fairy 0.6.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (186) hide show
  1. data/LICENSE +674 -0
  2. data/Makefile +116 -0
  3. data/README +15 -0
  4. data/bin/fairy +582 -0
  5. data/bin/fairy-cat +74 -0
  6. data/bin/fairy-cp +128 -0
  7. data/bin/fairy-rm +122 -0
  8. data/bin/subcmd/controller +41 -0
  9. data/bin/subcmd/inspector +81 -0
  10. data/bin/subcmd/master +43 -0
  11. data/bin/subcmd/node +47 -0
  12. data/bin/subcmd/processor +54 -0
  13. data/doc/programming-interface.html +240 -0
  14. data/doc/programming-interface.rd +300 -0
  15. data/etc/fairy.conf.tmpl +118 -0
  16. data/ext/simple_hash/extconf.rb +4 -0
  17. data/ext/simple_hash/simple_hash.c +42 -0
  18. data/fairy.gemspec +60 -0
  19. data/lib/fairy/client/addins.rb +20 -0
  20. data/lib/fairy/client/barrier.rb +29 -0
  21. data/lib/fairy/client/basic-group-by.rb +52 -0
  22. data/lib/fairy/client/cat.rb +41 -0
  23. data/lib/fairy/client/direct-product.rb +51 -0
  24. data/lib/fairy/client/equijoin.rb +79 -0
  25. data/lib/fairy/client/exec.rb +54 -0
  26. data/lib/fairy/client/filter.rb +62 -0
  27. data/lib/fairy/client/find.rb +35 -0
  28. data/lib/fairy/client/group-by.rb +194 -0
  29. data/lib/fairy/client/here.rb +84 -0
  30. data/lib/fairy/client/inject.rb +70 -0
  31. data/lib/fairy/client/input-file.rb +53 -0
  32. data/lib/fairy/client/input-iota.rb +49 -0
  33. data/lib/fairy/client/input-local-file.rb +188 -0
  34. data/lib/fairy/client/input-varray.rb +30 -0
  35. data/lib/fairy/client/input.rb +42 -0
  36. data/lib/fairy/client/io-filter.rb +26 -0
  37. data/lib/fairy/client/junction.rb +31 -0
  38. data/lib/fairy/client/map.rb +34 -0
  39. data/lib/fairy/client/merge-group-by.rb +71 -0
  40. data/lib/fairy/client/output-file.rb +64 -0
  41. data/lib/fairy/client/output-local-file.rb +60 -0
  42. data/lib/fairy/client/output-null.rb +47 -0
  43. data/lib/fairy/client/output-varray.rb +50 -0
  44. data/lib/fairy/client/output.rb +29 -0
  45. data/lib/fairy/client/roma-put.rb +62 -0
  46. data/lib/fairy/client/roma.rb +156 -0
  47. data/lib/fairy/client/seg-join.rb +61 -0
  48. data/lib/fairy/client/seg-map.rb +78 -0
  49. data/lib/fairy/client/seg-shuffle.rb +35 -0
  50. data/lib/fairy/client/seg-split.rb +27 -0
  51. data/lib/fairy/client/seg-zip.rb +60 -0
  52. data/lib/fairy/client/select.rb +38 -0
  53. data/lib/fairy/client/sort.rb +48 -0
  54. data/lib/fairy/client/sort18.rb +56 -0
  55. data/lib/fairy/client/sort19.rb +61 -0
  56. data/lib/fairy/client/there.rb +47 -0
  57. data/lib/fairy/client/top_n_into_roma.rb +34 -0
  58. data/lib/fairy/client/wc.rb +92 -0
  59. data/lib/fairy/controller.rb +1103 -0
  60. data/lib/fairy/logger.rb +107 -0
  61. data/lib/fairy/master/addins.rb +20 -0
  62. data/lib/fairy/master/atom.rb +17 -0
  63. data/lib/fairy/master/c-barrier.rb +283 -0
  64. data/lib/fairy/master/c-basic-group-by.rb +250 -0
  65. data/lib/fairy/master/c-cat.rb +159 -0
  66. data/lib/fairy/master/c-direct-product.rb +203 -0
  67. data/lib/fairy/master/c-exec.rb +68 -0
  68. data/lib/fairy/master/c-filter.rb +422 -0
  69. data/lib/fairy/master/c-find.rb +138 -0
  70. data/lib/fairy/master/c-group-by.rb +64 -0
  71. data/lib/fairy/master/c-here.rb +80 -0
  72. data/lib/fairy/master/c-inject.rb +119 -0
  73. data/lib/fairy/master/c-input-file.rb +46 -0
  74. data/lib/fairy/master/c-input-iota.rb +66 -0
  75. data/lib/fairy/master/c-input-local-file.rb +117 -0
  76. data/lib/fairy/master/c-input-varray.rb +53 -0
  77. data/lib/fairy/master/c-input.rb +24 -0
  78. data/lib/fairy/master/c-inputtable.rb +31 -0
  79. data/lib/fairy/master/c-inputtable18.rb +36 -0
  80. data/lib/fairy/master/c-inputtable19.rb +35 -0
  81. data/lib/fairy/master/c-io-filter.rb +28 -0
  82. data/lib/fairy/master/c-junction.rb +54 -0
  83. data/lib/fairy/master/c-map.rb +27 -0
  84. data/lib/fairy/master/c-merge-group-by.rb +241 -0
  85. data/lib/fairy/master/c-output-file.rb +84 -0
  86. data/lib/fairy/master/c-output-local-file.rb +19 -0
  87. data/lib/fairy/master/c-output-null.rb +45 -0
  88. data/lib/fairy/master/c-output-varray.rb +57 -0
  89. data/lib/fairy/master/c-output.rb +20 -0
  90. data/lib/fairy/master/c-seg-join.rb +141 -0
  91. data/lib/fairy/master/c-seg-map.rb +26 -0
  92. data/lib/fairy/master/c-seg-shuffle.rb +87 -0
  93. data/lib/fairy/master/c-seg-split.rb +110 -0
  94. data/lib/fairy/master/c-seg-zip.rb +132 -0
  95. data/lib/fairy/master/c-select.rb +27 -0
  96. data/lib/fairy/master/c-sort.rb +108 -0
  97. data/lib/fairy/master/c-there.rb +57 -0
  98. data/lib/fairy/master/c-wc.rb +232 -0
  99. data/lib/fairy/master/job-interpriter.rb +19 -0
  100. data/lib/fairy/master/scheduler.rb +24 -0
  101. data/lib/fairy/master.rb +329 -0
  102. data/lib/fairy/node/addins.rb +19 -0
  103. data/lib/fairy/node/p-barrier.rb +95 -0
  104. data/lib/fairy/node/p-basic-group-by.rb +252 -0
  105. data/lib/fairy/node/p-direct-product.rb +153 -0
  106. data/lib/fairy/node/p-exec.rb +30 -0
  107. data/lib/fairy/node/p-filter.rb +363 -0
  108. data/lib/fairy/node/p-find.rb +111 -0
  109. data/lib/fairy/node/p-group-by.rb +1534 -0
  110. data/lib/fairy/node/p-here.rb +21 -0
  111. data/lib/fairy/node/p-identity.rb +24 -0
  112. data/lib/fairy/node/p-inject.rb +127 -0
  113. data/lib/fairy/node/p-input-file.rb +108 -0
  114. data/lib/fairy/node/p-input-iota.rb +39 -0
  115. data/lib/fairy/node/p-input-local-file.rb +61 -0
  116. data/lib/fairy/node/p-input-varray.rb +26 -0
  117. data/lib/fairy/node/p-io-filter.rb +28 -0
  118. data/lib/fairy/node/p-map.rb +40 -0
  119. data/lib/fairy/node/p-merger-group-by.rb +48 -0
  120. data/lib/fairy/node/p-output-file.rb +104 -0
  121. data/lib/fairy/node/p-output-local-file.rb +14 -0
  122. data/lib/fairy/node/p-output-null.rb +32 -0
  123. data/lib/fairy/node/p-output-varray.rb +41 -0
  124. data/lib/fairy/node/p-seg-join.rb +82 -0
  125. data/lib/fairy/node/p-seg-map.rb +34 -0
  126. data/lib/fairy/node/p-seg-split.rb +61 -0
  127. data/lib/fairy/node/p-seg-zip.rb +79 -0
  128. data/lib/fairy/node/p-select.rb +40 -0
  129. data/lib/fairy/node/p-single-exportable.rb +90 -0
  130. data/lib/fairy/node/p-sort.rb +195 -0
  131. data/lib/fairy/node/p-task.rb +113 -0
  132. data/lib/fairy/node/p-there.rb +44 -0
  133. data/lib/fairy/node/p-wc.rb +266 -0
  134. data/lib/fairy/node.rb +187 -0
  135. data/lib/fairy/processor.rb +510 -0
  136. data/lib/fairy/share/base-app.rb +114 -0
  137. data/lib/fairy/share/block-source.rb +234 -0
  138. data/lib/fairy/share/conf.rb +396 -0
  139. data/lib/fairy/share/debug.rb +21 -0
  140. data/lib/fairy/share/encoding.rb +17 -0
  141. data/lib/fairy/share/fast-tempfile.rb +93 -0
  142. data/lib/fairy/share/file-place.rb +176 -0
  143. data/lib/fairy/share/hash-1.rb +20 -0
  144. data/lib/fairy/share/hash-md5.rb +28 -0
  145. data/lib/fairy/share/hash-murmur.rb +69 -0
  146. data/lib/fairy/share/hash-rb18.rb +20 -0
  147. data/lib/fairy/share/hash-simple-hash.rb +28 -0
  148. data/lib/fairy/share/inspector.rb +16 -0
  149. data/lib/fairy/share/lc/exceptions.rb +82 -0
  150. data/lib/fairy/share/lc/ja/exceptions.rb +81 -0
  151. data/lib/fairy/share/locale.rb +17 -0
  152. data/lib/fairy/share/log.rb +215 -0
  153. data/lib/fairy/share/pool-dictionary.rb +53 -0
  154. data/lib/fairy/share/port-marshaled-queue.rb +347 -0
  155. data/lib/fairy/share/port.rb +1697 -0
  156. data/lib/fairy/share/reference.rb +45 -0
  157. data/lib/fairy/share/stdout.rb +56 -0
  158. data/lib/fairy/share/tr.rb +16 -0
  159. data/lib/fairy/share/varray.rb +147 -0
  160. data/lib/fairy/share/vfile.rb +183 -0
  161. data/lib/fairy/version.rb +8 -0
  162. data/lib/fairy.rb +206 -0
  163. data/sample/grep.rb +46 -0
  164. data/sample/ping.rb +19 -0
  165. data/sample/sort.rb +102 -0
  166. data/sample/wordcount.rb +61 -0
  167. data/spec/README +12 -0
  168. data/spec/fairy1_spec.rb +31 -0
  169. data/spec/fairy2_spec.rb +42 -0
  170. data/spec/fairy3_spec.rb +126 -0
  171. data/spec/fairy4_spec.rb +63 -0
  172. data/spec/fairy5_spec.rb +45 -0
  173. data/spec/fairy6_spec.rb +52 -0
  174. data/spec/fairy7_spec.rb +58 -0
  175. data/spec/fairy8_spec.rb +48 -0
  176. data/spec/mkdat.rb +148 -0
  177. data/spec/run_all.sh +65 -0
  178. data/test/testc.rb +7111 -0
  179. data/tools/cap_recipe/Capfile +144 -0
  180. data/tools/cap_recipe/cluster.yml.sample +14 -0
  181. data/tools/fairy_perf_graph.rb +444 -0
  182. data/tools/git-tag +44 -0
  183. data/tools/log-analysis.rb +62 -0
  184. data/tools/svn-ls-diff +38 -0
  185. data/tools/svn-tags +37 -0
  186. metadata +298 -0
@@ -0,0 +1,138 @@
1
+ # encoding: UTF-8
2
+ #
3
+ # Copyright (C) 2007-2010 Rakuten, Inc.
4
+ #
5
+
6
+ require "forwardable"
7
+
8
+ require "deep-connect"
9
+
10
+ require "fairy/master/c-io-filter"
11
+
12
+ module Fairy
13
+ class CFind<CIOFilter
14
+ extend Forwardable
15
+
16
+ Controller.def_export self
17
+
18
+ def initialize(controller, opts, block_source)
19
+ super
20
+
21
+ @block_source = block_source
22
+
23
+ @clocal_find = CLocalFind.new(controller, opts, block_source)
24
+ @cfind_result = CFindResult.new(controller, opts, self)
25
+
26
+ end
27
+
28
+ def_delegator :@cfind_result, :value
29
+ def_delegator :@cfind_result, :output=
30
+
31
+ def input=(input)
32
+ @clocal_find.input = input
33
+ @cfind_result.input = @clocal_find
34
+ end
35
+
36
+ def update_find
37
+ @clocal_find.break_running
38
+ end
39
+
40
+ class CLocalFind<CIOFilter
41
+ def initialize(controller, opts, block_source)
42
+ super
43
+ @block_source = block_source
44
+
45
+ @no = 0
46
+ @first_node = nil
47
+ @first_node_mutex = Mutex.new
48
+ end
49
+
50
+ def node_class_name
51
+ "PLocalFind"
52
+ end
53
+
54
+ def njob_creation_params
55
+ [@block_source]
56
+ end
57
+
58
+ def each_assigned_filter(&block)
59
+ super
60
+
61
+ @first_node.export.output_no_import = @no
62
+ end
63
+
64
+ def each_export_by(njob, mapper, &block)
65
+ @first_node_mutex.synchronize do
66
+ @no += 1
67
+ if @first_node
68
+ njob.export.output = @first_node.export.output
69
+ njob.export.no = @no - 1
70
+ else
71
+ super
72
+ @first_node = njob
73
+ end
74
+ end
75
+ end
76
+
77
+ def bind_export(exp, imp)
78
+ # do nothing.
79
+ end
80
+
81
+ # def each_export(&block)
82
+ # no = 0
83
+ # first_node = nil
84
+ # each_node do |node|
85
+ # no += 1
86
+ # if first_node
87
+ # node.export.output = first_node.export.output
88
+ # else
89
+ # first_node = node
90
+ # block.call node.export, node
91
+ # end
92
+ # end
93
+ # first_node.export.output_no_import = no
94
+ # end
95
+
96
+ # def find_break
97
+ # # create node 中ならそれをとめる
98
+ # break_create_node
99
+ # # 各tasklettをとめる
100
+ # each_node do |tasklet|
101
+ # tasklet.find_break
102
+ # end
103
+ # end
104
+
105
+ end
106
+
107
+ class CFindResult<CIOFilter
108
+ def initialize(controller, opts, bfind)
109
+ super
110
+ @bfind = bfind
111
+
112
+ @find_mutex = Mutex.new
113
+ @findp = false
114
+ end
115
+
116
+ def node_class_name
117
+ "PFindResult"
118
+ end
119
+
120
+ def njob_creation_params
121
+ []
122
+ end
123
+
124
+ def value
125
+ each_node{|node| return node.value}
126
+ end
127
+
128
+ def update_find
129
+ @find_mutex.synchronize do
130
+ if !@findp
131
+ @findp = true
132
+ @bfind.update_find
133
+ end
134
+ end
135
+ end
136
+ end
137
+ end
138
+ end
@@ -0,0 +1,64 @@
1
+ # encoding: UTF-8
2
+ #
3
+ # Copyright (C) 2007-2010 Rakuten, Inc.
4
+ #
5
+
6
+ require "fairy/master/c-io-filter"
7
+ require "fairy/master/c-inputtable"
8
+ require "fairy/master/c-basic-group-by"
9
+
10
+ module Fairy
11
+ class CGroupBy<CBasicGroupBy
12
+ Controller.def_export self
13
+
14
+ def initialize(controller, opts, block_source)
15
+ super
16
+
17
+ @hash_seed = controller.hash_seed
18
+ end
19
+
20
+ attr_reader :hash_seed
21
+
22
+ def node_class_name
23
+ "PGroupBy"
24
+ end
25
+
26
+ class CPostFilter<CIOFilter
27
+ Controller.def_export self
28
+
29
+ def initialize(controller, opts, block_source)
30
+ super
31
+ @block_source = block_source
32
+ end
33
+
34
+ def node_class_name
35
+ "PGroupBy::PPostFilter"
36
+ end
37
+
38
+ def njob_creation_params
39
+ [@block_source]
40
+ end
41
+
42
+ def create_import(processor)
43
+ policy = @opts[:postfilter_prequeuing_policy]
44
+ policy ||= @opts[:prequeuing_policy]
45
+
46
+ processor.create_import(policy)
47
+ end
48
+ end
49
+
50
+ # class BPostAfterModFilter<BFilter
51
+ # Controller.def_export self
52
+
53
+ # def initialize(controller, opts, block_source)
54
+ # super
55
+ # @block_source = block_source
56
+ # end
57
+
58
+ # def node_class_name
59
+ # "NPostAfterModFilter"
60
+ # end
61
+ # end
62
+
63
+ end
64
+ end
@@ -0,0 +1,80 @@
1
+ # encoding: UTF-8
2
+ #
3
+ # Copyright (C) 2007-2010 Rakuten, Inc.
4
+ #
5
+
6
+ require "fairy/master/c-io-filter"
7
+ require "fairy/master/c-inputtable"
8
+
9
+ require "fairy/share/port"
10
+
11
+
12
+ module Fairy
13
+ class CHere<CIOFilter
14
+ Controller.def_export self
15
+
16
+ def node_class_name
17
+ "PHere"
18
+ end
19
+
20
+ def create_and_add_node(ntask, mapper, opts={})
21
+ node = create_node(ntask) {|node|
22
+ if opts[:init_njob]
23
+ opts[:init_njob].call(node)
24
+ end
25
+ mapper.bind_input(node)
26
+ # exp = node.start_export
27
+ }
28
+ node
29
+ end
30
+
31
+ def each(&block)
32
+ policy = @opts[:prequeuing_policy]
33
+ each_node do |node|
34
+ node.start_export
35
+ import = Import.new(policy)
36
+ import.set_log_callback do |n, key|
37
+ Log::verbose(self, "IMPORT POP key=#{key}: #{n}")
38
+ end
39
+ import.no_import = 1
40
+ node.export.output = import
41
+ import.each do |e|
42
+ block.call e
43
+ end
44
+ end
45
+ end
46
+ DeepConnect.def_method_spec(self, "each(){DVAL}")
47
+
48
+ def each_buf(&block)
49
+ threshold = @opts[:pool_threshold] || CONF.HERE_POOL_THRESHOLD
50
+ chunk = []
51
+
52
+ policy = @opts[:prequeuing_policy]
53
+ each_node do |node|
54
+ node.start_export
55
+ import = Import.new(policy)
56
+ import.set_log_callback do |n, key|
57
+ Log::verbose(self, "IMPORT POP key=#{key}: #{n}")
58
+ end
59
+ import.no_import = 1
60
+ node.export.output = import
61
+
62
+ import.each do |e|
63
+ #[REQ#89]
64
+
65
+ chunk.push e
66
+ if chunk.size > threshold
67
+ block.call chunk
68
+ chunk.clear
69
+ end
70
+ end
71
+ end
72
+ if !chunk.empty?
73
+ block.call chunk
74
+ end
75
+ end
76
+ DeepConnect.def_method_spec(self, "each_buf(){DVAL}")
77
+ end
78
+ end
79
+
80
+
@@ -0,0 +1,119 @@
1
+ # encoding: UTF-8
2
+ #
3
+ # Copyright (C) 2007-2010 Rakuten, Inc.
4
+ #
5
+
6
+ require "forwardable"
7
+
8
+ require "deep-connect"
9
+
10
+ require "fairy/master/c-io-filter"
11
+
12
+ module Fairy
13
+ class CInject<CIOFilter
14
+ extend Forwardable
15
+
16
+ Controller.def_export self
17
+
18
+ def initialize(controller, opts, block_source)
19
+ super
20
+
21
+ @block_source = block_source
22
+
23
+ @clocal_inject = CLocalInject.new(controller, opts, block_source)
24
+ @cwide_inject = CWideInject.new(controller, opts, block_source)
25
+
26
+ end
27
+
28
+ def_delegator :@cwide_inject, :value
29
+ def_delegator :@cwide_inject, :output=
30
+ def_delegator :@cwide_inject, :each_assigned_filter
31
+ def_delegator :@cwide_inject, :each_export_by
32
+ def_delegator :@cwide_inject, :bind_export
33
+
34
+
35
+ def input=(input)
36
+ @clocal_inject.input = input
37
+ @cwide_inject.input = @clocal_inject
38
+ end
39
+
40
+ class CLocalInject<CIOFilter
41
+ def initialize(controller, opts, block_source)
42
+ super
43
+ @block_source = block_source
44
+
45
+ @no = 0
46
+ @first_node = nil
47
+ @first_node_mutex = Mutex.new
48
+ end
49
+
50
+ def node_class_name
51
+ "PLocalInject"
52
+ end
53
+
54
+ def njob_creation_params
55
+ [@block_source]
56
+ end
57
+
58
+ def each_assigned_filter(&block)
59
+ super
60
+
61
+ @first_node.export.output_no_import = @no
62
+ end
63
+
64
+ def each_export_by(njob, mapper, &block)
65
+ @first_node_mutex.synchronize do
66
+ @no += 1
67
+ if @first_node
68
+ njob.export.output = @first_node.export.output
69
+ njob.export.no = @no - 1
70
+ else
71
+ super
72
+ @first_node = njob
73
+ end
74
+ end
75
+ end
76
+
77
+ def bind_export(exp, imp)
78
+ # do nothing.
79
+ end
80
+
81
+
82
+ # def each_export(&block)
83
+
84
+ # no = 0
85
+ # first_node = nil
86
+ # each_node do |node|
87
+ # no += 1
88
+ # if first_node
89
+ # node.export.output = first_node.export.output
90
+ # else
91
+ # first_node = node
92
+ # block.call node.export, node
93
+ # end
94
+ # end
95
+ # first_node.export.output_no_import = no
96
+ # end
97
+
98
+ end
99
+
100
+ class CWideInject<CIOFilter
101
+ def initialize(controller, opts, block_source)
102
+ super
103
+ @block_source = block_source
104
+ end
105
+
106
+ def node_class_name
107
+ "PWideInject"
108
+ end
109
+
110
+ def njob_creation_params
111
+ [@block_source]
112
+ end
113
+
114
+ def value
115
+ each_node{|node| return node.value}
116
+ end
117
+ end
118
+ end
119
+ end
@@ -0,0 +1,46 @@
1
+ # encoding: UTF-8
2
+ #
3
+ # Copyright (C) 2007-2010 Rakuten, Inc.
4
+ #
5
+
6
+ require "uri"
7
+
8
+ require "fairy/master/c-input"
9
+ require "fairy/share/vfile"
10
+ require "fairy/share/file-place"
11
+
12
+ module Fairy
13
+ class CInputFile<CInput
14
+ Controller.def_export self
15
+
16
+ def CInputFile.open(controller, opts, descripter)
17
+ bfile = CInputFile.new(controller, opts)
18
+ bfile.open(desctipter)
19
+ bfile
20
+ end
21
+ DeepConnect.def_single_method_spec(self, "REF open(REF, VAL, VAL)")
22
+
23
+ def node_class_name
24
+ "PInputFile"
25
+ end
26
+
27
+ def open(vf)
28
+ @vfile = vf
29
+ @cfile_place = CFilePlace.new(@vfile)
30
+
31
+ start_create_nodes
32
+ end
33
+ DeepConnect.def_method_spec(self, "REF open(DVAL)")
34
+
35
+ def input
36
+ @cfile_place
37
+ end
38
+
39
+ # def create_and_add_node(processor, mapper)
40
+ # node = super
41
+ # node.open(mapper.input.path)
42
+ # node
43
+ # end
44
+
45
+ end
46
+ end