[27036] | 1 | #!/usr/bin/perl
|
---|
| 2 |
|
---|
| 3 | # Pragma
|
---|
| 4 | use strict;
|
---|
| 5 | use warnings;
|
---|
| 6 |
|
---|
| 7 | # Configuration
|
---|
| 8 | my $debug = 0;
|
---|
| 9 |
|
---|
| 10 | # Requires setup.bash to have been sourced
|
---|
| 11 | BEGIN
|
---|
| 12 | {
|
---|
| 13 | die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
|
---|
| 14 | die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
|
---|
| 15 | die "HADOOP_PREFIX not set (set in <gsdl>/ext/parallel_processing/setup.bash)\n" unless defined $ENV{'HADOOP_PREFIX'};
|
---|
| 16 | die "HDFS HOST not set (set in <gsdl>/ext/parallel_processing/setup.bash)\n" unless defined $ENV{'HDFSHOST'};
|
---|
| 17 | die "HDFS PORT not set (set in <gsdl>/ext/parallel_processing/setup.bash)\n" unless defined $ENV{'HDFSPORT'};
|
---|
[27041] | 18 | # Ensure Greenstone Perl locations are in INC
|
---|
| 19 | unshift (@INC, $ENV{'GSDLHOME'} . '/perllib');
|
---|
| 20 | unshift (@INC, $ENV{'GSDLHOME'} . '/perllib/cpan');
|
---|
| 21 | # we'll need the perl version number
|
---|
[27124] | 22 | my ($version_number) = `perl-version.pl`;
|
---|
[27041] | 23 | if (defined $ENV{'GSDLEXTS'})
|
---|
| 24 | {
|
---|
| 25 | my @extensions = split(/:/,$ENV{'GSDLEXTS'});
|
---|
| 26 | foreach my $e (@extensions)
|
---|
| 27 | {
|
---|
| 28 | my $ext_prefix = $ENV{'GSDLHOME'} . '/ext/' . $e;
|
---|
| 29 | unshift (@INC, $ext_prefix . '/perllib');
|
---|
| 30 | unshift (@INC, $ext_prefix . '/perllib/cpan');
|
---|
| 31 | unshift (@INC, $ext_prefix . '/' . $ENV{'GSDLOS'} . '/lib/perl/' . $version_number);
|
---|
| 32 | }
|
---|
| 33 | }
|
---|
[27036] | 34 | }
|
---|
| 35 |
|
---|
| 36 | # Libraries (depends on unshift above
|
---|
| 37 | use Sort::Key::Natural qw(natsort);
|
---|
| 38 |
|
---|
| 39 | # Begin
|
---|
| 40 | print "===== Parse Hadoop Log =====\n";
|
---|
| 41 |
|
---|
| 42 | # 0. Init
|
---|
| 43 | if (!defined $ARGV[0])
|
---|
| 44 | {
|
---|
[27587] | 45 | die("usage: parse_task_info_from_hadoop_log.pl <results dir> [-debug]\n");
|
---|
[27036] | 46 | }
|
---|
| 47 | my $results_dir = $ARGV[0];
|
---|
| 48 | if (!-d $results_dir)
|
---|
| 49 | {
|
---|
| 50 | die("Error! Can't find results directory: " . $results_dir . "\n");
|
---|
| 51 | }
|
---|
| 52 | print " Results directory: " . $results_dir . "\n";
|
---|
| 53 |
|
---|
[27587] | 54 | if (defined $ARGV[1] && $ARGV[1] eq '-debug')
|
---|
| 55 | {
|
---|
| 56 | $debug = 1;
|
---|
| 57 | }
|
---|
| 58 |
|
---|
[27586] | 59 | # 1. Determine job ID
|
---|
[27036] | 60 | my $hadoop_log_path = &fileCat($results_dir, 'hadoop.log');
|
---|
| 61 | if (!-e $hadoop_log_path)
|
---|
| 62 | {
|
---|
| 63 | die("Error! Hadoop log file cannot be found: " . $hadoop_log_path . "\n");
|
---|
| 64 | }
|
---|
| 65 | print " Hadoop log path: " . $hadoop_log_path . "\n";
|
---|
[27586] | 66 | print " * Determine JobID: ";
|
---|
| 67 | my $job_id;
|
---|
| 68 | my $result = `grep "Running job:" "$hadoop_log_path"`;
|
---|
| 69 | if ($result =~ /Running job: job_(\d+_\d+)/)
|
---|
| 70 | {
|
---|
| 71 | $job_id = $1;
|
---|
| 72 | }
|
---|
| 73 | else
|
---|
| 74 | {
|
---|
| 75 | die("Error! Failed to locate JobID\n");
|
---|
| 76 | }
|
---|
| 77 | print $job_id . "\n";
|
---|
| 78 | # - we'll need the date to locate the appopriate log file
|
---|
[27588] | 79 | my $log_date_year = 0;
|
---|
| 80 | my $log_date_month = 0;
|
---|
| 81 | my $log_date_day = 0;
|
---|
[27586] | 82 | if ($job_id =~ /^(\d\d\d\d)(\d\d)(\d\d)/)
|
---|
| 83 | {
|
---|
[27588] | 84 | $log_date_year = $1;
|
---|
| 85 | $log_date_month = $2;
|
---|
| 86 | $log_date_day = $3;
|
---|
[27586] | 87 | }
|
---|
[27588] | 88 | else
|
---|
| 89 | {
|
---|
| 90 | die('Error! Failed to parse date from Job ID: ' . $job_id . "\n");
|
---|
| 91 | }
|
---|
[27036] | 92 |
|
---|
[27588] | 93 | # 2. Determine user and system details
|
---|
[27036] | 94 | my $username = `whoami`;
|
---|
| 95 | chomp($username);
|
---|
| 96 | print " Username: " . $username . "\n";
|
---|
| 97 | my $hostname = `hostname`;
|
---|
| 98 | chomp($hostname);
|
---|
| 99 | print " Hostname: " . $hostname . "\n";
|
---|
| 100 | my $data_locality_report_path = &fileCat($results_dir, 'data_locality.csv');
|
---|
| 101 | print " Report path: " . $data_locality_report_path . "\n";
|
---|
| 102 |
|
---|
[27586] | 103 | # 3. Parse log
|
---|
[27588] | 104 | print " * Parse JobTracker Log(s)...\n";
|
---|
[27036] | 105 | my $tid_2_splits = {};
|
---|
| 106 | my $tid_2_node = {};
|
---|
| 107 | my $aid_2_node = {};
|
---|
[27588] | 108 | my $job_complete = 0;
|
---|
| 109 | my $parsed_latest_log = 0;
|
---|
| 110 | while (!$job_complete && !$parsed_latest_log)
|
---|
[27036] | 111 | {
|
---|
[27588] | 112 | # - determine appropriate job tracker log
|
---|
| 113 | my $jobtracker_log_path_prefix = &fileCat($ENV{'HADOOP_PREFIX'}, 'logs', 'hadoop-' . $username . '-jobtracker-' . $hostname . '.log');
|
---|
| 114 | my $jobtracker_log_path = sprintf('%s.%04d-%02d-%02d', $jobtracker_log_path_prefix, $log_date_year, $log_date_month, $log_date_day);
|
---|
| 115 | # - maybe the log hasn't been rolled yet
|
---|
| 116 | if (!-e $jobtracker_log_path)
|
---|
[27036] | 117 | {
|
---|
[27588] | 118 | $jobtracker_log_path = $jobtracker_log_path_prefix;
|
---|
| 119 | # - nope, no applicable log found
|
---|
| 120 | if (!-e $jobtracker_log_path)
|
---|
[27036] | 121 | {
|
---|
[27588] | 122 | die('Error! Hadoop JobTracker log file cannot be found: ' . $jobtracker_log_path . "\n");
|
---|
[27036] | 123 | }
|
---|
[27588] | 124 | else
|
---|
[27036] | 125 | {
|
---|
[27588] | 126 | $parsed_latest_log = 1;
|
---|
[27036] | 127 | }
|
---|
[27588] | 128 | }
|
---|
| 129 | print " - parsing JobTracker log: " . $jobtracker_log_path . "\n";
|
---|
| 130 |
|
---|
| 131 | if (open(JTLIN, '<', $jobtracker_log_path))
|
---|
| 132 | {
|
---|
| 133 | my $line = '';
|
---|
| 134 | while ($line = <JTLIN>)
|
---|
[27036] | 135 | {
|
---|
[27588] | 136 | # Tips provide a match between task and file splits
|
---|
| 137 | if ($line =~ /tip:task_${job_id}(_m_\d+) has split on node:\/default-rack\/([^\.\r\n]+)/)
|
---|
[27036] | 138 | {
|
---|
[27588] | 139 | my $task_id = $job_id . $1;
|
---|
| 140 | my $compute_node = $2;
|
---|
| 141 | &debugPrint('found tip: ' . $task_id . ' => ' . $compute_node);
|
---|
| 142 | if (!defined $tid_2_splits->{$task_id})
|
---|
| 143 | {
|
---|
| 144 | $tid_2_splits->{$task_id} = [$compute_node];
|
---|
| 145 | }
|
---|
| 146 | else
|
---|
| 147 | {
|
---|
| 148 | push(@{$tid_2_splits->{$task_id}}, $compute_node);
|
---|
| 149 | }
|
---|
[27036] | 150 | }
|
---|
[27588] | 151 | # JobTracker (MAP) entries give us a mapping between task, attempt, and
|
---|
| 152 | # compute node
|
---|
| 153 | elsif ($line =~ /Adding task \(MAP\) 'attempt_${job_id}(_m_\d+)(_\d+)'.*tracker_([^\.]+).local/)
|
---|
| 154 | {
|
---|
| 155 | my $task_id = $job_id . $1;
|
---|
| 156 | my $attempt_id = $job_id . $1 . $2;
|
---|
| 157 | my $compute_node = $3;
|
---|
| 158 | &debugPrint('found MAP: ' . $attempt_id . ' => ' . $compute_node);
|
---|
| 159 | $aid_2_node->{$attempt_id} = {'compute_node' => $compute_node,
|
---|
| 160 | 'succeeded' => 0
|
---|
| 161 | };
|
---|
| 162 | }
|
---|
| 163 | # Watch for attempt successes (so we can weed out failures)
|
---|
| 164 | elsif ($line =~ /Task 'attempt_${job_id}(_m_\d+_\d+)' has completed .* successfully/)
|
---|
| 165 | {
|
---|
| 166 | my $attempt_id = $job_id . $1;
|
---|
| 167 | &debugPrint('successful attempt: ' . $attempt_id);
|
---|
| 168 | if (defined $aid_2_node->{$attempt_id})
|
---|
| 169 | {
|
---|
| 170 | $aid_2_node->{$attempt_id}->{'succeeded'} = 1;
|
---|
| 171 | }
|
---|
| 172 | }
|
---|
| 173 | # And job completion... so we can keep parsing other log files as
|
---|
| 174 | # necessary
|
---|
[27589] | 175 | elsif ($line =~ /Job job_${job_id} has completed successfully\./)
|
---|
[27588] | 176 | {
|
---|
| 177 | $job_complete = 1;
|
---|
| 178 | }
|
---|
[27036] | 179 | }
|
---|
[27588] | 180 | close(JTLIN);
|
---|
[27036] | 181 | }
|
---|
[27588] | 182 | else
|
---|
| 183 | {
|
---|
| 184 | die('Error! Failed to open JobTracker log for reading: ' . $jobtracker_log_path . "\n");
|
---|
| 185 | }
|
---|
| 186 | # Increment the day by one - unfortunately this leads to unpleasant date
|
---|
| 187 | # maths to ensure month and year roll over appropriately too
|
---|
| 188 | $log_date_day++;
|
---|
| 189 | # On to a new month?
|
---|
| 190 | # Leap Year Feb > 29 otherwise Feb > 28
|
---|
| 191 | # Apr, Jun, Sep, Nov > 30
|
---|
| 192 | # Rest > 31
|
---|
| 193 | if ($log_date_month == 2 && ((&isLeapYear($log_date_year) && $log_date_day > 29) || $log_date_day > 28))
|
---|
| 194 | {
|
---|
| 195 | $log_date_day = 1;
|
---|
| 196 | $log_date_month++;
|
---|
| 197 | }
|
---|
| 198 | elsif (($log_date_month == 4 || $log_date_month == 6 || $log_date_month == 9 || $log_date_month == 11) && $log_date_day > 30)
|
---|
| 199 | {
|
---|
| 200 | $log_date_day = 1;
|
---|
| 201 | $log_date_month++;
|
---|
| 202 | }
|
---|
| 203 | elsif ($log_date_day > 31)
|
---|
| 204 | {
|
---|
| 205 | $log_date_day = 1;
|
---|
| 206 | $log_date_month++;
|
---|
| 207 | }
|
---|
| 208 | # On to a new year?
|
---|
| 209 | if ($log_date_month > 12)
|
---|
| 210 | {
|
---|
| 211 | $log_date_month = 1;
|
---|
| 212 | $log_date_year++;
|
---|
| 213 | }
|
---|
[27036] | 214 | }
|
---|
[27589] | 215 | print " Done\n";
|
---|
[27588] | 216 |
|
---|
| 217 | if (!$job_complete)
|
---|
[27036] | 218 | {
|
---|
[27588] | 219 | print "Warning! Failed to parse in information for a complete job.\n";
|
---|
[27036] | 220 | }
|
---|
| 221 |
|
---|
[27586] | 222 | # 4. Write CSV of information
|
---|
[27036] | 223 | print " * Writing Job Information... ";
|
---|
[27588] | 224 | &debugPrint("AttemptID \tComputeNode\tSucceeded", 1);
|
---|
[27036] | 225 | foreach my $attempt_id (keys %{$aid_2_node})
|
---|
| 226 | {
|
---|
| 227 | &debugPrint($attempt_id . "\t" . $aid_2_node->{$attempt_id}->{'compute_node'} . "\t" . $aid_2_node->{$attempt_id}->{'succeeded'});
|
---|
| 228 | }
|
---|
[27588] | 229 | &debugPrint("TaskID \tComputeNodeSplits");
|
---|
| 230 | my $split_counter = 0;
|
---|
[27036] | 231 | foreach my $task_id (keys %{$tid_2_splits})
|
---|
| 232 | {
|
---|
| 233 | &debugPrint($task_id . "\t" . join(',', natsort(@{$tid_2_splits->{$task_id}})));
|
---|
[27588] | 234 | $split_counter++;
|
---|
[27036] | 235 | }
|
---|
[27588] | 236 | &debugPrint(' * Number of split records: ' . $split_counter);
|
---|
| 237 | &debugPrint('');
|
---|
[27412] | 238 |
|
---|
[27036] | 239 | # - open the CSV file and write out the combined information from above
|
---|
| 240 | if (open(CSVOUT, '>:utf8', $data_locality_report_path))
|
---|
| 241 | {
|
---|
| 242 | print CSVOUT "TaskNo,AttemptNo,Data Local,Compute Node,Splits\n";
|
---|
| 243 | foreach my $attempt_id (natsort(keys %{$aid_2_node}))
|
---|
| 244 | {
|
---|
| 245 | my ($job_id, $task_number, $attempt_number) = $attempt_id =~ /^(\d+_\d+)_m_(\d+)_(\d+)/;
|
---|
| 246 | my $task_id = $job_id . '_m_' . $task_number;
|
---|
| 247 | my $compute_node = $aid_2_node->{$attempt_id}->{'compute_node'};
|
---|
| 248 | my @splits = @{$tid_2_splits->{$task_id}};
|
---|
[27412] | 249 | my $data_local = 0;
|
---|
[27036] | 250 | if (grep($_ eq $compute_node, @splits))
|
---|
| 251 | {
|
---|
[27412] | 252 | $data_local = 1;
|
---|
[27036] | 253 | }
|
---|
[27412] | 254 | print CSVOUT $task_number . "," . $attempt_number . "," . $data_local . ",\"" . $compute_node . "\",\"" . join(',', natsort(@splits)) . "\"\n";
|
---|
[27589] | 255 | $tid_2_splits->{$task_id} = undef;
|
---|
[27036] | 256 | }
|
---|
[27588] | 257 |
|
---|
| 258 | # Report on any other splits that were recorded in the log, but for unknown
|
---|
| 259 | # reasons aren't matched with a 'successful' task
|
---|
| 260 | foreach my $task_id (keys %{$tid_2_splits})
|
---|
| 261 | {
|
---|
[27589] | 262 | if (defined $tid_2_splits->{$task_id})
|
---|
| 263 | {
|
---|
| 264 | my ($job_id, $task_number) = $task_id =~ /^(\d+_\d+)_m_(\d+)/;
|
---|
| 265 | my @splits = @{$tid_2_splits->{$task_id}};
|
---|
| 266 | print CSVOUT $task_number . ",-1,-1,\"\",\"" . join(',', natsort(@splits)) . "\"\n";
|
---|
| 267 | }
|
---|
[27588] | 268 | }
|
---|
| 269 |
|
---|
[27036] | 270 | close(CSVOUT);
|
---|
| 271 | }
|
---|
| 272 | else
|
---|
| 273 | {
|
---|
| 274 | die("Error! Failed to open file for writing: " . $data_locality_report_path);
|
---|
| 275 | }
|
---|
| 276 | print "Done\n";
|
---|
| 277 |
|
---|
[27586] | 278 | # 5. Done
|
---|
[27036] | 279 | print "===== Complete! =====\n\n";
|
---|
| 280 | exit;
|
---|
| 281 |
|
---|
| 282 | # Subs
|
---|
| 283 |
|
---|
| 284 | sub debugPrint
|
---|
| 285 | {
|
---|
[27588] | 286 | my ($msg, $force_newline) = @_;
|
---|
[27036] | 287 | if ($debug)
|
---|
| 288 | {
|
---|
[27588] | 289 | if (defined $force_newline && $force_newline)
|
---|
| 290 | {
|
---|
| 291 | print "\n[debug] " . $msg . "\n";
|
---|
| 292 | }
|
---|
| 293 | else
|
---|
| 294 | {
|
---|
| 295 | print '[debug] ' . $msg . "\n";
|
---|
| 296 | }
|
---|
[27036] | 297 | }
|
---|
| 298 | }
|
---|
| 299 |
|
---|
| 300 | sub fileCat
|
---|
| 301 | {
|
---|
| 302 | my $path = join('/', @_);
|
---|
| 303 | $path =~ s/\/\/+/\//g;
|
---|
| 304 | return $path;
|
---|
| 305 | }
|
---|
[27588] | 306 |
|
---|
| 307 | ## @function isLeapYear()
|
---|
| 308 | #
|
---|
| 309 | sub isLeapYear
|
---|
| 310 | {
|
---|
| 311 | my ($year) = @_;
|
---|
| 312 | if ($year % 400 == 0)
|
---|
| 313 | {
|
---|
| 314 | return 1;
|
---|
| 315 | }
|
---|
| 316 | elsif ($year % 100 == 0)
|
---|
| 317 | {
|
---|
| 318 | return 0;
|
---|
| 319 | }
|
---|
| 320 | elsif ($year % 4 == 0)
|
---|
| 321 | {
|
---|
| 322 | return 1;
|
---|
| 323 | }
|
---|
| 324 | return 0;
|
---|
| 325 | }
|
---|
| 326 | ## isLeapYear() ##
|
---|
| 327 |
|
---|
| 328 | 1;
|
---|