Changeset 27685

Show
Ignore:
Timestamp:
21.06.2013 12:24:54 (6 years ago)
Author:
jmt12
Message:

in the case of multiple attempts you need to retain the information about data location tips - whereas I was undefining it after the first task with that id (assuming they were unique). Also added in a check that data location tips are defined for a task before accessing them

Files:
1 modified

Legend:

Unmodified
Added
Removed
  • gs2-extensions/parallel-building/trunk/src/bin/script/parse_task_info_from_hadoop_log.pl

    r27589 r27685  
    246246    my $task_id = $job_id . '_m_' . $task_number; 
    247247    my $compute_node = $aid_2_node->{$attempt_id}->{'compute_node'}; 
    248     my @splits = @{$tid_2_splits->{$task_id}}; 
    249     my $data_local = 0; 
    250     if (grep($_ eq $compute_node, @splits)) 
    251     { 
    252       $data_local = 1; 
    253     } 
    254     print CSVOUT $task_number . "," . $attempt_number . "," . $data_local . ",\"" . $compute_node . "\",\"" . join(',', natsort(@splits)) . "\"\n"; 
    255     $tid_2_splits->{$task_id} = undef; 
     248    if (defined $tid_2_splits->{$task_id}) 
     249    { 
     250      my @splits = @{$tid_2_splits->{$task_id}}; 
     251      my $data_local = 0; 
     252      if (grep($_ eq $compute_node, @splits)) 
     253      { 
     254        $data_local = 1; 
     255      } 
     256      print CSVOUT $task_number . "," . $attempt_number . "," . $data_local . ",\"" . $compute_node . "\",\"" . join(',', natsort(@splits)) . "\"\n"; 
     257    } 
     258    else 
     259    { 
     260      print "Warning! Missing data location information for task: " . $task_id . "\n"; 
     261    } 
    256262  } 
    257263