Merge branch 'mwaw/test_service_ping_queries_agains_values_351085' into 'master'

Build a test to compare service ping output with output of the values from SQL generated

See merge request gitlab-org/gitlab!81002

lib/gitlab/usage_data_queries.rb

@@ -41,9 +41,19 @@ module Gitlab
       end
 
       def maximum_id(model, column = nil)
+        # no-op: shadowing super for performance reasons
       end
 
       def minimum_id(model, column = nil)
+        # no-op: shadowing super for performance reasons
+      end
+
+      def alt_usage_data(value = nil, fallback: FALLBACK, &block)
+        if block_given?
+          { alt_usage_data_block: block.to_s }
+        else
+          { alt_usage_data_value: value }
+        end
       end
 
       def redis_usage_data(counter = nil, &block)

spec/lib/gitlab/usage/service_ping_report_spec.rb

@@ -3,6 +3,8 @@
 require 'spec_helper'
 
 RSpec.describe Gitlab::Usage::ServicePingReport, :use_clean_rails_memory_store_caching do
+  include UsageDataHelpers
+
   let(:usage_data) { { uuid: "1111", counts: { issue: 0 } } }
 
   context 'when feature merge_service_ping_instrumented_metrics enabled' do
@@ -121,4 +123,97 @@ RSpec.describe Gitlab::Usage::ServicePingReport, :use_clean_rails_memory_store_c
       end
     end
   end
+
+  context 'cross test values against queries' do
+    # TODO: fix failing metrics https://gitlab.com/gitlab-org/gitlab/-/issues/353559
+    let(:failing_todo_metrics) do
+      ["counts.labels",
+       "counts.jira_imports_total_imported_issues_count",
+       "counts.in_product_marketing_email_create_0_sent",
+       "counts.in_product_marketing_email_create_0_cta_clicked",
+       "counts.in_product_marketing_email_create_1_sent",
+       "counts.in_product_marketing_email_create_1_cta_clicked",
+       "counts.in_product_marketing_email_create_2_sent",
+       "counts.in_product_marketing_email_create_2_cta_clicked",
+       "counts.in_product_marketing_email_verify_0_sent",
+       "counts.in_product_marketing_email_verify_0_cta_clicked",
+       "counts.in_product_marketing_email_verify_1_sent",
+       "counts.in_product_marketing_email_verify_1_cta_clicked",
+       "counts.in_product_marketing_email_verify_2_sent",
+       "counts.in_product_marketing_email_verify_2_cta_clicked",
+       "counts.in_product_marketing_email_trial_0_sent",
+       "counts.in_product_marketing_email_trial_0_cta_clicked",
+       "counts.in_product_marketing_email_trial_1_sent",
+       "counts.in_product_marketing_email_trial_1_cta_clicked",
+       "counts.in_product_marketing_email_trial_2_sent",
+       "counts.in_product_marketing_email_trial_2_cta_clicked",
+       "counts.in_product_marketing_email_team_0_sent",
+       "counts.in_product_marketing_email_team_0_cta_clicked",
+       "counts.in_product_marketing_email_team_1_sent",
+       "counts.in_product_marketing_email_team_1_cta_clicked",
+       "counts.in_product_marketing_email_team_2_sent",
+       "counts.in_product_marketing_email_team_2_cta_clicked",
+       "counts.in_product_marketing_email_experience_0_sent",
+       "counts.in_product_marketing_email_team_short_0_sent",
+       "counts.in_product_marketing_email_team_short_0_cta_clicked",
+       "counts.in_product_marketing_email_trial_short_0_sent",
+       "counts.in_product_marketing_email_trial_short_0_cta_clicked",
+       "counts.in_product_marketing_email_admin_verify_0_sent",
+       "counts.in_product_marketing_email_admin_verify_0_cta_clicked",
+       "counts.ldap_users",
+       "usage_activity_by_stage.create.projects_with_sectional_code_owner_rules",
+       "usage_activity_by_stage.monitor.clusters_integrations_prometheus",
+       "usage_activity_by_stage.monitor.projects_with_enabled_alert_integrations_histogram",
+       "usage_activity_by_stage_monthly.create.projects_with_sectional_code_owner_rules",
+       "usage_activity_by_stage_monthly.monitor.clusters_integrations_prometheus"]
+    end
+
+    def fetch_value_by_query(query)
+      # Because test cases are run inside a transaction, if any query raise and error all queries that follows
+      # it are automatically canceled by PostgreSQL, to avoid that problem, and to provide exhaustive information
+      # about every metric, queries are wrapped explicitly in sub transactions.
+      ApplicationRecord.transaction do
+        ApplicationRecord.connection.execute(query)&.first&.values&.first
+      end
+    rescue ActiveRecord::StatementInvalid => e
+      e.message
+    end
+
+    def build_payload_from_queries(payload, accumulator = [], key_path = [])
+      payload.each do |key, value|
+        if value.is_a?(Hash)
+          build_payload_from_queries(value, accumulator, key_path.dup << key)
+        elsif value.is_a?(String) && /SELECT .* FROM.*/ =~ value
+          accumulator << [key_path.dup << key, value, fetch_value_by_query(value)]
+        end
+      end
+      accumulator
+    end
+
+    before do
+      stub_usage_data_connections
+      stub_object_store_settings
+      stub_prometheus_queries
+      memoized_constatns = Gitlab::UsageData::CE_MEMOIZED_VALUES
+      memoized_constatns += Gitlab::UsageData::EE_MEMOIZED_VALUES if defined? Gitlab::UsageData::EE_MEMOIZED_VALUES
+      memoized_constatns.each { |v| Gitlab::UsageData.clear_memoization(v) }
+      stub_database_flavor_check('Cloud SQL for PostgreSQL')
+    end
+
+    let(:service_ping_payload) { described_class.for(output: :all_metrics_values) }
+    let(:metrics_queries_with_values) { build_payload_from_queries(described_class.for(output: :metrics_queries)) }
+
+    it 'generates queries that match collected data', :aggregate_failures do
+      message = "Expected %{query} result to match %{value} for %{key_path} metric"
+
+      metrics_queries_with_values.each do |key_path, query, value|
+        next if failing_todo_metrics.include?(key_path.join('.'))
+
+        expect(value).to(
+          eq(service_ping_payload.dig(*key_path)),
+          message % { query: query, value: (value || 'NULL'), key_path: key_path.join('.') }
+        )
+      end
+    end
+  end
 end

spec/lib/gitlab/usage_data_queries_spec.rb

@@ -45,6 +45,19 @@ RSpec.describe Gitlab::UsageDataQueries do
     end
   end
 
+  describe '.alt_usage_data' do
+    subject(:alt_usage_data) { described_class.alt_usage_data { 42 } }
+
+    it 'returns value when used with value' do
+      expect(described_class.alt_usage_data(1))
+        .to eq(alt_usage_data_value: 1)
+    end
+
+    it 'returns a stringified block for alt_usage_data with a block' do
+      expect(alt_usage_data[:alt_usage_data_block]).to start_with('#<Proc:')
+    end
+  end
+
   describe '.sum' do
     it 'returns the raw SQL' do
       expect(described_class.sum(Issue, :weight)).to eq('SELECT SUM("issues"."weight") FROM "issues"')