changed the meaning of 'num_required_workers' to "total estimated number of...

changed the meaning of 'num_required_workers' to "total estimated number of workers needed for this analysis" ( 'num_running_workers' is now included in it )

modules/Bio/EnsEMBL/Hive/AnalysisStats.pm

@@ -134,6 +134,8 @@ sub num_running_workers {
     return $self->{'_num_running_workers'};
 }
 
+# NB: the meaning of this field changes from "how many workers we need to add" to "how many workers we actually need"
+
 sub num_required_workers {
     my $self = shift;
     $self->{'_num_required_workers'} = shift if(@_);
@@ -293,7 +295,7 @@ sub print_stats {
 
     my $analysis = $self->get_analysis;
 
-    printf("%-27s(%2d) %11s jobs(Sem:%d, Rdy:%d, InProg:%d, Done+Pass:%d, Fail:%d)=%d Ave_msec:%d, ",
+    printf("%-27s(%2d) %11s jobs(Sem:%d, Rdy:%d, InProg:%d, Done+Pass:%d, Fail:%d)=%d Ave_msec:%d, workers(Running:%d, Reqired:%d) ",
         $analysis->logic_name,
         $self->analysis_id,
         $self->status,
@@ -306,9 +308,11 @@ sub print_stats {
         $self->total_job_count,
 
         $self->avg_msec_per_job,
+
+        $self->num_running_workers,
+        $self->num_required_workers,
     );
-    print 'req.workers:'.$self->num_required_workers
-         .'  h.cap:'    .( defined($self->hive_capacity) ? $self->hive_capacity : '-' )
+    print '  h.cap:'    .( defined($self->hive_capacity) ? $self->hive_capacity : '-' )
          .'  a.cap:'    .( defined($analysis->analysis_capacity) ? $analysis->analysis_capacity : '-')
          ."  (sync'd "  .$self->seconds_since_last_update." sec ago)\n",
 }

modules/Bio/EnsEMBL/Hive/DBSQL/AnalysisStatsAdaptor.pm

@@ -90,7 +90,7 @@ sub fetch_all_by_suitability_rc_id_meadow_type {
         # the ones that clearly have work to do:
         #
     my $primary_results = $self->_generic_fetch(
-        "ast.num_required_workers>0 AND ast.status in ('READY', 'WORKING')" ,
+        "ast.num_running_workers<ast.num_required_workers AND ast.status in ('READY', 'WORKING')" ,
         $join ,
         'ORDER BY a.priority DESC, ' . ( ($self->dbc->driver eq 'sqlite') ? 'RANDOM()' : 'RAND()' ),
     );

modules/Bio/EnsEMBL/Hive/Queen.pm

@@ -245,8 +245,8 @@ sub specialize_new_worker {
             if($stats->status eq 'BLOCKED') {
                 die "Analysis is BLOCKED, can't specialize a worker";
             }
-            if($stats->num_required_workers <= 0) {
-                die "Analysis requires 0 workers at the moment";
+            if($stats->num_running_workers >= $stats->num_required_workers) {
+                die "Analysis doesn't require extra workers at the moment";
             }
             if($stats->status eq 'DONE') {
                 die "Analysis is DONE, and doesn't require workers";
@@ -580,12 +580,9 @@ sub synchronize_AnalysisStats {
 
   if($self->db->hive_use_triggers()) {
 
-            my $job_count = $analysisStats->ready_job_count();
+            my $job_count           = $analysisStats->ready_job_count();
 
-            my $required_workers     = $scheduling_allowed
-                                    && ( POSIX::ceil( $job_count / $analysisStats->get_or_estimate_batch_size() )
-                                         - $analysisStats->num_running_workers() );
-            $required_workers = 0 if($required_workers < 0);
+            my $required_workers    = $scheduling_allowed && POSIX::ceil( $job_count / $analysisStats->get_or_estimate_batch_size() );
 
             $analysisStats->num_required_workers( $required_workers );
 
@@ -613,10 +610,7 @@ sub synchronize_AnalysisStats {
         if($status eq 'READY') {
             $analysisStats->ready_job_count($job_count);
 
-            my $required_workers     = $scheduling_allowed
-                                    && ( POSIX::ceil( $job_count / $analysisStats->get_or_estimate_batch_size() )
-                                         - $analysisStats->num_running_workers() );
-            $required_workers = 0 if($required_workers < 0);
+            my $required_workers    = $scheduling_allowed && POSIX::ceil( $job_count / $analysisStats->get_or_estimate_batch_size() );
 
             $analysisStats->num_required_workers( $required_workers );
 
@@ -843,7 +837,7 @@ sub suggest_analysis_to_specialize_by_rc_id_meadow_type {
 
             #synchronize and double check that it can be run:
         $self->safe_synchronize_AnalysisStats($stats);
-        return $stats if( ($stats->status ne 'BLOCKED') and ($stats->status ne 'SYNCHING') and ($stats->num_required_workers > 0) );
+        return $stats if( ($stats->status ne 'BLOCKED') and ($stats->status ne 'SYNCHING') and ($stats->num_running_workers < $stats->num_required_workers) );
     }
 
     return undef;

modules/Bio/EnsEMBL/Hive/Scheduler.pm

@@ -121,8 +121,10 @@ sub schedule_workers {
         next if($analysis_stats->status eq 'BLOCKED');
 
             # getting the initial worker requirement for this analysis (may be stale if not sync'ed recently)
-        my $workers_this_analysis = $analysis_stats->num_required_workers
-            or next;
+        my $extra_workers_this_analysis = $analysis_stats->num_required_workers - $analysis_stats->num_running_workers;
+
+            # if this analysis doesn't require any extra workers - just skip it:
+        next if ($extra_workers_this_analysis <= 0);
 
             # setting up all negotiating limiters:
         $queen_capacity->multiplier( $analysis_stats->hive_capacity );
@@ -135,22 +137,22 @@ sub schedule_workers {
 
             # negotiations:
         foreach my $limiter (@limiters) {
-            $workers_this_analysis = $limiter->preliminary_offer( $workers_this_analysis );
+            $extra_workers_this_analysis = $limiter->preliminary_offer( $extra_workers_this_analysis );
         }
 
             # do not continue with this analysis if limiters haven't agreed on a positive number:
-        next unless($workers_this_analysis);
+        next unless($extra_workers_this_analysis);
 
             # let all parties know the final decision of negotiations:
         foreach my $limiter (@limiters) {
-            $limiter->final_decision( $workers_this_analysis );
+            $limiter->final_decision( $extra_workers_this_analysis );
         }
 
         my $this_rc_name    = $analysis_id2rc_name->{ $analysis_stats->analysis_id };
-        $workers_to_submit_by_meadow_type_rc_name{ $this_meadow_type }{ $this_rc_name } += $workers_this_analysis;
-        $total_workers_to_submit                                                        += $workers_this_analysis;
+        $workers_to_submit_by_meadow_type_rc_name{ $this_meadow_type }{ $this_rc_name } += $extra_workers_this_analysis;
+        $total_workers_to_submit                                                        += $extra_workers_this_analysis;
         $analysis_stats->print_stats();
-        printf("Before checking the Valley for pending jobs, Scheduler allocated $workers_this_analysis x $this_meadow_type:$this_rc_name workers for '%s' [%.4f hive_load remaining]\n",
+        printf("Before checking the Valley for pending jobs, Scheduler allocated $extra_workers_this_analysis x $this_meadow_type:$this_rc_name extra workers for '%s' [%.4f hive_load remaining]\n",
             $analysis->logic_name,
             $queen_capacity->available_capacity,
         );