ich derailed_benchmark gem bin mit meiner App-Leistung zu verfolgen:Postgres Performance-Problem in meiner Rails app
$ PATH_TO_HIT="/api/v2/feed.json?per_page=30&page=1&category_name=Feed" USER_SERVER=webrick TEST_COUNT=20 bundle exec derailed exec perf:stackprof
==================================
Mode: cpu(1000)
Samples: 20708 (0.42% miss rate)
GC: 3219 (15.54%)
==================================
TOTAL (pct) SAMPLES (pct) FRAME
4720 (22.8%) 4694 (22.7%) block in ActiveRecord::ConnectionAdapters::PostgreSQLAdapter#exec_cache
542 (2.6%) 502 (2.4%) ActiveSupport::Inflector#underscore
413 (2.0%) 413 (2.0%) ActiveSupport::PerThreadRegistry#instance
364 (1.8%) 364 (1.8%) ActiveRecord::QueryMethods#validate_order_args
309 (1.5%) 309 (1.5%) block in ActiveSupport::Inflector#apply_inflections
282 (1.4%) 282 (1.4%) ThreadSafe::NonConcurrentCacheBackend#[]
257 (1.2%) 257 (1.2%) ActiveRecord::Relation#initialize
410 (2.0%) 235 (1.1%) ActiveRecord::Relation#initialize_copy
229 (1.1%) 229 (1.1%) ActiveRecord::Delegation::DelegateCache#relation_delegate_class
212 (1.0%) 212 (1.0%) block in ActiveRecord::Relation::Merger#merge
562 (2.7%) 198 (1.0%) ActiveRecord::QueryMethods#preprocess_order_args
190 (0.9%) 189 (0.9%) ActiveRecord::Core::ClassMethods#arel_table
181 (0.9%) 181 (0.9%) JSON#parse
175 (0.8%) 175 (0.8%) ActiveRecord::Relation#reset
165 (0.8%) 165 (0.8%) ActiveRecord::Attribute#initialize
153 (0.7%) 153 (0.7%) ActiveRecord::Relation#values
151 (0.7%) 151 (0.7%) ActiveRecord::Inheritance::ClassMethods#base_class
333 (1.6%) 151 (0.7%) ActiveRecord::Scoping::Default::ClassMethods#build_default_scope
144 (0.7%) 144 (0.7%) Skylight::Normalizers::ActiveRecord::SQL#extract_rust
142 (0.7%) 142 (0.7%) ActiveRecord::QueryMethods#joins_values
138 (0.7%) 138 (0.7%) block (4 levels) in Class#class_attribute
195 (0.9%) 133 (0.6%) ActiveRecord::DynamicMatchers#respond_to?
158 (0.8%) 121 (0.6%) ActiveRecord::QueryMethods#where_values=
125 (0.6%) 115 (0.6%) ActiveRecord::Reflection::AssociationReflection#klass
113 (0.5%) 113 (0.5%) ActiveRecord::Result#initialize_copy
110 (0.5%) 110 (0.5%) Arel::Table#initialize
193 (0.9%) 109 (0.5%) ActiveRecord::ConnectionAdapters::PostgreSQL::Utils#extract_schema_qualified_name
114 (0.6%) 106 (0.5%) Arel::Nodes::Binary#hash
104 (0.5%) 104 (0.5%) ActiveRecord::QueryMethods#extending_values
99 (0.5%) 99 (0.5%) ActiveRecord::QueryMethods#order_values
Wie kann ich den "Block in Activerecord :: ConnectionAdapters :: PostgreSQLAdapter # exec_cache" beheben Performance-Problem?
UPDATE
Nach demselben Befehl ausgeführt wird mit „config.middleware.delete "Active :: Querycache" in meiner config/application.rb die Ergebnisse sind:
==================================
Mode: cpu(1000)
Samples: 21116 (0.42% miss rate)
GC: 2213 (10.48%)
==================================
TOTAL (pct) SAMPLES (pct) FRAME
5619 (26.6%) 5600 (26.5%) block in ActiveRecord::ConnectionAdapters::PostgreSQLAdapter#exec_cache
2268 (10.7%) 2268 (10.7%) block in ActiveRecord::ConnectionAdapters::PostgreSQLAdapter#exec_no_cache
421 (2.0%) 383 (1.8%) ActiveSupport::Inflector#underscore
304 (1.4%) 304 (1.4%) ActiveSupport::PerThreadRegistry#instance
294 (1.4%) 294 (1.4%) block in ActiveSupport::Inflector#apply_inflections
270 (1.3%) 270 (1.3%) ThreadSafe::NonConcurrentCacheBackend#[]
245 (1.2%) 245 (1.2%) ActiveRecord::Relation#initialize
229 (1.1%) 229 (1.1%) ActiveRecord::QueryMethods#validate_order_args
219 (1.0%) 219 (1.0%) ActiveRecord::Delegation::DelegateCache#relation_delegate_class
207 (1.0%) 207 (1.0%) ActiveRecord::Inheritance::ClassMethods#base_class
285 (1.3%) 188 (0.9%) ActiveRecord::Relation#initialize_copy
184 (0.9%) 184 (0.9%) ActiveRecord::Attribute#initialize
181 (0.9%) 179 (0.8%) ActiveRecord::Core::ClassMethods#arel_table
175 (0.8%) 175 (0.8%) Skylight::Normalizers::ActiveRecord::SQL#extract_rust
165 (0.8%) 165 (0.8%) block in ActiveRecord::Relation::Merger#merge
147 (0.7%) 147 (0.7%) block (4 levels) in Class#class_attribute
374 (1.8%) 145 (0.7%) ActiveRecord::QueryMethods#preprocess_order_args
113 (0.5%) 113 (0.5%) ActiveRecord::Relation#values
112 (0.5%) 112 (0.5%) ActiveRecord::QueryMethods#joins_values
171 (0.8%) 109 (0.5%) ActiveRecord::ConnectionAdapters::PostgreSQL::Utils#extract_schema_qualified_name
99 (0.5%) 99 (0.5%) Arel::Table#initialize
97 (0.5%) 97 (0.5%) ActiveRecord::Relation#reset
271 (1.3%) 96 (0.5%) ActiveRecord::Scoping::Default::ClassMethods#build_default_scope
107 (0.5%) 95 (0.4%) ActiveRecord::Reflection::AssociationReflection#klass
93 (0.4%) 93 (0.4%) ActiveRecord::QueryMethods#order_values
125 (0.6%) 93 (0.4%) ActiveRecord::QueryMethods#where_values=
88 (0.4%) 88 (0.4%) ActiveRecord::Reflection::ThroughReflection#active_record
106 (0.5%) 87 (0.4%) Skylight::Trace#start
81 (0.4%) 81 (0.4%) ActiveRecord::QueryMethods#check_cached_relation
80 (0.4%) 80 (0.4%) ActiveRecord::QueryMethods#where_values
UDPATE 2
Nach dem Ausführen der Abfrage mit "Wandzeit" -Modus und nicht "CPU-Zeit" -Modus ist dies das Ergebnis:
==================================
Mode: wall(1000)
Samples: 41424 (1.92% miss rate)
GC: 3648 (8.81%)
==================================
TOTAL (pct) SAMPLES (pct) FRAME
4780 (11.5%) 4718 (11.4%) block in ActiveRecord::ConnectionAdapters::PostgreSQLAdapter#exec_cache
2783 (6.7%) 2783 (6.7%) block in ActiveRecord::ConnectionAdapters::PostgreSQLAdapter#exec_no_cache
1088 (2.6%) 1088 (2.6%) ActiveSupport::PerThreadRegistry#instance
934 (2.3%) 934 (2.3%) ThreadSafe::NonConcurrentCacheBackend#[]
1031 (2.5%) 926 (2.2%) ActiveSupport::Inflector#underscore
739 (1.8%) 739 (1.8%) block in ActiveSupport::Inflector#apply_inflections
626 (1.5%) 626 (1.5%) ActiveRecord::Relation#initialize
589 (1.4%) 589 (1.4%) ActiveRecord::Delegation::DelegateCache#relation_delegate_class
577 (1.4%) 577 (1.4%) ThreadSafe::NonConcurrentCacheBackend#get_or_default
549 (1.3%) 549 (1.3%) ActiveRecord::Attribute#initialize
497 (1.2%) 497 (1.2%) block in ActiveRecord::Relation::Merger#merge
497 (1.2%) 497 (1.2%) ActiveRecord::QueryMethods#validate_order_args
702 (1.7%) 424 (1.0%) ActiveRecord::Relation#initialize_copy
419 (1.0%) 417 (1.0%) ActiveRecord::Core::ClassMethods#arel_table
384 (0.9%) 384 (0.9%) ActiveRecord::Inheritance::ClassMethods#base_class
383 (0.9%) 383 (0.9%) block (4 levels) in Class#class_attribute
358 (0.9%) 358 (0.9%) Skylight::Normalizers::ActiveRecord::SQL#extract_rust
329 (0.8%) 329 (0.8%) ActiveRecord::Base.logger
321 (0.8%) 321 (0.8%) rescue in Net::BufferedIO#rbuf_fill
315 (0.8%) 315 (0.8%) ActiveRecord::Core#update_attributes_from_transaction_state
314 (0.8%) 314 (0.8%) ActiveRecord::ConnectionAdapters::AbstractAdapter#type_map
795 (1.9%) 298 (0.7%) ActiveRecord::QueryMethods#preprocess_order_args
284 (0.7%) 284 (0.7%) Arel::Table#initialize
279 (0.7%) 279 (0.7%) ActiveRecord::Relation#values
278 (0.7%) 278 (0.7%) ActiveRecord::Relation#reset
734 (1.8%) 264 (0.6%) ActiveRecord::Scoping::Default::ClassMethods#build_default_scope
263 (0.6%) 263 (0.6%) ActiveRecord::QueryMethods#joins_values
394 (1.0%) 258 (0.6%) ActiveRecord::ConnectionAdapters::PostgreSQL::Utils#extract_schema_qualified_name
15323 (37.0%) 249 (0.6%) ActiveRecord::Querying#find_by_sql
257 (0.6%) 246 (0.6%) ActiveRecord::Reflection::AssociationReflection#klass
Ich baue meine feed.json index.json.jbuilder verwenden, das ist, wie es aussieht:
json.battles @battles do |battle|
if (battle.products.size == 2)
battle_results = battle.calculate_results
json.(battle, :id)
vote = battle.votes.find_by(user_id: current_user.id)
json.voted vote.present?
if vote
json.product_voted vote.product.id == battle.products[0].id ? "first" : "second"
end
json.mybattle battle.try(:user).try(:id) == current_user.id
json.user do
username = ""
if (battle.try(:user).try(:nickname).present?)
username = battle.try(:user).try(:nickname)
else
username = battle.try(:user).try(:name).try(:downcase).try(:delete,' ')
end
json.username username
json.user_id battle.try(:user_id)
json.profile_image battle.try(:user).try(:image) || ""
json.full_name battle.try(:user).try(:name) || ""
end
json.votes battle_results[:votes]
json.created_at time_ago_in_words(battle.created_at) + " ago"
json.title battle.title
json.first_product do
first_product = battle.products[0]
json.id first_product.id
json.voted first_product.votes.find_by(user_id: current_user.id).present?
json.percentage battle_results[:percentage_product_one]
# json.percentage_after_voting battle_results[:percentage_after_voting_product_one]
json.name first_product.name
json.price SearchFunctions.convert_currency(first_product.price.to_s, current_user.currency_code, 'USD')
# json.price first_product.price.to_s
json.url first_product.url
if first_product.images["sub"] && first_product.images["sub"].kind_of?(Array)
first_product.images["sub"] = first_product.images["sub"].first(10)
end
json.images first_product.images
json.manufacturer first_product.manufacturer
json.description first_product.description
json.is_user_saved first_product.saved_products.find_by(user_id: current_user.id).present?
json.saved_count first_product.saved_products.length
end
json.second_product do
second_product = battle.products[1]
json.id second_product.id
json.voted second_product.votes.find_by(user_id: current_user.id).present?
json.percentage battle_results[:percentage_product_two]
# json.percentage_after_voting battle_results[:percentage_after_voting_product_two]
json.name second_product.name
json.price SearchFunctions.convert_currency(second_product.price.to_s, current_user.currency_code, 'USD')
# json.price second_product.price.to_s
json.url second_product.url
if second_product.images["sub"] && second_product.images["sub"].kind_of?(Array)
second_product.images["sub"] = second_product.images["sub"].first(10)
end
json.images second_product.images
json.manufacturer second_product.manufacturer
json.description second_product.description
json.is_user_saved second_product.saved_products.find_by(user_id: current_user.id).present?
json.saved_count second_product.saved_products.length
end
end
end
Können Sie Ihren Test mit Query-Caching deaktiviert ausführen? Fügen Sie dies zu Ihrer config/application.rb 'config.middleware.delete" ActiveRecord :: QueryCache "' –
Es ist ein bisschen komisch, dass Sie jetzt mehr Samples für zwischengespeicherte Abfragen als zuvor haben. Eine andere Frage verwendest du Mac OS X? Wenn ja, sollten Sie die Wandzeit und nicht die CPU-Zeit messen. –
Ich benutze Mac OS x, wie kann ich die Wandzeit messen? – gal