blazer.rb 5.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194
  1. require "csv"
  2. require "yaml"
  3. require "chartkick"
  4. require "safely/core"
  5. require "blazer/version"
  6. require "blazer/data_source"
  7. require "blazer/result"
  8. require "blazer/run_statement"
  9. require "blazer/adapters/base_adapter"
  10. require "blazer/adapters/athena_adapter"
  11. require "blazer/adapters/bigquery_adapter"
  12. require "blazer/adapters/drill_adapter"
  13. require "blazer/adapters/elasticsearch_adapter"
  14. require "blazer/adapters/mongodb_adapter"
  15. require "blazer/adapters/presto_adapter"
  16. require "blazer/adapters/sql_adapter"
  17. require "blazer/engine"
  18. module Blazer
  19. class Error < StandardError; end
  20. class TimeoutNotSupported < Error; end
  21. class << self
  22. attr_accessor :audit
  23. attr_reader :time_zone
  24. attr_accessor :user_name
  25. attr_writer :user_class
  26. attr_writer :user_method
  27. attr_accessor :before_action
  28. attr_accessor :from_email
  29. attr_accessor :cache
  30. attr_accessor :transform_statement
  31. attr_accessor :check_schedules
  32. attr_accessor :anomaly_checks
  33. attr_accessor :async
  34. attr_accessor :images
  35. attr_accessor :query_editable
  36. end
  37. self.audit = true
  38. self.user_name = :name
  39. self.check_schedules = ["5 minutes", "1 hour", "1 day"]
  40. self.anomaly_checks = false
  41. self.async = false
  42. self.images = false
  43. TIMEOUT_MESSAGE = "Query timed out :("
  44. TIMEOUT_ERRORS = [
  45. "canceling statement due to statement timeout", # postgres
  46. "canceling statement due to conflict with recovery", # postgres
  47. "cancelled on user's request", # redshift
  48. "canceled on user's request", # redshift
  49. "system requested abort", # redshift
  50. "maximum statement execution time exceeded" # mysql
  51. ]
  52. BELONGS_TO_OPTIONAL = {}
  53. BELONGS_TO_OPTIONAL[:optional] = true if Rails::VERSION::MAJOR >= 5
  54. def self.time_zone=(time_zone)
  55. @time_zone = time_zone.is_a?(ActiveSupport::TimeZone) ? time_zone : ActiveSupport::TimeZone[time_zone.to_s]
  56. end
  57. def self.user_class
  58. if !defined?(@user_class)
  59. @user_class = Blazer.settings.key?("user_class") ? Blazer.settings["user_class"] : (User rescue nil)
  60. end
  61. @user_class
  62. end
  63. def self.user_method
  64. if !defined?(@user_method)
  65. @user_method ||= Blazer.settings["user_method"]
  66. if user_class
  67. @user_method ||= "current_#{Blazer.user_class.to_s.downcase.singularize}"
  68. end
  69. end
  70. @user_method
  71. end
  72. def self.settings
  73. @settings ||= begin
  74. path = Rails.root.join("config", "blazer.yml").to_s
  75. if File.exist?(path)
  76. YAML.load(ERB.new(File.read(path)).result)
  77. else
  78. {}
  79. end
  80. end
  81. end
  82. def self.data_sources
  83. @data_sources ||= begin
  84. ds = Hash[
  85. settings["data_sources"].map do |id, s|
  86. [id, Blazer::DataSource.new(id, s)]
  87. end
  88. ]
  89. ds.default = ds.values.first
  90. ds
  91. # TODO Blazer 2.0
  92. # ds2 = Hash.new { |hash, key| raise Blazer::Error, "Unknown data source: #{key}" }
  93. # ds.each do |k, v|
  94. # ds2[k] = v
  95. # end
  96. # ds2
  97. end
  98. end
  99. def self.extract_vars(statement)
  100. # strip commented out lines
  101. # and regex {1} or {1,2}
  102. statement.gsub(/\-\-.+/, "").gsub(/\/\*.+\*\//m, "").scan(/\{\w*?\}/i).map { |v| v[1...-1] }.reject { |v| /\A\d+(\,\d+)?\z/.match(v) || v.empty? }.uniq
  103. end
  104. def self.run_checks(schedule: nil)
  105. checks = Blazer::Check.includes(:query)
  106. checks = checks.where(schedule: schedule) if schedule
  107. checks.find_each do |check|
  108. next if check.state == "disabled"
  109. Safely.safely { run_check(check) }
  110. end
  111. end
  112. def self.run_check(check)
  113. rows = nil
  114. error = nil
  115. tries = 1
  116. ActiveSupport::Notifications.instrument("run_check.blazer", check_id: check.id, query_id: check.query.id, state_was: check.state) do |instrument|
  117. # try 3 times on timeout errors
  118. data_source = data_sources[check.query.data_source]
  119. statement = check.query.statement
  120. Blazer.transform_statement.call(data_source, statement) if Blazer.transform_statement
  121. while tries <= 3
  122. result = data_source.run_statement(statement, refresh_cache: true, check: check, query: check.query)
  123. if result.timed_out?
  124. Rails.logger.info "[blazer timeout] query=#{check.query.name}"
  125. tries += 1
  126. sleep(10)
  127. elsif result.error.to_s.start_with?("PG::ConnectionBad")
  128. data_source.reconnect
  129. Rails.logger.info "[blazer reconnect] query=#{check.query.name}"
  130. tries += 1
  131. sleep(10)
  132. else
  133. break
  134. end
  135. end
  136. check.reload # in case state has changed since job started
  137. check.update_state(result)
  138. # TODO use proper logfmt
  139. Rails.logger.info "[blazer check] query=#{check.query.name} state=#{check.state} rows=#{result.rows.try(:size)} error=#{result.error}"
  140. instrument[:statement] = statement
  141. instrument[:data_source] = data_source
  142. instrument[:state] = check.state
  143. instrument[:rows] = result.rows.try(:size)
  144. instrument[:error] = result.error
  145. instrument[:tries] = tries
  146. end
  147. end
  148. def self.send_failing_checks
  149. emails = {}
  150. Blazer::Check.includes(:query).where(state: ["failing", "error", "timed out", "disabled"]).find_each do |check|
  151. check.split_emails.each do |email|
  152. (emails[email] ||= []) << check
  153. end
  154. end
  155. emails.each do |email, checks|
  156. Safely.safely do
  157. Blazer::CheckMailer.failing_checks(email, checks).deliver_now
  158. end
  159. end
  160. end
  161. def self.adapters
  162. @adapters ||= {}
  163. end
  164. def self.register_adapter(name, adapter)
  165. adapters[name] = adapter
  166. end
  167. end
  168. Blazer.register_adapter "drill", Blazer::Adapters::DrillAdapter
  169. Blazer.register_adapter "bigquery", Blazer::Adapters::BigQueryAdapter
  170. Blazer.register_adapter "athena", Blazer::Adapters::AthenaAdapter
  171. Blazer.register_adapter "elasticsearch", Blazer::Adapters::ElasticsearchAdapter
  172. Blazer.register_adapter "mongodb", Blazer::Adapters::MongodbAdapter
  173. Blazer.register_adapter "presto", Blazer::Adapters::PrestoAdapter
  174. Blazer.register_adapter "sql", Blazer::Adapters::SqlAdapter