Merge branch '299448-add-base-class-for-metric-instrumentation' into 'master'

Add base framework and tooling for Usage Data metrics instrumentation

See merge request gitlab-org/gitlab!59649

config/metrics/license/20210201124933_uuid.yml

@@ -11,6 +11,7 @@ milestone: "9.1"
 introduced_by_url: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/1521
 time_frame: none
 data_source: database
+instrumentation_class: 'Gitlab::Usage::Metrics::Instrumentations::UuidMetric'
 distribution:
 - ee
 - ce

config/metrics/schema.json

@@ -51,6 +51,10 @@
       "type": "string",
       "enum": ["database", "redis", "redis_hll", "prometheus", "ruby"]
     },
+    "instrumentation_class": {
+      "type": "string",
+      "pattern": "^(Gitlab::Usage::Metrics::Instrumentations::)(([A-Z][a-z]+)+::)*(([A-Z][a-z]+)+)$"
+    },
     "distribution": {
       "type": "array",
       "items": {

lib/gitlab/usage/metric_definition.rb

@@ -65,6 +65,10 @@ module Gitlab
           @definitions ||= load_all!
         end
 
+        def all
+          @all ||= definitions.map { |_key_path, definition| definition }
+        end
+
         def schemer
           @schemer ||= ::JSONSchemer.schema(Pathname.new(METRIC_SCHEMA_PATH))
         end

lib/gitlab/usage/metrics/instrumentations/base_metric.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      module Instrumentations
+        class BaseMetric
+          include Gitlab::Utils::UsageData
+
+          attr_reader :time_frame
+
+          def initialize(time_frame:)
+            @time_frame = time_frame
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage/metrics/instrumentations/database_metric.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      module Instrumentations
+        class DatabaseMetric < BaseMetric
+          # Usage Example
+          #
+          # class CountUsersCreatingIssuesMetric < DatabaseMetric
+          #   operation :distinct_count, column: :author_id
+          #
+          #   relation do |database_time_constraints|
+          #     ::Issue.where(database_time_constraints)
+          #   end
+          # end
+          class << self
+            def start(&block)
+              @metric_start = block
+            end
+
+            def finish(&block)
+              @metric_finish = block
+            end
+
+            def relation(&block)
+              @metric_relation = block
+            end
+
+            def operation(symbol, column: nil)
+              @metric_operation = symbol
+              @column = column
+            end
+
+            attr_reader :metric_operation, :metric_relation, :metric_start, :metric_finish, :column
+          end
+
+          def value
+            method(self.class.metric_operation)
+              .call(relation,
+                    self.class.column,
+                    start: self.class.metric_start&.call,
+                    finish: self.class.metric_finish&.call)
+          end
+
+          def relation
+            self.class.metric_relation.call.where(time_constraints)
+          end
+
+          private
+
+          def time_constraints
+            case time_frame
+            when '28d'
+              { created_at: 30.days.ago..2.days.ago }
+            when 'all'
+              {}
+            when 'none'
+              nil
+            else
+              raise "Unknown time frame: #{time_frame} for DatabaseMetric"
+            end
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage/metrics/instrumentations/generic_metric.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      module Instrumentations
+        class GenericMetric < BaseMetric
+          # Usage example
+          #
+          # class UuidMetric < GenericMetric
+          #   value do
+          #     Gitlab::CurrentSettings.uuid
+          #   end
+          # end
+          class << self
+            def value(&block)
+              @metric_value = block
+            end
+
+            attr_reader :metric_value
+          end
+
+          def value
+            alt_usage_data do
+              self.class.metric_value.call
+            end
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage/metrics/instrumentations/redis_hll_metric.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      module Instrumentations
+        class RedisHLLMetric < BaseMetric
+          # Usage example
+          #
+          # class CountUsersVisitingAnalyticsValuestreamMetric < RedisHLLMetric
+          #   event_names :g_analytics_valuestream
+          # end
+          class << self
+            def event_names(events = nil)
+              @mentric_events = events
+            end
+
+            attr_reader :metric_events
+          end
+
+          def value
+            redis_usage_data do
+              event_params = time_constraints.merge(event_names: self.class.metric_events)
+
+              Gitlab::UsageDataCounters::HLLRedisCounter.unique_events(**event_params)
+            end
+          end
+
+          private
+
+          def time_constraints
+            case time_frame
+            when '28d'
+              { start_date: 4.weeks.ago.to_date, end_date: Date.current }
+            when '7d'
+              { start_date: 7.days.ago.to_date, end_date: Date.current }
+            else
+              raise "Unknown time frame: #{time_frame} for TimeConstraint"
+            end
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage/metrics/instrumentations/uuid_metric.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      module Instrumentations
+        class UuidMetric < GenericMetric
+          value do
+            Gitlab::CurrentSettings.uuid
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage/metrics/key_path_processor.rb

+# frozen_string_literal: true
+
+module Gitlab
+  module Usage
+    module Metrics
+      class KeyPathProcessor
+        class << self
+          def process(key_path, value)
+            unflatten(key_path.split('.'), value)
+          end
+
+          private
+
+          def unflatten(keys, value)
+            loop do
+              value = { keys.pop.to_sym => value }
+
+              break if keys.blank?
+            end
+
+            value
+          end
+        end
+      end
+    end
+  end
+end

lib/gitlab/usage_data_metrics.rb

+# frozen_string_literal: true
+
+module Gitlab
+  class UsageDataMetrics
+    class << self
+      # Build the Usage Ping JSON payload from metrics YAML definitions which have instrumentation class set
+      def uncached_data
+        ::Gitlab::Usage::MetricDefinition.all.map do |definition|
+          instrumentation_class = definition.attributes[:instrumentation_class]
+
+          if instrumentation_class.present?
+            metric_value = instrumentation_class.constantize.new(time_frame: definition.attributes[:time_frame]).value
+
+            metric_payload(definition.key_path, metric_value)
+          else
+            {}
+          end
+        end.reduce({}, :deep_merge)
+      end
+
+      private
+
+      def metric_payload(key_path, value)
+        ::Gitlab::Usage::Metrics::KeyPathProcessor.process(key_path, value)
+      end
+    end
+  end
+end

lib/tasks/gitlab/usage_data.rake

@@ -29,5 +29,10 @@ namespace :gitlab do
       items = Gitlab::Usage::MetricDefinition.definitions
       Gitlab::Usage::Docs::Renderer.new(items).write
     end
+
+    desc 'GitLab | UsageDataMetrics | Generate usage ping from metrics definition YAML files in JSON'
+    task generate_from_yaml: :environment do
+      puts Gitlab::Json.pretty_generate(Gitlab::UsageDataMetrics.uncached_data)
+    end
   end
 end

rubocop/rubocop-code_reuse.yml

@@ -31,6 +31,7 @@ CodeReuse/ActiveRecord:
     - lib/gitlab/import_export/**/*.rb
     - lib/gitlab/project_authorizations.rb
     - lib/gitlab/sql/**/*.rb
+    - lib/gitlab/usage/metrics/instrumentations/**/*.rb
     - lib/system_check/**/*.rb
     - qa/**/*.rb
     - rubocop/**/*.rb

spec/lib/gitlab/usage/metric_definition_spec.rb

@@ -25,6 +25,12 @@ RSpec.describe Gitlab::Usage::MetricDefinition do
   let(:definition) { described_class.new(path, attributes) }
   let(:yaml_content) { attributes.deep_stringify_keys.to_yaml }
 
+  around do |example|
+    described_class.instance_variable_set(:@definitions, nil)
+    example.run
+    described_class.instance_variable_set(:@definitions, nil)
+  end
+
   def write_metric(metric, path, content)
     path = File.join(metric, path)
     dir = File.dirname(path)
@@ -62,6 +68,9 @@ RSpec.describe Gitlab::Usage::MetricDefinition do
       :distribution       | 'test'
       :tier               | %w(test ee)
       :name               | 'count_<adjective_describing>_boards'
+
+      :instrumentation_class | 'Gitlab::Usage::Metrics::Instrumentations::Metric_Class'
+      :instrumentation_class | 'Gitlab::Usage::Metrics::MetricClass'
     end
 
     with_them do

spec/lib/gitlab/usage/metrics/instrumentations/uuid_metric_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::Usage::Metrics::Instrumentations::UuidMetric do
+  it_behaves_like 'a correct instrumented metric value', { time_frame: 'none' }, Gitlab::CurrentSettings.uuid
+end

spec/lib/gitlab/usage/metrics/key_path_processor_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::Usage::Metrics::KeyPathProcessor do
+  describe '#unflatten_default_path' do
+    using RSpec::Parameterized::TableSyntax
+
+    where(:key_path, :value, :expected_hash) do
+      'uuid'                                     | nil    | { uuid: nil }
+      'uuid'                                     | '1111' | { uuid: '1111' }
+      'counts.issues'                            | nil    | { counts: { issues: nil } }
+      'counts.issues'                            | 100    | { counts: { issues: 100 } }
+      'usage_activity_by_stage.verify.ci_builds' | 100    | { usage_activity_by_stage: { verify: { ci_builds: 100 } } }
+    end
+
+    with_them do
+      subject { described_class.process(key_path, value) }
+
+      it { is_expected.to eq(expected_hash) }
+    end
+  end
+end

spec/lib/gitlab/usage_data_metrics_spec.rb

+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Gitlab::UsageDataMetrics do
+  describe '.uncached_data' do
+    subject { described_class.uncached_data }
+
+    around do |example|
+      described_class.instance_variable_set(:@definitions, nil)
+      example.run
+      described_class.instance_variable_set(:@definitions, nil)
+    end
+
+    before do
+      allow(ActiveRecord::Base.connection).to receive(:transaction_open?).and_return(false)
+    end
+
+    context 'whith instrumentation_class' do
+      it 'includes top level keys' do
+        expect(subject).to include(:uuid)
+      end
+    end
+  end
+end

spec/support/gitlab/usage/metrics_instrumentation_shared_examples.rb

+# frozen_string_literal: true
+
+RSpec.shared_examples 'a correct instrumented metric value' do |options, expected_value|
+  let(:time_frame) { options[:time_frame] }
+
+  before do
+    allow(ActiveRecord::Base.connection).to receive(:transaction_open?).and_return(false)
+  end
+
+  it 'has correct value' do
+    expect(described_class.new(time_frame: time_frame).value).to eq(expected_value)
+  end
+end