Queen.pm 17.7 KB
Newer Older
Jessica Severin's avatar
Jessica Severin committed
1
2
3
4
5
6
7
8
#
# You may distribute this module under the same terms as perl itself
#
# POD documentation - main docs before the code

=pod 

=head1 NAME
9

10
  Bio::EnsEMBL::Hive::Queen
Jessica Severin's avatar
Jessica Severin committed
11
12

=head1 SYNOPSIS
13

14
  The Queen of the Hive based job control system
Jessica Severin's avatar
Jessica Severin committed
15
16

=head1 DESCRIPTION
17

18
19
20
21
22
23
24
25
26
27
28
29
  The Queen of the Hive based job control system is responsible to 'birthing' the
  correct number of workers of the right type so that they can find jobs to do.
  It will also free up jobs of Workers that died unexpectantly so that other workers
  can claim them to do.

  Hive based processing is a concept based on a more controlled version
  of an autonomous agent type system.  Each worker is not told what to do
  (like a centralized control system - like the current pipeline system)
  but rather queries a central database for jobs (give me jobs).

  Each worker is linked to an analysis_id, registers its self on creation
  into the Hive, creates a RunnableDB instance of the Analysis->module,
30
  gets $worker->batch_size() jobs from the analysis_job table, does its
31
  work, creates the next layer of analysis_job entries by interfacing to
32
33
34
  the DataflowRuleAdaptor to determine the analyses it needs to pass its
  output data to and creates jobs on the next analysis database.
  It repeats this cycle until it has lived its lifetime or until there are no
35
36
37
38
39
40
41
42
43
44
  more jobs left.
  The lifetime limit is just a safety limit to prevent these from 'infecting'
  a system.

  The Queens job is to simply birth Workers of the correct analysis_id to get the
  work down.  The only other thing the Queen does is free up jobs that were
  claimed by Workers that died unexpectantly so that other workers can take
  over the work.

  The Beekeeper is in charge of interfacing between the Queen and a compute resource
45
  or 'compute farm'.  Its job is to query Queens if they need any workers and to
46
47
48
  send the requested number of workers to open machines via the runWorker.pl script.
  It is also responsible for interfacing with the Queen to identify worker which died
  unexpectantly.
Jessica Severin's avatar
Jessica Severin committed
49
50

=head1 CONTACT
51

52
53
  Contact Jessica Severin on EnsEMBL::Hive implemetation/design detail: jessica@ebi.ac.uk
  Contact Ewan Birney on EnsEMBL in general: birney@sanger.ac.uk
Jessica Severin's avatar
Jessica Severin committed
54
55

=head1 APPENDIX
56

57
58
  The rest of the documentation details each of the object methods. 
  Internal methods are usually preceded with a _
Jessica Severin's avatar
Jessica Severin committed
59
60
61
62
63
64

=cut

package Bio::EnsEMBL::Hive::Queen;

use strict;
65
use POSIX;
66
67
68
use Bio::EnsEMBL::Utils::Argument;
use Bio::EnsEMBL::Utils::Exception;

Jessica Severin's avatar
Jessica Severin committed
69
70
71
use Bio::EnsEMBL::Hive::Worker;
use Bio::EnsEMBL::DBSQL::BaseAdaptor;
use Sys::Hostname;
72
use Bio::EnsEMBL::Hive::DBSQL::AnalysisCtrlRuleAdaptor;
Jessica Severin's avatar
Jessica Severin committed
73

74

Jessica Severin's avatar
Jessica Severin committed
75
76
77
our @ISA = qw(Bio::EnsEMBL::DBSQL::BaseAdaptor);

#
78
# PUBLIC METHODS
Jessica Severin's avatar
Jessica Severin committed
79
80
81
82
83
#
################

=head2 create_new_worker

84
  Arg [1]    : $analysis_id (optional)
Jessica Severin's avatar
Jessica Severin committed
85
  Example    :
86
87
88
89
90
91
  Description: If analysis_id is specified it will try to create a worker based
               on that analysis.  If not specified the queen will analyze the hive
               and pick the analysis that has the most amount of work to be done.
               It creates an entry in the hive table, and returns a Worker object 
               based on that insert.  This guarantees that each worker registered
               in this queens hive is properly registered.
Jessica Severin's avatar
Jessica Severin committed
92
93
94
95
96
97
98
  Returntype : Bio::EnsEMBL::Hive::Worker
  Exceptions :
  Caller     :

=cut

sub create_new_worker {
99
100
  my ($self, @args) = @_;

101
102
  my ($analysis_id, $beekeeper ,$pid) =
     rearrange([qw(analysis_id beekeeper process_id) ], @args);
Jessica Severin's avatar
Jessica Severin committed
103

104
  my $analStatsDBA = $self->db->get_AnalysisStatsAdaptor;
105
  return undef unless($analStatsDBA);
106

107
  return undef if($self->get_hive_current_load() >= 1.5);
108
  
109
110
111
112
113
  my $analysisStats;
  if($analysis_id) {
    $analysisStats = $analStatsDBA->fetch_by_analysis_id($analysis_id);
  } else {
    ($analysisStats) = @{$analStatsDBA->fetch_by_needed_workers(1)};
114
  }
115
116
  
  return undef unless($analysisStats);
117
  $analStatsDBA->decrement_needed_workers($analysisStats->analysis_id);
118
119
120
121
122
123
124
125
126
  $analysisStats->print_stats;
  
  if($analysisStats->status eq 'BLOCKED') {
    print("Analysis is BLOCKED, can't create workers\n");
    return undef;
  }
  if($analysisStats->status eq 'DONE') {
    print("Analysis is DONE, don't need to create workers\n");
    return undef;
Jessica Severin's avatar
Jessica Severin committed
127
128
129
  }

  my $host = hostname;
130
  $pid = getppid unless($pid);
131
  $beekeeper = '' unless($beekeeper);
Jessica Severin's avatar
Jessica Severin committed
132

133
  my $sql = "INSERT INTO hive SET born=now(), last_check_in=now(), " .
134
            "process_id=?, analysis_id=?, beekeeper=?, host=?";
Jessica Severin's avatar
Jessica Severin committed
135
136

  my $sth = $self->prepare($sql);
137
  $sth->execute($pid, $analysisStats->analysis_id, $beekeeper, $host);
Jessica Severin's avatar
Jessica Severin committed
138
139
140
141
142
143
  my $hive_id = $sth->{'mysql_insertid'};
  $sth->finish;

  my $worker = $self->_fetch_by_hive_id($hive_id);
  $worker=undef unless($worker and $worker->analysis);

144
  if($worker and $analysisStats) {
145
    $analysisStats->update_status('WORKING');
Jessica Severin's avatar
Jessica Severin committed
146
147
148
149
150
151
152
153
154
  }
  return $worker;
}


sub register_worker_death {
  my ($self, $worker) = @_;

  return unless($worker);
155
156
157
158

  # if called without a defined cause_of_death, assume catastrophic failure
  $worker->cause_of_death('FATALITY') unless(defined($worker->cause_of_death));
  
Jessica Severin's avatar
Jessica Severin committed
159
160
161
162
163
164
165
166
  my $sql = "UPDATE hive SET died=now(), last_check_in=now()";
  $sql .= " ,work_done='" . $worker->work_done . "'";
  $sql .= " ,cause_of_death='". $worker->cause_of_death ."'";
  $sql .= " WHERE hive_id='" . $worker->hive_id ."'";

  my $sth = $self->prepare($sql);
  $sth->execute();
  $sth->finish;
167
168

  if($worker->cause_of_death eq "NO_WORK") {
169
    $self->db->get_AnalysisStatsAdaptor->update_status($worker->analysis->dbID, "ALL_CLAIMED");
170
171
172
  }
  if($worker->cause_of_death eq "FATALITY") {
    #print("FATAL DEATH Arrrrgggghhhhhhhh (hive_id=",$worker->hive_id,")\n");
173
    $self->db->get_AnalysisJobAdaptor->reset_dead_jobs_for_worker($worker);
174
  }
175
176
177
  
  # always re-sync the analysis_stats when a worker dies
  $self->synchronize_AnalysisStats($worker->analysis->stats);
Jessica Severin's avatar
Jessica Severin committed
178
179
180
}


181
sub worker_check_in {
Jessica Severin's avatar
Jessica Severin committed
182
183
184
185
  my ($self, $worker) = @_;

  return unless($worker);
  my $sql = "UPDATE hive SET last_check_in=now()";
186
  $sql .= " ,work_done='" . $worker->work_done . "'";
Jessica Severin's avatar
Jessica Severin committed
187
188
189
190
191
  $sql .= " WHERE hive_id='" . $worker->hive_id ."'";

  my $sth = $self->prepare($sql);
  $sth->execute();
  $sth->finish;
192
193
194
195
196
197
198
  
  # if analysis_stats for this worker's analysis is more than a minutes
  # out of date, then re-synchronize it
  my $stats = $worker->analysis->stats;
  if($stats->seconds_since_last_update >= 60) {
    $self->synchronize_AnalysisStats($stats);
  }
Jessica Severin's avatar
Jessica Severin committed
199
200
201
}


202
203
204
205
206
207
208
209
210
211
sub fetch_overdue_workers {
  my ($self,$overdue_secs) = @_;

  $overdue_secs = 3600 unless(defined($overdue_secs));

  my $constraint = "h.cause_of_death='' ".
                   "AND (UNIX_TIMESTAMP()-UNIX_TIMESTAMP(h.last_check_in))>$overdue_secs";
  return $self->_generic_fetch($constraint);
}

Jessica Severin's avatar
Jessica Severin committed
212

213
214
215
216
217
218
219
220
221
222
223
224
225
=head2 synchronize_hive

  Example    : $queen->synchronize_hive();
  Description: Runs through all analyses in the system and synchronizes
              the analysis_stats summary with the states in the analysis_job 
              and hive tables.  Then follows by checking all the blocking rules
              and blocks/unblocks analyses as needed.
  Exceptions : none
  Caller     : general

=cut

sub synchronize_hive {
226
227
  my $self = shift;

228
  my $start_time = time();
229

230
231
232
233
234
235
  my $allAnalysis = $self->db->get_AnalysisAdaptor->fetch_all;
  print("analyze ", scalar(@$allAnalysis), "\n");
  foreach my $analysis (@$allAnalysis) {
    my $stats = $analysis->stats;
    $self->synchronize_AnalysisStats($stats);
  }
236
  foreach my $analysis (@$allAnalysis) {
237
238
239
240
    $self->check_blocking_control_rules_for_AnalysisStats($analysis->stats);
  }
  print((time() - $start_time), " secs to synchronize_hive\n");
}
241

242

243
=head2 synchronize_AnalysisStats
244

245
246
247
248
249
250
251
252
  Arg [1]    : Bio::EnsEMBL::Hive::AnalysisStats object
  Example    : $self->synchronize($analysisStats);
  Description: Queries the analysis_job and hive tables to get summary counts
               and rebuilds the AnalysisStats object.  Then updates the
               analysis_stats table with the new summary info
  Returntype : newly synced Bio::EnsEMBL::Hive::AnalysisStats object
  Exceptions : none
  Caller     : general
253

254
=cut
255

256
sub synchronize_AnalysisStats {
257
  my $self = shift;
258
  my $analysisStats = shift;
259

260
  return $analysisStats unless($analysisStats);
261
262
263
  return $analysisStats unless($analysisStats->analysis_id);
  return $analysisStats if($analysisStats->status eq 'BLOCKED');
  return $analysisStats if($analysisStats->status eq 'SYNCHING');
264
  
265
266
  $analysisStats->update_status('SYNCHING');
  
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
  $analysisStats->total_job_count(0);
  $analysisStats->unclaimed_job_count(0);
  $analysisStats->done_job_count(0);
  $analysisStats->failed_job_count(0);
  $analysisStats->num_required_workers(0);

  my $sql = "SELECT status, count(*) FROM analysis_job ".
            "WHERE analysis_id=? GROUP BY status";
  my $sth = $self->prepare($sql);
  $sth->execute($analysisStats->analysis_id);

  while (my ($status, $count)=$sth->fetchrow_array()) {

    my $total = $analysisStats->total_job_count();
    $analysisStats->total_job_count($total + $count);

    if($status eq 'READY') {
      $analysisStats->unclaimed_job_count($count);
      my $numWorkers = $count/$analysisStats->batch_size;
      $numWorkers=1 if($numWorkers<1);
      if($analysisStats->hive_capacity>0 and $numWorkers > $analysisStats->hive_capacity) {
        $numWorkers=$analysisStats->hive_capacity;
      }
      $analysisStats->num_required_workers($numWorkers);
    }
    if($status eq 'DONE') { $analysisStats->done_job_count($count); }
    if($status eq 'FAILED') { $analysisStats->failed_job_count($count); }
  }
  $sth->finish;
  $analysisStats->determine_status();

298
299
300
301
  #
  # adjust_stats_for_living_workers
  #
  
302
303
304
305
306
307
308
309
310
311
312
313
314
315
  if($analysisStats->hive_capacity > 0) {
    my $sql = "SELECT count(*) FROM hive WHERE cause_of_death='' and analysis_id=?";
    $sth = $self->prepare($sql);
    $sth->execute($analysisStats->analysis_id);
    my($liveCount)=$sth->fetchrow_array();
    $sth->finish;

    my $numWorkers = $analysisStats->num_required_workers;

    my $capacityAdjust = ($numWorkers + $liveCount) - $analysisStats->hive_capacity;
    $numWorkers -= $capacityAdjust if($capacityAdjust > 0);
    $numWorkers=0 if($numWorkers<0);

    $analysisStats->num_required_workers($numWorkers);
316
317
318
319
  }
  
  $analysisStats->update;
  
320
  return $analysisStats;
321
}
322
323


324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
sub check_blocking_control_rules_for_AnalysisStats
{
  my $self = shift;
  my $stats = shift;
  
  return unless($stats);

  #print("check ctrl on analysis ");  $stats->print_stats;
  my $ctrlRules = $self->db->get_AnalysisCtrlRuleAdaptor->
                  fetch_by_ctrled_analysis_id($stats->analysis_id);
  my $allRulesDone = 1;
  if(scalar @$ctrlRules > 0) {
    #print("HAS blocking_ctrl_rules to check\n");
    foreach my $ctrlrule (@{$ctrlRules}) {
      #use this method because the condition_analysis objects can be
      #network distributed to a different database so use it's adaptor to get
      #the AnalysisStats object
      #$ctrlrule->print_rule;
      my $condAnalysis = $ctrlrule->condition_analysis;
      my $condStats = $condAnalysis->stats if($condAnalysis);
      $allRulesDone = 0 unless($condStats and $condStats->status eq 'DONE');
      #print("  "); $condStats->print_stats;
    }
347

348
349
350
351
352
353
354
355
    if($allRulesDone) {
      if($stats->status eq 'BLOCKED') {
        #print("  UNBLOCK analysis : all conditions met\n");
        $stats->update_status('READY');
      }
    } else {
      #print("  RE-BLOCK analysis : some conditions failed\n");
      $stats->update_status('BLOCKED');
356
357
    }
  }
358
359
360
}


361
362
363
sub get_hive_current_load {
  my $self = shift;
  my $sql = "SELECT sum(1/analysis_stats.hive_capacity) FROM hive, analysis_stats ".
364
365
            "WHERE hive.analysis_id=analysis_stats.analysis_id and cause_of_death ='' ".
            "AND analysis_stats.hive_capacity>0";
366
367
368
369
370
371
372
373
374
375
  my $sth = $self->prepare($sql);
  $sth->execute();
  (my $load)=$sth->fetchrow_array();
  $sth->finish;
  $load=0 unless($load);
  print("current hive load = $load\n");
  return $load;
}


376
sub get_num_needed_workers {
377
378
  my $self = shift;

379
380
  my $neededAnals = $self->db->get_AnalysisStatsAdaptor->fetch_by_needed_workers();
  return 0 unless($neededAnals);
381

382
383
384
385
386
  my $availableLoad = 1.0 - $self->get_hive_current_load();
  return 0 if($availableLoad <0.0);

  my $numWorkers = 0;
  foreach my $analysis_stats (@{$neededAnals}) {
387
388
    #$analysis_stats->print_stats();

389
390
391
392
393
394
    my $thisLoad = 0.0;
    if($analysis_stats->hive_capacity>0) {
      $thisLoad = $analysis_stats->num_required_workers * (1/$analysis_stats->hive_capacity);
    }

    if(($analysis_stats->hive_capacity<=0) or ($thisLoad < $availableLoad)) {
395
396
397
398
399
      $numWorkers += $analysis_stats->num_required_workers;
      $availableLoad -= $thisLoad;
      printf("  %d (%1.9f) ", $numWorkers, $availableLoad);
      $analysis_stats->print_stats();
    } else {
400
      my $workerCount = POSIX::ceil($availableLoad * $analysis_stats->hive_capacity                     );
401
402
403
404
405
      $numWorkers += $workerCount;
      $availableLoad -=  $workerCount * (1/$analysis_stats->hive_capacity);
      printf("  %d (%1.9f) use only %d ", $numWorkers, $availableLoad, $workerCount);
      $analysis_stats->print_stats();
      last;
406
    }
407
    last if($availableLoad <= 0.0);
408
  }
409

410
  printf("need $numWorkers workers (availLoad=%1.5f)\n", $availableLoad);
411
  return $numWorkers;
412
413
414
}


415
416
417
418
419
sub print_hive_status
{
  my $self = shift;

  my $allStats = $self->db->get_AnalysisStatsAdaptor->fetch_all();
420
 
421
422
423
  foreach my $analysis_stats (@{$allStats}) {
    $analysis_stats->print_stats;
  }
424
425
426
427
428
429
430
431
432
433
434
435

  print("HIVE LIVE WORKERS====\n");
  my $sql = "select logic_name, count(*) from hive, analysis ".
            "where hive.analysis_id=analysis.analysis_id and hive.cause_of_death='' ".
            "group by hive.analysis_id";
  my $sth = $self->prepare($sql);
  $sth->execute();
  while((my $logic_name, my $count)=$sth->fetchrow_array()) {
    printf("%20s : %d workers\n", $logic_name, $count);
  }
  print("=====================\n");
  $sth->finish;  
436
}
437

438
439


Jessica Severin's avatar
Jessica Severin committed
440
441
442
443
444
445
446
447
448
#
# INTERNAL METHODS
#
###################

=head2 _fetch_by_hive_id

  Arg [1]    : int $id
               the unique database identifier for the feature to be obtained
449
  Example    : $feat = $queen->fetch_by_dbID(1234);
Jessica Severin's avatar
Jessica Severin committed
450
451
452
453
454
455
456
457
458
459
460
461
  Description: Returns the feature created from the database defined by the
               the id $id.
  Returntype : Bio::EnsEMBL::Hive::Worker
  Exceptions : thrown if $id is not defined
  Caller     : general

=cut

sub _fetch_by_hive_id{
  my ($self,$id) = @_;

  unless(defined $id) {
462
    throw("fetch_by_dbID must have an id");
Jessica Severin's avatar
Jessica Severin committed
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
  }

  my @tabs = $self->_tables;

  my $constraint = "h.hive_id = $id";

  #return first element of _generic_fetch list
  my ($obj) = @{$self->_generic_fetch($constraint)};
  return $obj;
}

=head2 _generic_fetch

  Arg [1]    : (optional) string $constraint
               An SQL query constraint (i.e. part of the WHERE clause)
  Arg [2]    : (optional) string $logic_name
               the logic_name of the analysis of the features to obtain
  Example    : $fts = $a->_generic_fetch('contig_id in (1234, 1235)', 'Swall');
  Description: Performs a database fetch and returns feature objects in
               contig coordinates.
  Returntype : listref of Bio::EnsEMBL::SeqFeature in contig coordinates
  Exceptions : none
  Caller     : BaseFeatureAdaptor, ProxyDnaAlignFeatureAdaptor::_generic_fetch

=cut

sub _generic_fetch {
  my ($self, $constraint, $join) = @_;

  my @tables = $self->_tables;
  my $columns = join(', ', $self->_columns());

  if ($join) {
    foreach my $single_join (@{$join}) {
      my ($tablename, $condition, $extra_columns) = @{$single_join};
      if ($tablename && $condition) {
        push @tables, $tablename;

        if($constraint) {
          $constraint .= " AND $condition";
        } else {
          $constraint = " $condition";
        }
      }
      if ($extra_columns) {
        $columns .= ", " . join(', ', @{$extra_columns});
      }
    }
  }

  #construct a nice table string like 'table1 t1, table2 t2'
  my $tablenames = join(', ', map({ join(' ', @$_) } @tables));

  my $sql = "SELECT $columns FROM $tablenames";

  my $default_where = $self->_default_where_clause;
  my $final_clause = $self->_final_clause;

  #append a where clause if it was defined
  if($constraint) {
    $sql .= " WHERE $constraint ";
    if($default_where) {
      $sql .= " AND $default_where ";
    }
  } elsif($default_where) {
    $sql .= " WHERE $default_where ";
  }

  #append additional clauses which may have been defined
  $sql .= " $final_clause";

  my $sth = $self->prepare($sql);
  $sth->execute;

#  print STDERR $sql,"\n";

  return $self->_objs_from_sth($sth);
}

sub _tables {
  my $self = shift;

  return (['hive', 'h']);
}

sub _columns {
  my $self = shift;

  return qw (h.hive_id
             h.analysis_id
553
             h.beekeeper
Jessica Severin's avatar
Jessica Severin committed
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
             h.host
             h.process_id
             h.work_done
             h.born
             h.last_check_in
             h.died
             h.cause_of_death
            );
}

sub _objs_from_sth {
  my ($self, $sth) = @_;

  my %column;
  $sth->bind_columns( \( @column{ @{$sth->{NAME_lc} } } ));

  my @workers = ();

  while ($sth->fetch()) {
    my $worker = new Bio::EnsEMBL::Hive::Worker;
    $worker->init;

    $worker->hive_id($column{'hive_id'});
577
    $worker->beekeeper($column{'beekeeper'});
Jessica Severin's avatar
Jessica Severin committed
578
579
580
581
582
583
584
    $worker->host($column{'host'});
    $worker->process_id($column{'process_id'});
    $worker->work_done($column{'work_done'});
    $worker->born($column{'born'});
    $worker->last_check_in($column{'last_check_in'});
    $worker->died($column{'died'});
    $worker->cause_of_death($column{'cause_of_death'});
585
    $worker->queen($self);
Jessica Severin's avatar
Jessica Severin committed
586
587
    $worker->db($self->db);

588
589
    if($column{'analysis_id'} and $self->db->get_AnalysisAdaptor) {
      $worker->analysis($self->db->get_AnalysisAdaptor->fetch_by_dbID($column{'analysis_id'}));
Jessica Severin's avatar
Jessica Severin committed
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
    }

    push @workers, $worker;
  }
  $sth->finish;

  return \@workers
}

sub _default_where_clause {
  my $self = shift;
  return '';
}

sub _final_clause {
  my $self = shift;
  return '';
}


1;