Queen.pm 29.8 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
our @ISA = qw(Bio::EnsEMBL::DBSQL::BaseAdaptor);

77
############################
Jessica Severin's avatar
Jessica Severin committed
78
#
79
# PUBLIC API for Workers
Jessica Severin's avatar
Jessica Severin committed
80
#
81
############################
Jessica Severin's avatar
Jessica Severin committed
82 83 84

=head2 create_new_worker

85
  Arg [1]    : $analysis_id (optional)
Jessica Severin's avatar
Jessica Severin committed
86
  Example    :
87 88 89 90 91 92
  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
93 94 95 96 97 98 99
  Returntype : Bio::EnsEMBL::Hive::Worker
  Exceptions :
  Caller     :

=cut

sub create_new_worker {
100 101
  my ($self, @args) = @_;

102 103
  my ($analysis_id, $beekeeper ,$pid, $job, $no_write) =
     rearrange([qw(analysis_id beekeeper process_id job no_write) ], @args);
Jessica Severin's avatar
Jessica Severin committed
104

105
  my $analStatsDBA = $self->db->get_AnalysisStatsAdaptor;
106
  return undef unless($analStatsDBA);
107

108
  $analysis_id = $job->analysis_id if(defined($job));
109
  
110
  my $analysisStats;
111

112 113
  if($analysis_id) {
    $analysisStats = $analStatsDBA->fetch_by_analysis_id($analysis_id);
114
    $self->safe_synchronize_AnalysisStats($analysisStats);
115
    #return undef unless(($analysisStats->status ne 'BLOCKED') and ($analysisStats->num_required_workers > 0));
116
  } else {
117
    $analysisStats = $self->_pick_best_analysis_for_new_worker;
118
  }
119
  return undef unless($analysisStats);
120

121 122
  unless($job) {
    #go into autonomous mode
123
    return undef if($self->get_hive_current_load() >= 1.1);
124
    
125 126
    $analStatsDBA->decrease_needed_workers($analysisStats->analysis_id);
    $analStatsDBA->increase_running_workers($analysisStats->analysis_id);
127 128 129 130 131 132 133 134 135 136
    $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;
    }
137
  }
138
  
Jessica Severin's avatar
Jessica Severin committed
139
  my $host = hostname;
140
  $pid = $$ unless($pid);
141
  $beekeeper = '' unless($beekeeper);
Jessica Severin's avatar
Jessica Severin committed
142

143 144 145
  my $sql = q{INSERT INTO hive 
              (born, last_check_in, process_id, analysis_id, beekeeper, host)
              VALUES (NOW(), NOW(), ?,?,?,?)};
Jessica Severin's avatar
Jessica Severin committed
146 147

  my $sth = $self->prepare($sql);
148
  $sth->execute($pid, $analysisStats->analysis_id, $beekeeper, $host);
Jessica Severin's avatar
Jessica Severin committed
149 150 151 152 153 154
  my $hive_id = $sth->{'mysql_insertid'};
  $sth->finish;

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

155
  if($worker and $analysisStats) {
156
    $analysisStats->update_status('WORKING');
Jessica Severin's avatar
Jessica Severin committed
157
  }
158 159 160 161
  
  $worker->_specific_job($job) if(defined($job));
  $worker->execute_writes(0) if($no_write);
  
Jessica Severin's avatar
Jessica Severin committed
162 163 164 165 166 167 168 169
  return $worker;
}


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

  return unless($worker);
170 171
  # if called without a defined cause_of_death, assume catastrophic failure
  $worker->cause_of_death('FATALITY') unless(defined($worker->cause_of_death));
172 173 174 175 176
  unless ($worker->cause_of_death() eq "HIVE_OVERLOAD") {
    ## HIVE_OVERLOAD occurs after a successful update of the analysis_stats teble. (c.f. Worker.pm)
    $worker->analysis->stats->adaptor->decrease_running_workers($worker->analysis->stats->analysis_id);
  }

Jessica Severin's avatar
Jessica Severin committed
177
  my $sql = "UPDATE hive SET died=now(), last_check_in=now()";
178
  $sql .= " ,status='DEAD'";
Jessica Severin's avatar
Jessica Severin committed
179 180 181 182 183 184 185
  $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;
186 187

  if($worker->cause_of_death eq "NO_WORK") {
188
    $self->db->get_AnalysisStatsAdaptor->update_status($worker->analysis->dbID, "ALL_CLAIMED");
189 190 191
  }
  if($worker->cause_of_death eq "FATALITY") {
    #print("FATAL DEATH Arrrrgggghhhhhhhh (hive_id=",$worker->hive_id,")\n");
192
    $self->db->get_AnalysisJobAdaptor->reset_dead_jobs_for_worker($worker);
193
  }
194
  
195
  # re-sync the analysis_stats when a worker dies as part of dynamic sync system
196
  if($self->safe_synchronize_AnalysisStats($worker->analysis->stats)->status ne 'DONE') {
197 198
    # since I'm dying I should make sure there is someone to take my place after I'm gone ...
    # above synch still sees me as a 'living worker' so I need to compensate for that
199
    $self->db->get_AnalysisStatsAdaptor->increase_needed_workers($worker->analysis->dbID);
200 201
  }

Jessica Severin's avatar
Jessica Severin committed
202 203 204
}


205
sub worker_check_in {
Jessica Severin's avatar
Jessica Severin committed
206 207 208 209
  my ($self, $worker) = @_;

  return unless($worker);
  my $sql = "UPDATE hive SET last_check_in=now()";
210
  $sql .= " ,work_done='" . $worker->work_done . "'";
Jessica Severin's avatar
Jessica Severin committed
211 212 213 214 215
  $sql .= " WHERE hive_id='" . $worker->hive_id ."'";

  my $sth = $self->prepare($sql);
  $sth->execute();
  $sth->finish;
216
  
217
  $self->safe_synchronize_AnalysisStats($worker->analysis->stats);
Jessica Severin's avatar
Jessica Severin committed
218 219 220
}


221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
=head2 worker_grab_jobs

  Arg [1]           : Bio::EnsEMBL::Hive::Worker object $worker
  Example: 
    my $jobs  = $queen->worker_grab_jobs();
  Description: 
    For the specified worker, it will search available jobs, 
    and using the workers requested batch_size, claim/fetch that
    number of jobs, and then return them.
  Returntype : 
    reference to array of Bio::EnsEMBL::Hive::AnalysisJob objects
  Exceptions :
  Caller     :

=cut

sub worker_grab_jobs {
  my $self            = shift;
  my $worker          = shift;
  
  my $jobDBA = $self->db->get_AnalysisJobAdaptor;
  my $claim = $jobDBA->claim_jobs_for_worker($worker);
  my $jobs = $jobDBA->fetch_by_claim_analysis($claim, $worker->analysis->dbID);
  return $jobs;
}

247
=head2 reset_and_fetch_job_by_dbID
248 249 250

  Arg [1]: int $analysis_job_id
  Example: 
251
    my $job = $queen->reset_and_fetch_job_by_dbID($analysis_job_id);
252 253
  Description: 
    For the specified analysis_job_id it will fetch just that job, 
254 255 256
    reset it completely as if it has never run, and return it.  
    Specifying a specific job bypasses the safety checks, 
    thus multiple workers could be running the 
257 258
    same job simultaneously (use only for debugging).
  Returntype : 
259
    Bio::EnsEMBL::Hive::AnalysisJob object
260
  Exceptions :
261
  Caller     : beekeepers, runWorker.pl scripts
262 263 264

=cut

265 266
sub reset_and_fetch_job_by_dbID {
  my $self = shift;
267
  my $analysis_job_id = shift;
268
  
269
  my $jobDBA = $self->db->get_AnalysisJobAdaptor;
270 271 272 273 274
  $jobDBA->reset_job_by_dbID($analysis_job_id); 

  my $job = $jobDBA->fetch_by_dbID($analysis_job_id); 
  my $stats = $self->db->get_AnalysisStatsAdaptor->fetch_by_analysis_id($job->analysis_id);
  $self->synchronize_AnalysisStats($stats);
275
  
276 277 278 279 280 281 282 283 284 285 286 287
  return $job;
}


sub worker_reclaim_job {
  my $self   = shift;
  my $worker = shift;
  my $job    = shift;

  return undef unless($job and $worker);
  $job->hive_id($worker->hive_id);
  $self->db->get_AnalysisJobAdaptor->reclaim_job($job);
288 289 290 291 292 293 294 295 296 297 298 299
  return $job;
}


sub worker_register_job_done {
  my $self = shift;
  my $worker = shift;
  my $job = shift;
  
  return unless($job);
  return unless($job->dbID and $job->adaptor and $job->hive_id);
  return unless($worker and $worker->analysis and $worker->analysis->dbID);
300
  
301 302 303 304 305 306 307 308 309
  $job->update_status('DONE');
}


sub flow_output_job {
  my $self = shift;
  my $job = shift;
  
  return unless($job);
310
  my $create_blocked_job = 0;
311
  $create_blocked_job = 1 if($job->status and ($job->status eq 'BLOCKED'));
312
  
313
  my @output_jobs;
314 315
  my $rules = $self->db->get_DataflowRuleAdaptor->fetch_from_analysis_job($job);
  foreach my $rule (@{$rules}) {
316
    my $job_id = Bio::EnsEMBL::Hive::DBSQL::AnalysisJobAdaptor->CreateNewJob (
317 318 319
        -input_id       => $job->input_id,
        -analysis       => $rule->to_analysis,
        -input_job_id   => $job->dbID,
320
        -block          => $create_blocked_job
321
    );
322 323 324
    my $job_url =  $rule->to_analysis->adaptor->db->dbc->url;
    $job_url .= "/analysis_job?dbID=" . $job_id;
    push @output_jobs, $job_url;
325
  }
326
  return \@output_jobs;
327 328 329 330 331 332 333 334 335
}


######################################
#
# Public API interface for beekeeper
#
######################################

336 337 338 339 340 341 342 343 344 345
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);
}

346 347 348 349 350 351
sub fetch_failed_workers {
  my $self = shift;
  my $constraint = "h.cause_of_death='FATALITY' ";
  return $self->_generic_fetch($constraint);
}

Jessica Severin's avatar
Jessica Severin committed
352

353 354 355 356 357 358 359 360 361 362 363 364 365
=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 {
366 367
  my $self = shift;

368
  my $start_time = time();
369

370 371 372 373 374 375
  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);
  }
376
  foreach my $analysis (@$allAnalysis) {
377 378 379 380
    $self->check_blocking_control_rules_for_AnalysisStats($analysis->stats);
  }
  print((time() - $start_time), " secs to synchronize_hive\n");
}
381

382

383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404
=head2 safe_synchronize_AnalysisStats

  Arg [1]    : Bio::EnsEMBL::Hive::AnalysisStats object
  Example    : $self->synchronize($analysisStats);
  Description: Prewrapper around synchronize_AnalysisStats that does
               checks and grabs sync_lock before proceeding with sync.
               Used by distributed worker sync system to avoid contention.
  Exceptions : none
  Caller     : general

=cut

sub safe_synchronize_AnalysisStats {
  my $self = shift;
  my $stats = shift;

  return $stats unless($stats);
  return $stats unless($stats->analysis_id);
  return $stats if($stats->status eq 'SYNCHING');
  return $stats if($stats->status eq 'DONE');
  return $stats if($stats->sync_lock);
  return $stats if(($stats->status eq 'WORKING') and
405
                   ($stats->seconds_since_last_update < 3*60));
406 407 408 409

  # OK try to claim the sync_lock
  my $sql = "UPDATE analysis_stats SET status='SYNCHING', sync_lock=1 ".
            "WHERE sync_lock=0 and analysis_id=" . $stats->analysis_id;
410
  #print("$sql\n");
411 412
  my $row_count = $self->dbc->do($sql);  
  return $stats unless($row_count == 1);
413
  #printf("got sync_lock on analysis_stats(%d)\n", $stats->analysis_id);
414 415 416
  
  #OK have the lock, go and do the sync
  $self->synchronize_AnalysisStats($stats);
417

418 419 420 421
  return $stats;
}


422
=head2 synchronize_AnalysisStats
423

424 425 426 427 428 429 430 431
  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
432

433
=cut
434

435
sub synchronize_AnalysisStats {
436
  my $self = shift;
437
  my $analysisStats = shift;
438

439 440 441 442 443 444 445
  # Trying to make hive not synchronize if there is a high load in the
  # server, e.g. during blasts (max 450 workers). The best thing I
  # could find is the combination of these two numbers
  if (($self->get_hive_current_load("silent") > 0.9) && $self->get_num_running_workers("silent") > 400) {
    return $analysisStats;
  }

446
  return $analysisStats unless($analysisStats);
447
  return $analysisStats unless($analysisStats->analysis_id);
448
  
449
  $analysisStats->refresh(); ## Need to get the new hive_capacity for dynamic analyses
450 451 452 453 454 455
  $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);

456 457 458 459 460 461
#   my $sql = "SELECT status, count(*) FROM analysis_job ".
#             "WHERE analysis_id=? GROUP BY status";
# This should be better in terms of performance 
# http://www.mysqlperformanceblog.com/2007/08/16/how-much-overhead-is-caused-by-on-disk-temporary-tables/
  my $sql = "SELECT status, count(status) FROM analysis_job ".
            "WHERE analysis_id=? GROUP BY status ORDER BY NULL LIMIT 10";
462 463 464
  my $sth = $self->prepare($sql);
  $sth->execute($analysisStats->analysis_id);

465 466
  my $hive_capacity = $analysisStats->hive_capacity;

467
  while (my ($status, $count)=$sth->fetchrow_array()) {
468
# print STDERR "$status - $count\n";
469 470 471 472 473 474

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

    if($status eq 'READY') {
      $analysisStats->unclaimed_job_count($count);
475 476 477 478 479 480 481 482 483
      my $numWorkers;
      if($analysisStats->batch_size > 0) {
        $numWorkers = POSIX::ceil($count / $analysisStats->batch_size);
      } else {
        my $job_msec = $analysisStats->avg_msec_per_job;
        $job_msec = 100 if($job_msec>0 and $job_msec<100);
        $numWorkers = POSIX::ceil(($count * $job_msec) / (3*60*1000)); 
        # guess num needed workers by total jobs / (num jobs a worker could do in 3 minutes)
      }
484
      $numWorkers=$count if($numWorkers==0);
485 486 487 488 489
      if($analysisStats->hive_capacity>0 and $numWorkers > $analysisStats->hive_capacity) {
        $numWorkers=$analysisStats->hive_capacity;
      }
      $analysisStats->num_required_workers($numWorkers);
    }
490 491
    if ($status eq 'DONE') { $analysisStats->done_job_count($count); }
    if ($status eq 'FAILED') { $analysisStats->failed_job_count($count); }
492 493
  }
  $sth->finish;
494

495
  $self->check_blocking_control_rules_for_AnalysisStats($analysisStats);
496

497 498 499
  if($analysisStats->status ne 'BLOCKED') {
    $analysisStats->determine_status();
  }
500

501 502 503 504
  #
  # adjust_stats_for_living_workers
  #
  
505
  if($analysisStats->hive_capacity > 0) {
506
    my $liveCount = $analysisStats->get_running_worker_count();
507 508 509 510 511 512 513 514

    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);
515
  }
516

517
  $analysisStats->update;  #update and release sync_lock
518

519
  return $analysisStats;
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
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;
    }
546

547 548 549
    if($allRulesDone) {
      if($stats->status eq 'BLOCKED') {
        #print("  UNBLOCK analysis : all conditions met\n");
550
        $stats->update_status('LOADING'); #trigger sync
551 552 553 554
      }
    } else {
      #print("  RE-BLOCK analysis : some conditions failed\n");
      $stats->update_status('BLOCKED');
555 556
    }
  }
557 558 559
}


560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
sub get_num_failed_analyses
{
  my $self = shift;
  my $analysis = shift;

  my $statsDBA = $self->db->get_AnalysisStatsAdaptor;
  my $failed_analyses = $statsDBA->fetch_by_status('FAILED');
  if ($analysis) {
    foreach my $this_failed_analysis (@$failed_analyses) {
      if ($this_failed_analysis->analysis_id == $analysis->dbID) {
        print "#########################################################\n",
            " Too many jobs failed for analysis ".$analysis->logic_name.". FAIL!!\n",
            "#########################################################\n\n";
        return 1;
      }
    }
    return 0;
  }

  if (@$failed_analyses) {
    print "##################################################\n",
        " Too many failed jobs. FAIL!!\n",
        "##################################################\n";
  }

  return scalar(@$failed_analyses);
}


589 590
sub get_hive_current_load {
  my $self = shift;
591 592
  my $silent = shift;
  my $sql = "SELECT /*! SQL_BUFFER_RESULT */ sum(1/analysis_stats.hive_capacity) FROM hive, analysis_stats ".
593 594
            "WHERE hive.analysis_id=analysis_stats.analysis_id and cause_of_death ='' ".
            "AND analysis_stats.hive_capacity>0";
595 596 597 598 599
  my $sth = $self->prepare($sql);
  $sth->execute();
  (my $load)=$sth->fetchrow_array();
  $sth->finish;
  $load=0 unless($load);
600 601
  print("current hive load = $load\n") unless (defined($silent));
  print("*") if ($silent eq 'silent');
602 603 604 605
  return $load;
}


606 607
sub get_num_running_workers {
  my $self = shift;
608
  my $silent = shift;
609 610 611 612 613 614
  my $sql = "SELECT count(*) FROM hive WHERE cause_of_death =''";
  my $sth = $self->prepare($sql);
  $sth->execute();
  (my $runningCount)=$sth->fetchrow_array();
  $sth->finish;
  $runningCount=0 unless($runningCount);
615
  print("current hive num_running_workers = $runningCount\n") unless (defined($silent));
616 617 618
  return $runningCount;
}

619 620 621 622 623 624
sub enter_status {
  my ($self, $worker, $status) = @_;

  $self->dbc->do("UPDATE hive SET status = '$status' WHERE hive_id = ".$worker->hive_id);
}

625 626
=head2 get_num_needed_workers

627
  Arg[1]     : Bio::EnsEMBL::Analysis object (optional)
628 629 630 631
  Example    : $count = $queen->get_num_needed_workers();
  Description: Runs through the analyses in the system which are waiting
               for workers to be created for them.  Calculates the maximum
               number of workers needed to fill the current needs of the system
632
               If Arg[1] is defined, does it only for the given analysis.
633
  Exceptions : none
634
  Caller     : beekeepers and other external processes
635 636 637

=cut

638
sub get_num_needed_workers {
639
  my $self = shift;
640
  my $analysis = shift;
641

642
  my $statsDBA = $self->db->get_AnalysisStatsAdaptor;
643
  my $neededAnals = $statsDBA->fetch_by_needed_workers(undef,$self->{maximise_concurrency});
644 645 646
  my $deeper_stats_list = $statsDBA->fetch_by_status('LOADING', 'BLOCKED');
  push @$neededAnals, @$deeper_stats_list;

647
  return 0 unless($neededAnals);
648

649 650 651 652 653
  my $availableLoad = 1.0 - $self->get_hive_current_load();
  return 0 if($availableLoad <0.0);

  my $numWorkers = 0;
  foreach my $analysis_stats (@{$neededAnals}) {
654 655
    next if (defined $analysis && $analysis->dbID != $analysis_stats->analysis_id);

656 657
    #$analysis_stats->print_stats();

658 659 660 661 662 663 664 665
    #digging deeper under the surface so need to sync
    if(($analysis_stats->status eq 'LOADING') or ($analysis_stats->status eq 'BLOCKED')) {
      $self->synchronize_AnalysisStats($analysis_stats);
    }

    next if($analysis_stats->status eq 'BLOCKED');
    next if($analysis_stats->num_required_workers == 0);

666 667 668 669 670 671
    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)) {
672 673 674
      $numWorkers += $analysis_stats->num_required_workers;
      $availableLoad -= $thisLoad;
      $analysis_stats->print_stats();
675
      printf("  %5d workers (%1.3f remaining-hive-load)\n", $numWorkers, $availableLoad);
676
    } else {
677
      my $workerCount = POSIX::ceil($availableLoad * $analysis_stats->hive_capacity);
678 679 680
      $numWorkers += $workerCount;
      $availableLoad -=  $workerCount * (1/$analysis_stats->hive_capacity);
      $analysis_stats->print_stats();
681
      printf("  %5d workers (%1.3f remaining-hive-load) use only %3d workers\n", $numWorkers, $availableLoad, $workerCount);
682
      last;
683
    }
684
    last if($availableLoad <= 0.0);
685
  }
686

687
  printf("need $numWorkers workers (availLoad=%1.5f)\n", $availableLoad);
688
  return $numWorkers;
689 690 691
}


692 693 694
sub get_hive_progress
{
  my $self = shift;
695
  my $sql = "SELECT /*! SQL_BUFFER_RESULT */ sum(done_job_count), sum(failed_job_count), sum(total_job_count), ".
696 697
            "sum(unclaimed_job_count * analysis_stats.avg_msec_per_job)/1000/60/60 ".
            "FROM analysis_stats";
698 699
  my $sth = $self->prepare($sql);
  $sth->execute();
700
  my ($done, $failed, $total, $cpuhrs) = $sth->fetchrow_array();
701
  $sth->finish;
702
  $done=0 unless($done);
703
  $failed=0 unless($failed);
704 705
  $total=0 unless($total);
  my $completed=0.0;
706
  $completed = ((100.0 * ($done+$failed))/$total)  if($total>0);
707 708 709 710
  my $remaining = $total - $done - $failed;
  printf("hive %1.3f%% complete (< %1.3f CPU_hrs) (%d todo + %d done + %d failed = %d total)\n", 
          $completed, $cpuhrs, $remaining, $done, $failed, $total);
  return $done, $total, $cpuhrs;
711 712
}

713
sub print_hive_status
714 715 716 717 718 719 720 721
{
  my $self = shift;
  $self->print_analysis_status;
  $self->print_running_worker_status;
}


sub print_analysis_status
722 723 724 725
{
  my $self = shift;

  my $allStats = $self->db->get_AnalysisStatsAdaptor->fetch_all();
726
 
727
  foreach my $analysis_stats (@{$allStats}) {
Albert Vilella's avatar
Albert Vilella committed
728
    $analysis_stats->print_stats($self->{'verbose_stats'});
729
  }
730 731 732 733 734 735
}


sub print_running_worker_status
{
  my $self = shift;
736

737
  my $total = 0;
738 739 740 741 742 743 744 745
  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);
746
    $total += $count;
747
  }
748
  printf("  %d total workers\n", $total);
749
  print("=====================\n");
750
  $sth->finish;
751
}
752

753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781
=head2 monitor

  Arg[1]     : --none--
  Example    : $queen->monitor();
  Description: Monitors current throughput and store the result in the monitor
               table
  Exceptions : none
  Caller     : beekeepers and other external processes

=cut

sub monitor
{
  my $self = shift;
  my $sql = qq{
      INSERT INTO monitor
      SELECT
          now(),
          count(*),
          sum(work_done/TIME_TO_SEC(TIMEDIFF(now(),born))),
          sum(work_done/TIME_TO_SEC(TIMEDIFF(now(),born)))/count(*),
          group_concat(DISTINCT logic_name)
      FROM hive left join analysis USING (analysis_id)
      WHERE cause_of_death = ""};
      
  my $sth = $self->prepare($sql);
  $sth->execute();
}

782

Jessica Severin's avatar
Jessica Severin committed
783 784 785 786 787
#
# INTERNAL METHODS
#
###################

788 789 790 791 792 793
sub _pick_best_analysis_for_new_worker {
  my $self = shift;

  my $statsDBA = $self->db->get_AnalysisStatsAdaptor;
  return undef unless($statsDBA);

794
  my ($stats) = @{$statsDBA->fetch_by_needed_workers(1,$self->{maximise_concurrency})};
795 796 797 798 799
  if($stats) {
    #synchronize and double check that it can be run
    $self->safe_synchronize_AnalysisStats($stats);
    return $stats if(($stats->status ne 'BLOCKED') and ($stats->num_required_workers > 0));
  }
800 801

  # ok so no analyses 'need' workers.
802 803 804
  if ($self->get_num_failed_analyses()) {
    return undef;
  }
805 806 807 808 809 810 811
  # see if anything needs an update, in case there are
  # hidden jobs that haven't made it into the summary stats

  print("QUEEN: no obvious needed workers, need to dig deeper\n");
  my $stats_list = $statsDBA->fetch_by_status('LOADING', 'BLOCKED');
  foreach $stats (@$stats_list) {
    #$stats->print_stats();
812
    $self->safe_synchronize_AnalysisStats($stats);
813 814
    #$stats->print_stats();

815
    return $stats if(($stats->status ne 'BLOCKED') and ($stats->num_required_workers > 0));
816 817
  }

818
  ($stats) = @{$statsDBA->fetch_by_needed_workers(1,$self->{maximise_concurrency})};
819 820 821 822 823 824
  return $stats if($stats);

  return undef;
}


Jessica Severin's avatar
Jessica Severin committed
825 826 827 828
=head2 _fetch_by_hive_id

  Arg [1]    : int $id
               the unique database identifier for the feature to be obtained
829
  Example    : $feat = $queen->fetch_by_dbID(1234);
Jessica Severin's avatar
Jessica Severin committed
830 831 832 833 834 835 836 837
  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

838
sub _fetch_by_hive_id {
Jessica Severin's avatar
Jessica Severin committed
839 840 841
  my ($self,$id) = @_;

  unless(defined $id) {
842
    throw("fetch_by_dbID must have an id");
Jessica Severin's avatar
Jessica Severin committed
843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
  }

  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
933
             h.beekeeper
Jessica Severin's avatar
Jessica Severin committed
934 935 936
             h.host
             h.process_id
             h.work_done
937
             h.status
Jessica Severin's avatar
Jessica Severin committed
938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957
             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'});
958
    $worker->beekeeper($column{'beekeeper'});
Jessica Severin's avatar
Jessica Severin committed
959 960 961
    $worker->host($column{'host'});
    $worker->process_id($column{'process_id'});
    $worker->work_done($column{'work_done'});
962
    $worker->status($column{'status'});
Jessica Severin's avatar
Jessica Severin committed
963 964 965 966
    $worker->born($column{'born'});
    $worker->last_check_in($column{'last_check_in'});
    $worker->died($column{'died'});
    $worker->cause_of_death($column{'cause_of_death'});
967
    $worker->queen($self);
Jessica Severin's avatar
Jessica Severin committed
968 969
    $worker->db($self->db);

970 971
    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
972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992
    }

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

  return \@workers
}

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

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


1;