is_data_local = $data_was_local->{$task_no};
|
---|
| 100 | my $task_record = {'host'=>'', 'cpu'=>0, 'job' => $job_no, 'task' => $task_no, 'start'=>0, 'end'=>0, 'cpu_time'=>0, 'data_locality'=>$is_data_local, 'file'=>''};
|
---|
| 101 | print ' - Reading and parsing "' . $file . '"... ';
|
---|
| 102 | my $task_log_path = $results_path . '/' . $file;
|
---|
| 103 | my $io_time = 0;
|
---|
| 104 | if (open(TIN, '<:utf8', $task_log_path))
|
---|
| 105 | {
|
---|
| 106 | my $io_start_time = 0;
|
---|
| 107 | while (my $line = <TIN>)
|
---|
| 108 | {
|
---|
| 109 | if ($line =~ /\[Started:(\d+)\]/)
|
---|
| 110 | {
|
---|
| 111 | $task_record->{'start'} = $1;
|
---|
| 112 | }
|
---|
| 113 | elsif ($line =~ /\[Host:([^\]]+)\]/)
|
---|
| 114 | {
|
---|
| 115 | $task_record->{'host'} = $1;
|
---|
| 116 | }
|
---|
| 117 | elsif ($line =~ /\[CPU:(\d+)\]/)
|
---|
| 118 | {
|
---|
| 119 | $task_record->{'cpu'} = $1;
|
---|
| 120 | }
|
---|
[27560] | 121 | elsif ($line =~ /\[Map:([^\]]+)=>1\]/)
|
---|
[27549] | 122 | {
|
---|
| 123 | $task_record->{'file'} = $1;
|
---|
| 124 | }
|
---|
| 125 | elsif ($line =~ /\[IOS:(\d+)\]/)
|
---|
| 126 | {
|
---|
| 127 | $io_start_time = $1;
|
---|
| 128 | }
|
---|
| 129 | elsif ($io_start_time > 0 && $line =~ /\[IOE:(\d+)\]/)
|
---|
| 130 | {
|
---|
| 131 | $io_time += ($1 - $io_start_time);
|
---|
| 132 | $io_start_time = 0;
|
---|
| 133 | }
|
---|
| 134 | elsif ($line =~ /\[Completed:(\d+)\]/)
|
---|
| 135 | {
|
---|
| 136 | my $end_time = $1;
|
---|
| 137 | $task_record->{'end'} = $end_time;
|
---|
| 138 | if ($io_start_time > 0)
|
---|
| 139 | {
|
---|
| 140 | $io_time += ($end_time - $io_start_time);
|
---|
| 141 | }
|
---|
| 142 | }
|
---|
| 143 | }
|
---|
| 144 | close(TIN);
|
---|
| 145 | # Calculate CPU time (total time - IO time)
|
---|
| 146 | $task_record->{'cpu_time'} = $task_record->{'end'} - $task_record->{'start'} - $io_time;
|
---|
| 147 | # Store this record
|
---|
| 148 | $task_records->{$task_no} = $task_record;
|
---|
| 149 | }
|
---|
| 150 | else
|
---|
| 151 | {
|
---|
| 152 | die('Error! Failed to open file for reading: ' . $task_log_path);
|
---|
| 153 | }
|
---|
| 154 | print "Done!\n";
|
---|
| 155 | }
|
---|
| 156 | }
|
---|
| 157 | close(DH);
|
---|
| 158 | }
|
---|
| 159 | else
|
---|
| 160 | {
|
---|
| 161 | die('Error! Failed to open directory for reading: ' . $results_path);
|
---|
| 162 | }
|
---|
| 163 | print ' - Processed ' . scalar(keys(%{$task_records})) . " records\n";
|
---|
| 164 |
|
---|
| 165 | # Generate compute-node records
|
---|
| 166 | print ' * Generating compute node records... ';
|
---|
| 167 | my $node_records;
|
---|
| 168 | foreach my $taskno (sort keys %{$task_records})
|
---|
| 169 | {
|
---|
| 170 | my $task_record = $task_records->{$taskno};
|
---|
| 171 | my $node_record = {'host'=>'', 'cpu'=>0, 'job' => '', 'start'=>0, 'end'=>0, 'cpu_time'=>0};
|
---|
| 172 | # - retrieve any existing record
|
---|
| 173 | my $worker_id = $task_record->{'host'} . '#' . $task_record->{'cpu'};
|
---|
| 174 | if (defined $node_records->{$worker_id})
|
---|
| 175 | {
|
---|
| 176 | $node_record = $node_records->{$worker_id};
|
---|
| 177 | }
|
---|
| 178 | if ($node_record->{'host'} eq '')
|
---|
| 179 | {
|
---|
| 180 | $node_record->{'host'} = $task_record->{'host'};
|
---|
| 181 | }
|
---|
| 182 | if ($node_record->{'cpu'} == 0)
|
---|
| 183 | {
|
---|
| 184 | $node_record->{'cpu'} = $task_record->{'cpu'};
|
---|
| 185 | }
|
---|
| 186 | if ($node_record->{'job'} eq '')
|
---|
| 187 | {
|
---|
| 188 | $node_record->{'job'} = $task_record->{'job'};
|
---|
| 189 | }
|
---|
| 190 | if ($node_record->{'start'} == 0 || $task_record->{'start'} < $node_record->{'start'})
|
---|
| 191 | {
|
---|
| 192 | $node_record->{'start'} = $task_record->{'start'};
|
---|
| 193 | }
|
---|
| 194 | if ($node_record->{'end'} == 0 || $task_record->{'end'} > $node_record->{'end'})
|
---|
| 195 | {
|
---|
| 196 | $node_record->{'end'} = $task_record->{'end'};
|
---|
| 197 | }
|
---|
| 198 | $node_record->{'cpu_time'} += $task_record->{'cpu_time'};
|
---|
| 199 | # - store it
|
---|
| 200 | $node_records->{$worker_id} = $node_record;
|
---|
| 201 | }
|
---|
| 202 | print "Done!\n";
|
---|
| 203 |
|
---|
| 204 | # Write out CSV of all information
|
---|
| 205 | my $report_csv_path = $results_path . '/timing.csv';
|
---|
| 206 | if (open(CSVOUT, '>:utf8', $report_csv_path))
|
---|
| 207 | {
|
---|
| 208 | my $row_counter = 1;
|
---|
| 209 | # Header
|
---|
| 210 | print CSVOUT "id,name,hostname,start,end,cputime,dl,pid,filename\n";
|
---|
| 211 | # Master Record
|
---|
| 212 | print CSVOUT $row_counter . ',M0,' . $job_record->{'host'} . ',' . $job_record->{'start'} . ',' . $job_record->{'end'} . ',' . ($job_record->{'cpu_time'} / 1000) . ",0,0,NA\n";
|
---|
| 213 | $row_counter++;
|
---|
| 214 | # For each compute node record
|
---|
| 215 | foreach my $worker_id (sort keys %{$node_records})
|
---|
| 216 | {
|
---|
| 217 | my $node_record = $node_records->{$worker_id};
|
---|
| 218 | my $node_id = $row_counter;
|
---|
| 219 | $row_counter++;
|
---|
| 220 | print CSVOUT $node_id . ',W' . $node_record->{'cpu'} . ',' . $node_record->{'host'} . ',' . $node_record->{'start'} . ',' . $node_record->{'end'} . ',' . $node_record->{'cpu_time'} . ",0,1,NA\n";
|
---|
| 221 | # List the child task records
|
---|
| 222 | foreach my $taskno (sort keys %{$task_records})
|
---|
| 223 | {
|
---|
| 224 | my $task_record = $task_records->{$taskno};
|
---|
| 225 | if ($task_record->{'host'} . '#' . $task_record->{'cpu'} eq $worker_id)
|
---|
| 226 | {
|
---|
| 227 | print CSVOUT $row_counter . ',T' . ($task_record->{'task'} + 0) . ',' . $task_record->{'host'} . ',' . $task_record->{'start'} . ',' . $task_record->{'end'} . ',' . $task_record->{'cpu_time'} . ',' . $task_record->{'data_locality'} . ',' . $node_id . ',' . $task_record->{'file'} . "\n";
|
---|
| 228 | $row_counter++;
|
---|
| 229 | }
|
---|
| 230 | }
|
---|
| 231 | }
|
---|
| 232 | close(CSVOUT);
|
---|
| 233 | }
|
---|
| 234 | else
|
---|
| 235 | {
|
---|
| 236 | die('Error! Failed to open file for writing: ' . $report_csv_path);
|
---|
| 237 | }
|
---|
| 238 |
|
---|
| 239 | print "Complete!\n\n";
|
---|
| 240 | exit;
|
---|
| 241 |
|
---|