Queen.pm 16.1 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
  Bio::EnsEMBL::Hive::Queen
Jessica Severin's avatar
Jessica Severin committed
10
11

=head1 SYNOPSIS
12
  The Queen of the Hive based job control system
Jessica Severin's avatar
Jessica Severin committed
13
14

=head1 DESCRIPTION
15
16
17
18
19
20
21
22
23
24
25
26
  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,
27
  gets $worker->batch_size() jobs from the analysis_job table, does its
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
  work, creates the next layer of analysis_job entries by interfacing to
  the DataflowRuleAdaptor to determine the analyses it needs to pass it's
  output data to and creates jobs on the next analysis's database.
  It repeats this cycle until it's lived it's lifetime or until there are no
  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
  or 'compute farm'.  It's job is to query Queens if they need any workers and to
  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
46
47

=head1 CONTACT
48
49
  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
50
51

=head1 APPENDIX
52
53
  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
54
55
56
57
58
59

=cut

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

use strict;
60
use POSIX;
61
62
63
use Bio::EnsEMBL::Utils::Argument;
use Bio::EnsEMBL::Utils::Exception;

Jessica Severin's avatar
Jessica Severin committed
64
65
66
use Bio::EnsEMBL::Hive::Worker;
use Bio::EnsEMBL::DBSQL::BaseAdaptor;
use Sys::Hostname;
67
use Bio::EnsEMBL::Hive::DBSQL::AnalysisCtrlRuleAdaptor;
Jessica Severin's avatar
Jessica Severin committed
68

69

Jessica Severin's avatar
Jessica Severin committed
70
71
72
our @ISA = qw(Bio::EnsEMBL::DBSQL::BaseAdaptor);

#
73
# PUBLIC METHODS
Jessica Severin's avatar
Jessica Severin committed
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
#
################

=head2 create_new_worker

  Arg [1]    : $analysis_id
  Example    :
  Description:
  Returntype : Bio::EnsEMBL::Hive::Worker
  Exceptions :
  Caller     :

=cut

sub create_new_worker {
89
90
  my ($self, @args) = @_;

91
92
  my ($analysis_id, $beekeeper ,$pid) =
     rearrange([qw(analysis_id beekeeper process_id) ], @args);
Jessica Severin's avatar
Jessica Severin committed
93

94
  my $analStatsDBA = $self->db->get_AnalysisStatsAdaptor;
95
  return undef unless($analStatsDBA);
96

97
  return undef if($self->get_hive_current_load() >= 1.5);
98
  
99
100
101
102
103
104
  unless($analysis_id) {
    my ($anal_stats) = @{$analStatsDBA->fetch_by_needed_workers(1)};
    return undef unless($anal_stats);
    $analysis_id = $anal_stats->analysis_id;
    $analStatsDBA->decrement_needed_workers($analysis_id);
  }
105
  
106
  my $analysisStats = $analStatsDBA->fetch_by_analysis_id($analysis_id);
107
108
109
110
111
112
113
114
115
116
  return undef unless($analysisStats);
  $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
117
118
119
  }

  my $host = hostname;
120
  $pid = getppid unless($pid);
121
  $beekeeper = '' unless($beekeeper);
Jessica Severin's avatar
Jessica Severin committed
122
123

  my $sql = "INSERT INTO hive SET born=now(), last_check_in=now()".
124
            ",process_id='$pid' ".
Jessica Severin's avatar
Jessica Severin committed
125
            ",analysis_id='$analysis_id' ".
126
            ",beekeeper='$beekeeper' ".
Jessica Severin's avatar
Jessica Severin committed
127
128
129
130
131
132
133
134
135
136
            ",host='$host'";

  my $sth = $self->prepare($sql);
  $sth->execute();
  my $hive_id = $sth->{'mysql_insertid'};
  $sth->finish;

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

137
  if($worker and $analysisStats) {
138
    $analStatsDBA->update_status($analysis_id, 'WORKING');
139
    $worker->batch_size($analysisStats->batch_size);
Jessica Severin's avatar
Jessica Severin committed
140
141
142
143
144
145
146
147
148
  }
  return $worker;
}


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

  return unless($worker);
149
150
151
152

  # 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
153
154
155
156
157
158
159
160
  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;
161
162

  if($worker->cause_of_death eq "NO_WORK") {
163
    $self->db->get_AnalysisStatsAdaptor->update_status($worker->analysis->dbID, "ALL_CLAIMED");
164
165
166
  }
  if($worker->cause_of_death eq "FATALITY") {
    #print("FATAL DEATH Arrrrgggghhhhhhhh (hive_id=",$worker->hive_id,")\n");
167
    $self->db->get_AnalysisJobAdaptor->reset_dead_jobs_for_worker($worker);
168
  }
Jessica Severin's avatar
Jessica Severin committed
169
170
171
}


172
sub worker_check_in {
Jessica Severin's avatar
Jessica Severin committed
173
174
175
176
  my ($self, $worker) = @_;

  return unless($worker);
  my $sql = "UPDATE hive SET last_check_in=now()";
177
  $sql .= " ,work_done='" . $worker->work_done . "'";
Jessica Severin's avatar
Jessica Severin committed
178
179
180
181
182
183
184
185
  $sql .= " WHERE hive_id='" . $worker->hive_id ."'";

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


186
187
188
189
190
191
192
193
194
195
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
196

197
198
199
200
201
202
203
204
sub update_analysis_stats {
  my $self = shift;

  my $sql = "SELECT analysis.analysis_id, status, count(*) ".
            "FROM analysis_job, analysis ".
            "WHERE analysis_job.analysis_id=analysis.analysis_id ".
            "GROUP BY analysis_job.analysis_id, status";

205
  my $statsDBA = $self->db->get_AnalysisStatsAdaptor;
206
207
208
209
210
211
  my $analysisStats = undef;

  my $sth = $self->prepare($sql);
  $sth->execute();
  while (my ($analysis_id, $status, $count)=$sth->fetchrow_array()) {
    unless(defined($analysisStats) and $analysisStats->analysis_id==$analysis_id) {
212
      $analysisStats->determine_status()->update() if($analysisStats);
213
214
215
216
217

      $analysisStats = $statsDBA->fetch_by_analysis_id($analysis_id);
      $analysisStats->total_job_count(0);
      $analysisStats->unclaimed_job_count(0);
      $analysisStats->done_job_count(0);
218
      $analysisStats->failed_job_count(0);
219
220
221
222
223
224
225
226
227
      $analysisStats->num_required_workers(0);
    }

    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;
228
      $numWorkers=1 if($numWorkers<1);
229
      if($analysisStats->hive_capacity>0 and $numWorkers > $analysisStats->hive_capacity) {
230
231
232
233
        $numWorkers=$analysisStats->hive_capacity;
      }
      $analysisStats->num_required_workers($numWorkers);
    }
234
235
    if($status eq 'DONE') { $analysisStats->done_job_count($count); }
    if($status eq 'FAILED') { $analysisStats->failed_job_count($count); }
236
  }
237
238
239
240
241
242
243
244
245
246
  $analysisStats->determine_status()->update() if($analysisStats);
  $sth->finish;

  $self->adjust_stats_for_living_workers();
}


sub adjust_stats_for_living_workers {
  my $self = shift;

247
  my $statsDBA = $self->db->get_AnalysisStatsAdaptor;
248
249
250
251
252
253
  
  my $sql = "SELECT analysis_id, count(*) FROM hive ".
            "WHERE cause_of_death='' GROUP BY analysis_id";
  my $sth = $self->prepare($sql);
  $sth->execute();
  while (my ($analysis_id, $liveCount)=$sth->fetchrow_array()) {
254

255
256
257
258
259
260
261
262
263
264
265
266
267
    my $analysis_stats = $statsDBA->fetch_by_analysis_id($analysis_id);

    if($analysis_stats->hive_capacity > 0) {
      my $numWorkers = $analysis_stats->num_required_workers;

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

      $analysis_stats->num_required_workers($numWorkers);
      $analysis_stats->update;
    }
  }
268
269
270
271
  $sth->finish;
}


272
273
274
sub get_hive_current_load {
  my $self = shift;
  my $sql = "SELECT sum(1/analysis_stats.hive_capacity) FROM hive, analysis_stats ".
275
276
            "WHERE hive.analysis_id=analysis_stats.analysis_id and cause_of_death ='' ".
            "AND analysis_stats.hive_capacity>0";
277
278
279
280
281
282
283
284
285
286
  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;
}


287
sub get_num_needed_workers {
288
289
  my $self = shift;

290
291
  my $neededAnals = $self->db->get_AnalysisStatsAdaptor->fetch_by_needed_workers();
  return 0 unless($neededAnals);
292

293
294
295
296
297
  my $availableLoad = 1.0 - $self->get_hive_current_load();
  return 0 if($availableLoad <0.0);

  my $numWorkers = 0;
  foreach my $analysis_stats (@{$neededAnals}) {
298
299
    #$analysis_stats->print_stats();

300
301
302
303
304
305
    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)) {
306
307
308
309
310
      $numWorkers += $analysis_stats->num_required_workers;
      $availableLoad -= $thisLoad;
      printf("  %d (%1.9f) ", $numWorkers, $availableLoad);
      $analysis_stats->print_stats();
    } else {
311
      my $workerCount = POSIX::ceil($availableLoad * $analysis_stats->hive_capacity                     );
312
313
314
315
316
      $numWorkers += $workerCount;
      $availableLoad -=  $workerCount * (1/$analysis_stats->hive_capacity);
      printf("  %d (%1.9f) use only %d ", $numWorkers, $availableLoad, $workerCount);
      $analysis_stats->print_stats();
      last;
317
    }
318
    last if($availableLoad <= 0.0);
319
  }
320

321
  printf("need $numWorkers workers (availLoad=%1.5f)\n", $availableLoad);
322
  return $numWorkers;
323
324
325
}


326
327
328
329
sub check_blocking_control_rules
{
  my $self = shift;

330
  my $analysisStatsList = $self->db->get_AnalysisStatsAdaptor->fetch_all();
331
332
333
334
  foreach my $stats (@{$analysisStatsList}) {
    #print("BLOCKED analysis ");  $stats->print_stats;
    my $ctrlRules = $self->db->get_AnalysisCtrlRuleAdaptor->
                    fetch_by_ctrled_analysis_id($stats->analysis_id);
335
336
337
338
339
340
341
342
343
344
345
346
347
    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;
      }
348

349
      if($allRulesDone) {
350
351
352
353
        if($stats->status eq 'BLOCKED') {
          #print("  UNBLOCK analysis : all conditions met\n");
          $stats->adaptor->update_status($stats->analysis_id, 'READY');
        }
354
355
356
357
      } else {
        #print("  RE-BLOCK analysis : all conditions met\n");
        $stats->adaptor->update_status($stats->analysis_id, 'BLOCKED');
      }
358
359
360
361
    }
  }
}

362
363
364
365
366
sub print_hive_status
{
  my $self = shift;

  my $allStats = $self->db->get_AnalysisStatsAdaptor->fetch_all();
367
 
368
369
370
  foreach my $analysis_stats (@{$allStats}) {
    $analysis_stats->print_stats;
  }
371
372
373
374
375
376
377
378
379
380
381
382

  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;  
383
}
384

385
386


Jessica Severin's avatar
Jessica Severin committed
387
388
389
390
391
392
393
394
395
#
# INTERNAL METHODS
#
###################

=head2 _fetch_by_hive_id

  Arg [1]    : int $id
               the unique database identifier for the feature to be obtained
396
  Example    : $feat = $queen->fetch_by_dbID(1234);
Jessica Severin's avatar
Jessica Severin committed
397
398
399
400
401
402
403
404
405
406
407
408
  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) {
409
    throw("fetch_by_dbID must have an id");
Jessica Severin's avatar
Jessica Severin committed
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
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
  }

  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
500
             h.beekeeper
Jessica Severin's avatar
Jessica Severin committed
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
             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'});
524
    $worker->beekeeper($column{'beekeeper'});
Jessica Severin's avatar
Jessica Severin committed
525
526
527
528
529
530
531
    $worker->host($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'});
532
    $worker->queen($self);
Jessica Severin's avatar
Jessica Severin committed
533
534
    $worker->db($self->db);

535
536
    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
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
    }

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

  return \@workers
}

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

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


1;