Fixing Performance and Memory Problems (RubyWine)

Transcript

1. Fixing performance & Memory problems @fglc2 / dressipi
2. Part 1: The follow events happen in real-time
3. Realtime recommender
4. Realtime recommender Batch Predictions
5. Realtime recommender Batch Predictions Event Stream Recommender Realtime Predictions
6. Realtime recommender Batch Predictions Event Stream Recommender Realtime Predictions Results
7. We deployed to production
9. API Combine Serialize Batch Predictions Realtime Predictions
12. class ScoreMap def serialize to_int_array.pack("l<*") end def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1).to_i] end end def scale_factor @scores.values.max_by {|n| n.abs } .abs / 1000 end end
13. class ScoreMap def serialize to_int_array.pack("l<*") end def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1)).to_i] end end def scale_factor @scores.values.max_by {|n| n.abs }
14. class ScoreMap def serialize to_int_array.pack("l<*") end def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1)).to_i] end end def scale_factor @scores.values.max_by {|n| n.abs } .abs / 1000
15. def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1)).to_i] end end def scale_factor @scores.values.max_by {|n| n.abs } .abs / 1000 end end
16. Ruby-Prof is a profiler • Records behaviour of your code • Produces reports on where your code spends time • Use to analyse problems not detect them
18. require 'ruby-prof'
19. require 'ruby-prof' scores = load_example_data
20. require 'ruby-prof' scores = load_example_data result = RubyProf.profile do ScoreMap.new(scores).serialize end
21. require 'ruby-prof' scores = load_example_data result = RubyProf.profile do ScoreMap.new(scores).serialize end printer = RubyProf::GraphHtmlPrinter.new(result) File.open('out.html', 'w') do |file| printer.print(file) end
29. def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1)).to_i] end end def scale_factor @scores.values.max_by {|n| n.abs } .abs / 1000 end
30. def to_int_array @scores.flat_map do |key, value| [ key, (value/(scale_factor + 1)).to_i] end end def scale_factor @scale_factor ||= @scores.values.max_by {|n| n.abs } .abs / 1000 end
32. Slows down your code
33. Can distort things
34. ruby-prof is exact
35. Sampling profilers
36. Sampling profilers • Sampling profilers monitor (sample) call stack periodically and reports how often each method seen
37. Sampling profilers • Sampling profilers monitor (sample) call stack periodically and reports how often each method seen • Sampling frequency controls accuracy vs overhead tradeoff
38. Stackprof StackProf.run(mode: :cpu, out: 'stackprof.dump') do ScoreMap.new(scores).serialize end
39. Stackprof StackProf.run(mode: :cpu, out: 'stackprof.dump') do ScoreMap.new(scores).serialize end
40. ================================== Mode: cpu(1000) Samples: 435 (0.00% miss rate) GC: 2 (0.46%) ================================== TOTAL (pct) SAMPLES (pct) FRAME 433 (99.5%) 433 (99.5%) ScoreMap#scale_factor 2 (0.5%) 2 (0.5%) (garbage collection) 433 (99.5%) 0 (0.0%) ScoreMap#to_int_array 433 (99.5%) 0 (0.0%) ScoreMap#serialize 433 (99.5%) 0 (0.0%) block in <main> 433 (99.5%) 0 (0.0%) <main> 433 (99.5%) 0 (0.0%) <main>
41. ================================== Mode: cpu(1000) Samples: 435 (0.00% miss rate) GC: 2 (0.46%) ================================== TOTAL (pct) SAMPLES (pct) FRAME 433 (99.5%) 433 (99.5%) ScoreMap#scale_factor 2 (0.5%) 2 (0.5%) (garbage collection) 433 (99.5%) 0 (0.0%) ScoreMap#to_int_array 433 (99.5%) 0 (0.0%) ScoreMap#serialize 433 (99.5%) 0 (0.0%) block in <main> 433 (99.5%) 0 (0.0%) <main> 433 (99.5%) 0 (0.0%) <main>
42. Overhead
43. Overhead • Code on its own: 6.05 s
44. Overhead • Code on its own: 6.05 s • Rubyprof: ~ 27s
45. Overhead • Code on its own: 6.05 s • Rubyprof: ~ 27s • Stackprof (default settings): 6.09s
46. Part 1 Summary
47. Part 1 Summary • Use monitoring (or angry users) to detect problems
48. Part 1 Summary • Use monitoring (or angry users) to detect problems • Refine with stackprof (if needed)
49. Part 1 Summary • Use monitoring (or angry users) to detect problems • Refine with stackprof (if needed) • Reproduce the problem in a small benchmark
50. Part 1 Summary • Use monitoring (or angry users) to detect problems • Refine with stackprof (if needed) • Reproduce the problem in a small benchmark • Use rubyprof to understand / fix the problem
51. Part 1 Summary • Use monitoring (or angry users) to detect problems • Refine with stackprof (if needed) • Reproduce the problem in a small benchmark • Use rubyprof to understand / fix the problem • Profit!
52. Part 2: Memory
53. • Bitmap GC (1.9) • Generational GC (2.1) • Incremental GC (2.2) • Symbol GC (2.2) • Transient Heap (2.6) • Compacting GC ( 2.7 )
54. You still have to worry about memory
55. NoMemoryError: failed to allocate
57. 0 100 200 300 400
58. Memory leak
59. Where do they come from
60. Where do they come from • Leaks in native code (C extensions)
61. Where do they come from • Leaks in native code (C extensions) • Ruby thinks you are still using it (long lived objects, caches, etc.)
62. Where do they come from • Leaks in native code (C extensions) • Ruby thinks you are still using it (long lived objects, caches, etc.) • Not understanding object lifecycles
63. class LeakingMiddleware def initialize(app) @app = app end def call(env) ... end def cached_thing(id) @things ||= {} @things[id] ||= Thing.find id end end
64. class LeakingMiddleware def initialize(app) @app = app end def call(env) ... end def cached_thing(id) @things ||= {} @things[id] ||= Thing.find id end end
65. Leaks are scary because
66. Leaks are scary because • Manifest as global problem
67. Leaks are scary because • Manifest as global problem • Memory usage is complicated and noisy (caches, threads, lazy load)
68. Leaks are scary because • Manifest as global problem • Memory usage is complicated and noisy (caches, threads, lazy load) • Easily hidden in a dependency (see bundler-leak)
69. Back to the problem
70. At least a year
71. At least a year • Slow or rare leak ~ 2 weeks to exhaust memory at 3M requests / day
72. At least a year • Slow or rare leak ~ 2 weeks to exhaust memory at 3M requests / day • App usually restarted often enough (deploys, autoscaling)
73. At least a year • Slow or rare leak ~ 2 weeks to exhaust memory at 3M requests / day • App usually restarted often enough (deploys, autoscaling) • Failed instances automatically detected as unhealthy and replaced
75. But seriously
76. I couldn't reproduce the leak
77. What could be happening?
78. What could be happening? • Not picking the right endpoint
79. What could be happening? • Not picking the right endpoint • Different usage patterns
80. What could be happening? • Not picking the right endpoint • Different usage patterns • Environment differences
81. Debugging in production
82. rbtrace
83. rbtrace • Count / trace method calls, GCs
84. rbtrace • Count / trace method calls, GCs • Dump backtraces
85. rbtrace • Count / trace method calls, GCs • Dump backtraces • Execute arbitrary ruby
86. trace_object_allocations
87. trace_object_allocations • Part of Objectspace (in stdlib)
88. trace_object_allocations • Part of Objectspace (in stdlib) • Records details about each memory allocation (when/ where/what)
89. trace_object_allocations • Part of Objectspace (in stdlib) • Records details about each memory allocation (when/ where/what) • Dump to a file for analysis ( heap dump )
90. Note that this feature introduces a huge performance decrease and huge memory consumption.
91. The plan
92. The plan • Only work on one instance
93. The plan • Only work on one instance • Use rbtrace to dump heap from single rails process
94. The plan • Only work on one instance • Use rbtrace to dump heap from single rails process • Analyse heap dump offline
95. I had to google this!
96. bundle exec rbtrace --pid $PID -e ' Thread.new { require 'objspace' ObjectSpace.trace_object_allocations_start }' Start Tracing
97. bundle exec rbtrace --pid $PID -e ' Thread.new { require 'objspace' ObjectSpace.trace_object_allocations_start }' Start Tracing
98. Dump GC.start io=File.open("/tmp/heap.dump", "w"); ObjectSpace.dump_all(output: io); io.close
99. What's in a heap dump?
101. "address": "0x7ff5bbdfc3f8",
102. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8",
103. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8", "file": "...gems/mongoid-7.0.5/lib/mongoid/document.rb",
 "line": 335,
 "method": "allocate",
104. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8", "file": "...gems/mongoid-7.0.5/lib/mongoid/document.rb",
 "line": 335,
 "method": "allocate", "generation": 62,
105. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8", "file": "...gems/mongoid-7.0.5/lib/mongoid/document.rb",
 "line": 335,
 "method": "allocate", "generation": 62, "ivars": 3,
 "memsize": 40,
106. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8", "file": "...gems/mongoid-7.0.5/lib/mongoid/document.rb",
 "line": 335,
 "method": "allocate", "generation": 62, "ivars": 3,
 "memsize": 40, "references": [
 "0x7ff5bbe1c4f0"
 ],
107. "address": "0x7ff5bbdfc3f8", "type": "OBJECT",
 "class": "0x7ff5ba88b8e8", "file": "...gems/mongoid-7.0.5/lib/mongoid/document.rb",
 "line": 335,
 "method": "allocate", "generation": 62, "ivars": 3,
 "memsize": 40, "references": [
 "0x7ff5bbe1c4f0"
 ], "flags": {
 "wb_protected": true
 },
108. heapy
109. heapy $ heapy read good_heap.dump
110. heapy $ heapy read good_heap.dump Analyzing Heap
 ==============
111. heapy $ heapy read good_heap.dump Analyzing Heap
 ============== Generation: nil object count: 475742, mem: 0.0 kb
112. heapy $ heapy read good_heap.dump Analyzing Heap
 ============== Generation: nil object count: 475742, mem: 0.0 kb Generation: 61 object count: 5334, mem: 1782.5 kb
 Generation: 62 object count: 918, mem: 1324.8 kb
 Generation: 63 object count: 4, mem: 0.3 kb
113. heapy $ heapy read good_heap.dump Analyzing Heap
 ============== Generation: nil object count: 475742, mem: 0.0 kb Generation: 61 object count: 5334, mem: 1782.5 kb
 Generation: 62 object count: 918, mem: 1324.8 kb
 Generation: 63 object count: 4, mem: 0.3 kb Generation: 2222 object count: 1, mem: 0.0 kb
 Generation: 2223 object count: 420, mem: 44.6 kb
 Generation: 2224 object count: 635, mem: 42.8 kb
114. Analyzing Heap ============== Generation: nil object count: 498011, mem: 0.0 kb Generation: 61 object count: 834, mem: 82.2 kb Generation: 62 object count: 36, mem: 3.9 kb Generation: 63 object count: 9, mem: 0.8 kb Generation: 64 object count: 3, mem: 0.4 kb Generation: 65 object count: 23, mem: 1.8 kb Generation: 66 object count: 16, mem: 1.4 kb Generation: 67 object count: 12, mem: 1.2 kb ...
115. Analyzing Heap ============== Generation: nil object count: 498011, mem: 0.0 kb Generation: 61 object count: 834, mem: 82.2 kb Generation: 62 object count: 36, mem: 3.9 kb Generation: 63 object count: 9, mem: 0.8 kb Generation: 64 object count: 3, mem: 0.4 kb Generation: 65 object count: 23, mem: 1.8 kb Generation: 66 object count: 16, mem: 1.4 kb Generation: 67 object count: 12, mem: 1.2 kb ...
116. Generation: 218 object count: 11, mem: 1.2 kb Generation: 219 object count: 10, mem: 1.1 kb Generation: 220 object count: 12, mem: 1.2 kb Generation: 221 object count: 12, mem: 1.2 kb Generation: 222 object count: 13, mem: 9.2 kb Generation: 223 object count: 10, mem: 1.0 kb Generation: 224 object count: 12, mem: 1.2 kb Generation: 225 object count: 12, mem: 1.2 kb Generation: 226 object count: 9, mem: 0.9 kb Generation: 227 object count: 11, mem: 1.2 kb ...
117. Generation: 1874 object count: 8, mem: 0.9 kb Generation: 1875 object count: 8, mem: 1.0 kb Generation: 1876 object count: 10, mem: 1.0 kb Generation: 1877 object count: 6, mem: 0.7 kb Generation: 1878 object count: 8, mem: 1.0 kb Generation: 1879 object count: 10, mem: 1.0 kb Generation: 1880 object count: 6, mem: 0.7 kb Generation: 1881 object count: 8, mem: 1.0 kb Generation: 1882 object count: 9, mem: 0.8 kb Generation: 1883 object count: 6, mem: 0.7 kb ...
118. $ heapy read heap.dump 1879
119. $ heapy read heap.dump 1879 allocated by memory (978) (in bytes)
 ==============================
 858 .../action_view/template.rb:296
 120 .../active_support/core_ext/module/remove_method.rb:4
120. $ heapy read heap.dump 1879 allocated by memory (978) (in bytes)
 ==============================
 858 .../action_view/template.rb:296
 120 .../active_support/core_ext/module/remove_method.rb:4 _app_views_mobile_layouts_widgets_html_haml__3857206368_702097383 _app_views_mobile_widgets_outfits_html_haml__3715565700_702097206 _app_views_mobile_layouts_widgets_html_haml__3857206368_702096884
121. $ heapy read heap.dump 1879 allocated by memory (978) (in bytes)
 ==============================
 858 .../action_view/template.rb:296
 120 .../active_support/core_ext/module/remove_method.rb:4 _app_views_mobile_layouts_widgets_html_haml__3857206368_702097383 _app_views_mobile_widgets_outfits_html_haml__3715565700_702097206 _app_views_mobile_layouts_widgets_html_haml__3857206368_702096884 "flags":{"wb_protected":true, "old":true, "uncollectible":true }
122. 2 Questions
123. 2 Questions • Why is the method name uncollectible?
124. 2 Questions • Why is the method name uncollectible? • Why so many of them?
125. Ruby 2.2: Symbols can be garbage collected
126. Ruby 2.2: Symbols can be garbage collected *Terms and conditions may apply Your statutory rights are not affected
127. 2 Kinds of Symbols
128. 2 Kinds of Symbols • Mortal symbols: created from foo.to_sym
129. 2 Kinds of Symbols • Mortal symbols: created from foo.to_sym • Immortal symbols: C-level references (rb_intern)
130. 2 Kinds of Symbols • Mortal symbols: created from foo.to_sym • Immortal symbols: C-level references (rb_intern) Defining a method creates an immortal symbol
131. Action View template caching
132. Action View template caching • Each template is compiled to a method with random name
133. Action View template caching • Each template is compiled to a method with random name • Each entry in the view search path is a resolver
134. Action View template caching • Each template is compiled to a method with random name • Each entry in the view search path is a resolver • Resolvers return compiled template from cache or load from disk
135. before_action :set_view_path def set_view_path if mobile_site_requested? prepend_view_path "app/views/mobile" end end
136. before_action :set_view_path def set_view_path if mobile_site_requested? prepend_view_path "app/views/mobile" end end
137. before_action :set_view_path def set_view_path if mobile_site_requested? prepend_view_path "app/views/mobile" end end
138. devise/app/views app/views View Paths
139. devise/app/views app/views View Paths
140. devise/app/views app/views View Paths app/views/mobile
141. devise/app/views app/views View Paths app/views/mobile
142. devise/app/views app/views View Paths
143. devise/app/views app/views View Paths app/views/mobile
144. New (empty) cache on each request!
145. Each request leaks a symbol
148. Part 2 Summary
149. Part 2 Summary • Don't be afraid of production!
150. Part 2 Summary • Don't be afraid of production! • Understand what/how using memory with heap dumps
151. Part 2 Summary • Don't be afraid of production! • Understand what/how using memory with heap dumps • Pay attention to your objects' lifecycles
152. Thank you! @fglc2 / dressipi
153. Further Reading • https://samsaffron.com/archive/2015/03/31/debugging-memory- leaks-in-ruby • https://samsaffron.com/archive/2019/10/08/debugging-unmanaged- and-hidden-memory-leaks-in-ruby • https://blog.codeship.com/tracking-object-allocations-in-ruby/ • https://blog.codeship.com/debugging-a-memory-leak-on-heroku/ • https://www.slideshare.net/authorNari/symbol-gc • https://github.com/schneems/derailed_benchmarks • https://www.schneems.com/2019/11/07/why-does-my-apps- memory-usage-grow-asymptotically-over-time/