Class: Elasticsearch::Rails::HA::ParallelIndexer

Inherits:
Object
  • Object
show all
Defined in:
lib/elasticsearch/rails/ha/parallel_indexer.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(opts) ⇒ ParallelIndexer

leverage multiple cores to run indexing in parallel



12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 12

def initialize(opts)
  @klass    = opts[:klass] or fail "klass required"
  @idx_name = opts[:idx_name] or fail "idx_name required"
  @nprocs   = opts[:nprocs] or fail "nprocs required"
  @batch_size = opts[:batch_size] or fail "batch_size required"
  @max        = opts[:max]
  @force      = opts[:force]
  @verbose    = opts[:verbose]
  @scope      = opts[:scope]

  # make sure klass is not a simple string
  if @klass.is_a?(String)
    @klass = @klass.constantize
  end

  # calculate array of offsets based on nprocs
  @total_expected = klass.count
  @pool_size = (@total_expected / @nprocs.to_f).ceil
end

Instance Attribute Details

#batch_size ⇒ Object (readonly)

Returns the value of attribute batch_size.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def batch_size
  @batch_size
end

#force ⇒ Object (readonly)

Returns the value of attribute force.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def force
  @force
end

#idx_name ⇒ Object (readonly)

Returns the value of attribute idx_name.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def idx_name
  @idx_name
end

#klass ⇒ Object (readonly)

Returns the value of attribute klass.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def klass
  @klass
end

#max ⇒ Object (readonly)

Returns the value of attribute max.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def max
  @max
end

#nprocs ⇒ Object (readonly)

Returns the value of attribute nprocs.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def nprocs
  @nprocs
end

#scope ⇒ Object (readonly)

Returns the value of attribute scope.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def scope
  @scope
end

#verbose ⇒ Object (readonly)

Returns the value of attribute verbose.



9
10
11
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 9

def verbose
  @verbose
end

Instance Method Details

#process_child_results(results) ⇒ Object 



80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 80

def process_child_results(results)
  # check exit status of each child so we know if we should throw exception
  results.each do |pair|
    pid = pair[0]
    pstat = pair[1]
    exit_ok = true
    if pstat.exited?
      @verbose and puts ::ANSI.blue{ "PID #{pid} exited with #{pstat.exitstatus}" }
    end
    if pstat.signaled?
      puts ::ANSI.red{ " >> #{pid} exited with uncaught signal #{pstat.termsig}" }
      exit_ok = false
    end

    if !pstat.success?
      puts ::ANSI.red{ " >> #{pid} was not successful" }
      exit_ok = false
    end

    if pair[1].exitstatus != 0
      puts ::ANSI.red{ " >> #{pid} exited with non-zero status" }
      exit_ok = false
    end

    if !exit_ok
      raise ::ANSI.red{ "PID #{pair[0]} exited abnormally, so the whole reindex fails" }
    end
  end
end

#run ⇒ Object 



32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 32

def run
  return if @pool_size < 1

  # get all ids since we can't assume there are no holes in the PK sequencing
  ids = klass.order('id ASC').pluck(:id)
  offsets = []
  ids.each_slice(@pool_size) do |chunk|
    #puts "chunk: size=#{chunk.size} #{chunk.first}..#{chunk.last}"
    offsets.push( chunk.first )
  end
  if @verbose
    puts ::ANSI.blue{ "Parallel Indexer: index=#{@idx_name} total=#{@total_expected} nprocs=#{@nprocs} pool_size=#{@pool_size} offsets=#{offsets} " }
  end

  if @force
    @verbose and puts ::ANSI.blue{ "Force creating new index" }
    klass.__elasticsearch__.create_index! force: true, index: idx_name
    klass.__elasticsearch__.refresh_index! index: idx_name
  end

  @current_db_config = ActiveRecord::Base.connection_config
  # IMPORTANT before forks in offsets loop
  ActiveRecord::Base.connection.disconnect!

  child_pids = []
  offsets.each do |start_at|
    child_pid = fork do
      run_child(start_at)
    end
    if child_pid
      child_pids << child_pid
    end
  end

  # reconnect in parent
  ActiveRecord::Base.establish_connection(@current_db_config)

  # Process.waitall seems to hang during tests. Do it manually.
  child_results = []

  child_pids.each do |pid|
    Process.wait(pid)
    child_results.push [pid, $?]
  end

  process_child_results(child_results)
end

#run_child(start_at) ⇒ Object 



110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
 
# File 'lib/elasticsearch/rails/ha/parallel_indexer.rb', line 110

def run_child(start_at)
  # IMPORTANT after fork
  ActiveRecord::Base.establish_connection(@current_db_config)

  # IMPORTANT for tests to determine whether at_end should run
  ENV["I_AM_HA_CHILD"] = "true"

  completed = 0
  errors    = []
  @verbose and puts ::ANSI.blue{ "Start worker #{$$} at offset #{start_at}" }
  pbar = ::ANSI::Progressbar.new("#{klass} [#{$$}]", @pool_size, STDOUT) rescue nil
  checkpoint = false
  if pbar
    win_width = pbar.__send__ :get_width
    title_width = (win_width / 4).to_i
    pbar.format("#{klass} [#{$$}]: %3d%% %s %s", :percentage, :bar, :stat)
    pbar.__send__ :show
    pbar.bar_mark = '='
  else
    checkpoint = true
  end

  @klass.__elasticsearch__.import return: 'errors',
    index: @idx_name,
    start: start_at,
    scope: @scope,
    batch_size: @batch_size    do |resp|
      # show errors immediately (rather than buffering them)
      errors += resp['items'].select { |k, v| k.values.first['error'] }
      completed += resp['items'].size
      if pbar && @verbose
        pbar.inc resp['items'].size
      end
      if checkpoint && @verbose
        puts ::ANSI.blue{ "[#{$$}] #{Time.now.utc.iso8601} : #{completed} records completed" }
      end
      STDERR.flush
      STDOUT.flush
      if errors.size > 0
        STDOUT.puts "ERRORS in #{$$}:"
        STDOUT.puts errors.pretty_inspect
      end
      if completed >= @pool_size || (@max && @max.to_i == completed)
        pbar.finish if pbar
        @verbose and puts ::ANSI.blue{ "Worker #{$$} finished #{completed} records" }
        exit!(true) # exit child worker
      end
    end # end do |resp| block
end