ddtrace 0.49.0 → 1.12.1

data/lib/datadog/tracing/contrib/delayed_job/configuration/settings.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require_relative '../../../span_operation'
+require_relative '../../configuration/settings'
+require_relative '../ext'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        module Configuration
+          # Custom settings for the DelayedJob integration
+          # @public_api
+          class Settings < Contrib::Configuration::Settings
+            option :enabled do |o|
+              o.default { env_to_bool(Ext::ENV_ENABLED, true) }
+              o.lazy
+            end
+
+            option :analytics_enabled do |o|
+              o.default { env_to_bool(Ext::ENV_ANALYTICS_ENABLED, false) }
+              o.lazy
+            end
+
+            option :analytics_sample_rate do |o|
+              o.default { env_to_float(Ext::ENV_ANALYTICS_SAMPLE_RATE, 1.0) }
+              o.lazy
+            end
+
+            option :service_name
+            option :client_service_name
+            option :error_handler, default: Tracing::SpanOperation::Events::DEFAULT_ON_ERROR
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/delayed_job/ext.rb

@@ -0,0 +1,26 @@
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        # DelayedJob integration constants
+        # @public_api Changing resource names, tag names, or environment variables creates breaking changes.
+        module Ext
+          ENV_ENABLED = 'DD_TRACE_DELAYED_JOB_ENABLED'.freeze
+          ENV_ANALYTICS_ENABLED = 'DD_TRACE_DELAYED_JOB_ANALYTICS_ENABLED'.freeze
+          ENV_ANALYTICS_SAMPLE_RATE = 'DD_TRACE_DELAYED_JOB_ANALYTICS_SAMPLE_RATE'.freeze
+          SPAN_JOB = 'delayed_job'.freeze
+          SPAN_ENQUEUE = 'delayed_job.enqueue'.freeze
+          SPAN_RESERVE_JOB = 'delayed_job.reserve_job'.freeze
+          TAG_ATTEMPTS = 'delayed_job.attempts'.freeze
+          TAG_ID = 'delayed_job.id'.freeze
+          TAG_PRIORITY = 'delayed_job.priority'.freeze
+          TAG_QUEUE = 'delayed_job.queue'.freeze
+          TAG_COMPONENT = 'delayed_job'.freeze
+          TAG_OPERATION_ENQUEUE = 'enqueue'.freeze
+          TAG_OPERATION_JOB = 'job'.freeze
+          TAG_OPERATION_RESERVE_JOB = 'reserve_job'.freeze
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/delayed_job/integration.rb

@@ -0,0 +1,41 @@
+require_relative '../integration'
+require_relative 'configuration/settings'
+require_relative 'patcher'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        # Description of DelayedJob integration
+        class Integration
+          include Contrib::Integration
+
+          MINIMUM_VERSION = Gem::Version.new('4.1')
+
+          # @public_api Changing the integration name or integration options can cause breaking changes
+          register_as :delayed_job
+
+          def self.version
+            Gem.loaded_specs['delayed_job'] && Gem.loaded_specs['delayed_job'].version
+          end
+
+          def self.loaded?
+            !defined?(::Delayed).nil?
+          end
+
+          def self.compatible?
+            super && version >= MINIMUM_VERSION
+          end
+
+          def new_configuration
+            Configuration::Settings.new
+          end
+
+          def patcher
+            Patcher
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/delayed_job/patcher.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require_relative '../patcher'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        # Patcher enables patching of 'delayed_job' module.
+        module Patcher
+          include Contrib::Patcher
+
+          module_function
+
+          def target_version
+            Integration.version
+          end
+
+          def patch
+            require_relative 'plugin'
+            add_instrumentation(::Delayed::Worker)
+            patch_server_internals
+          end
+
+          def add_instrumentation(klass)
+            klass.plugins << Plugin
+          end
+
+          def patch_server_internals
+            require_relative 'server_internal_tracer/worker'
+            ::Delayed::Worker.prepend(ServerInternalTracer::Worker)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/delayed_job/plugin.rb

@@ -0,0 +1,108 @@
+# frozen_string_literal: true
+
+require 'delayed/plugin'
+
+require_relative '../../metadata/ext'
+require_relative '../analytics'
+require_relative 'ext'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        # DelayedJob plugin that instruments invoke_job hook
+        class Plugin < Delayed::Plugin
+          def self.instrument_invoke(job, &block)
+            return yield(job) unless Tracing.enabled?
+
+            Tracing.trace(
+              Ext::SPAN_JOB,
+              service: configuration[:service_name],
+              resource: job_name(job),
+              on_error: configuration[:error_handler]
+            ) do |span|
+              set_sample_rate(span)
+
+              # Measure service stats
+              Contrib::Analytics.set_measured(span)
+
+              span.set_tag(Ext::TAG_ID, job.id)
+              span.set_tag(Ext::TAG_QUEUE, job.queue) if job.queue
+              span.set_tag(Ext::TAG_PRIORITY, job.priority)
+              span.set_tag(Ext::TAG_ATTEMPTS, job.attempts)
+              span.span_type = Tracing::Metadata::Ext::AppTypes::TYPE_WORKER
+
+              span.set_tag(Tracing::Metadata::Ext::TAG_COMPONENT, Ext::TAG_COMPONENT)
+              span.set_tag(Tracing::Metadata::Ext::TAG_OPERATION, Ext::TAG_OPERATION_JOB)
+
+              span.set_tag(Tracing::Metadata::Ext::TAG_KIND, Tracing::Metadata::Ext::SpanKind::TAG_CONSUMER)
+
+              span.set_tag(Contrib::Ext::Messaging::TAG_SYSTEM, Ext::TAG_COMPONENT)
+
+              yield job
+            end
+          end
+
+          def self.instrument_enqueue(job, &block)
+            return yield(job) unless Tracing.enabled?
+
+            Tracing.trace(
+              Ext::SPAN_ENQUEUE,
+              service: configuration[:client_service_name],
+              resource: job_name(job)
+            ) do |span|
+              set_sample_rate(span)
+
+              # Measure service stats
+              Contrib::Analytics.set_measured(span)
+
+              span.set_tag(Ext::TAG_QUEUE, job.queue) if job.queue
+              span.set_tag(Ext::TAG_PRIORITY, job.priority)
+              span.span_type = Tracing::Metadata::Ext::AppTypes::TYPE_WORKER
+
+              span.set_tag(Tracing::Metadata::Ext::TAG_COMPONENT, Ext::TAG_COMPONENT)
+              span.set_tag(Tracing::Metadata::Ext::TAG_OPERATION, Ext::TAG_OPERATION_ENQUEUE)
+
+              span.set_tag(Tracing::Metadata::Ext::TAG_KIND, Tracing::Metadata::Ext::SpanKind::TAG_PRODUCER)
+
+              span.set_tag(Contrib::Ext::Messaging::TAG_SYSTEM, Ext::TAG_COMPONENT)
+
+              yield job
+            end
+          end
+
+          def self.flush(worker, &block)
+            yield worker
+
+            Tracing.shutdown! if Tracing.enabled?
+          end
+
+          def self.configuration
+            Datadog.configuration.tracing[:delayed_job]
+          end
+
+          def self.job_name(job)
+            # When DelayedJob is used through ActiveJob, we need to parse the payload differentely
+            # to get the actual job name
+            return job.payload_object.job_data['job_class'] if job.payload_object.respond_to?(:job_data)
+
+            job.name
+          end
+
+          def self.set_sample_rate(span)
+            # Set analytics sample rate
+            if Contrib::Analytics.enabled?(configuration[:analytics_enabled])
+              Contrib::Analytics.set_sample_rate(span, configuration[:analytics_sample_rate])
+            end
+          end
+
+          callbacks do |lifecycle|
+            lifecycle.around(:invoke_job) { |job, &block| instrument_invoke(job, &block) }
+            lifecycle.around(:enqueue) { |job, &block| instrument_enqueue(job, &block) }
+            lifecycle.around(:execute) { |worker, &block| flush(worker, &block) }
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/delayed_job/server_internal_tracer/worker.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Tracing
+    module Contrib
+      module DelayedJob
+        module ServerInternalTracer
+          # Trace when DelayedJob looks for a new job to work on
+          module Worker
+            def reserve_job
+              configuration = Datadog.configuration.tracing[:delayed_job]
+
+              Datadog::Tracing.trace(Ext::SPAN_RESERVE_JOB, service: configuration[:service_name]) do |span|
+                span.span_type = Datadog::Tracing::Metadata::Ext::AppTypes::TYPE_WORKER
+
+                span.set_tag(Contrib::Ext::Messaging::TAG_SYSTEM, Ext::TAG_COMPONENT)
+
+                span.set_tag(Datadog::Tracing::Metadata::Ext::TAG_COMPONENT, Ext::TAG_COMPONENT)
+                span.set_tag(Datadog::Tracing::Metadata::Ext::TAG_OPERATION, Ext::TAG_OPERATION_RESERVE_JOB)
+
+                # Set analytics sample rate
+                if Contrib::Analytics.enabled?(configuration[:analytics_enabled])
+                  Contrib::Analytics.set_sample_rate(span, configuration[:analytics_sample_rate])
+                end
+
+                super
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/elasticsearch/configuration/settings.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+require_relative '../../configuration/settings'
+require_relative '../ext'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Elasticsearch
+        module Configuration
+          # Custom settings for the Elasticsearch integration
+          # @public_api
+          class Settings < Contrib::Configuration::Settings
+            option :enabled do |o|
+              o.default { env_to_bool(Ext::ENV_ENABLED, true) }
+              o.lazy
+            end
+
+            option :analytics_enabled do |o|
+              o.default { env_to_bool(Ext::ENV_ANALYTICS_ENABLED, false) }
+              o.lazy
+            end
+
+            option :analytics_sample_rate do |o|
+              o.default { env_to_float(Ext::ENV_ANALYTICS_SAMPLE_RATE, 1.0) }
+              o.lazy
+            end
+
+            option :quantize, default: {}
+
+            option :service_name do |o|
+              o.default do
+                Contrib::SpanAttributeSchema.fetch_service_name(
+                  Ext::ENV_SERVICE_NAME,
+                  Ext::DEFAULT_PEER_SERVICE_NAME
+                )
+              end
+              o.lazy
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/elasticsearch/ext.rb

@@ -0,0 +1,27 @@
+module Datadog
+  module Tracing
+    module Contrib
+      module Elasticsearch
+        # Elasticsearch integration constants
+        # @public_api Changing resource names, tag names, or environment variables creates breaking changes.
+        module Ext
+          ENV_ENABLED = 'DD_TRACE_ELASTICSEARCH_ENABLED'.freeze
+          ENV_SERVICE_NAME = 'DD_TRACE_ELASTICSEARCH_SERVICE_NAME'.freeze
+          ENV_ANALYTICS_ENABLED = 'DD_TRACE_ELASTICSEARCH_ANALYTICS_ENABLED'.freeze
+          ENV_ANALYTICS_SAMPLE_RATE = 'DD_TRACE_ELASTICSEARCH_ANALYTICS_SAMPLE_RATE'.freeze
+          DEFAULT_PEER_SERVICE_NAME = 'elasticsearch'.freeze
+          SPAN_QUERY = 'elasticsearch.query'.freeze
+          SPAN_TYPE_QUERY = 'elasticsearch'.freeze
+          TAG_BODY = 'elasticsearch.body'.freeze
+          TAG_METHOD = 'elasticsearch.method'.freeze
+          TAG_PARAMS = 'elasticsearch.params'.freeze
+          TAG_URL = 'elasticsearch.url'.freeze
+          TAG_COMPONENT = 'elasticsearch'.freeze
+          TAG_OPERATION_QUERY = 'query'.freeze
+
+          TAG_SYSTEM = 'elasticsearch'.freeze
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/elasticsearch/integration.rb

@@ -0,0 +1,48 @@
+require_relative '../integration'
+require_relative 'configuration/settings'
+require_relative 'patcher'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Elasticsearch
+        # Description of Elasticsearch integration
+        class Integration
+          include Contrib::Integration
+
+          MINIMUM_VERSION = Gem::Version.new('1.0.0')
+
+          # @public_api Changing the integration name or integration options can cause breaking changes
+          register_as :elasticsearch, auto_patch: true
+
+          def self.version
+            # elastic-transport gem for version >= 8.0.0
+            # elasticsearch-transport gem for version < 8.0.0
+            Gem.loaded_specs['elastic-transport'] \
+              && Gem.loaded_specs['elastic-transport'].version || \
+              Gem.loaded_specs['elasticsearch-transport'] \
+                && Gem.loaded_specs['elasticsearch-transport'].version
+          end
+
+          def self.loaded?
+            # Elastic::Transport gem for version >= 8.0.0
+            # Elasticsearch::Transport gem for version < 8.0.0
+            !defined?(::Elastic::Transport).nil? || !defined?(::Elasticsearch::Transport).nil?
+          end
+
+          def self.compatible?
+            super && version >= MINIMUM_VERSION
+          end
+
+          def new_configuration
+            Configuration::Settings.new
+          end
+
+          def patcher
+            Patcher
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/elasticsearch/patcher.rb

@@ -0,0 +1,155 @@
+require_relative '../../metadata/ext'
+require_relative '../analytics'
+require_relative 'ext'
+require_relative '../ext'
+require_relative '../integration'
+require_relative '../patcher'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Elasticsearch
+        # Patcher enables patching of 'elasticsearch' module.
+        module Patcher
+          include Contrib::Patcher
+
+          module_function
+
+          def target_version
+            Integration.version
+          end
+
+          def patch
+            require 'uri'
+            require 'json'
+            require_relative 'quantize'
+
+            patch_elasticsearch_transport_client
+          end
+
+          SELF_DEPRECATION_ONLY_ONCE = Core::Utils::OnlyOnce.new
+
+          # rubocop:disable Metrics/MethodLength
+          # rubocop:disable Metrics/AbcSize
+          # rubocop:disable Metrics/CyclomaticComplexity
+          # rubocop:disable Metrics/PerceivedComplexity
+          def patch_elasticsearch_transport_client
+            # rubocop:disable Metrics/BlockLength
+            transport_module::Client.class_eval do
+              alias_method :perform_request_without_datadog, :perform_request
+              remove_method :perform_request
+
+              def perform_request(*args)
+                # DEV-2.0: Remove this access, as `Client#self` in this context is not exposed to the user
+                # since `elasticsearch` v8.0.0. In contrast, `Client#transport` is always available across
+                # all `elasticsearch` gem versions and should be used instead.
+                service = Datadog.configuration_for(self, :service_name)
+
+                if service
+                  SELF_DEPRECATION_ONLY_ONCE.run do
+                    Datadog.logger.warn(
+                      'Providing configuration though the Elasticsearch client object is deprecated.' \
+                      'Configure the `client#transport` object instead: ' \
+                      'Datadog.configure_onto(client.transport, service_name: service_name, ...)'
+                    )
+                  end
+                end
+
+                # `Client#transport` is most convenient object both this integration and the library
+                # user have shared access to across all `elasticsearch` versions.
+                #
+                # `Client#self` in this context is an internal object that the library user
+                # does not have access to since `elasticsearch` v8.0.0.
+                service ||= Datadog.configuration_for(transport, :service_name) || datadog_configuration[:service_name]
+
+                method = args[0]
+                path = args[1]
+                params = args[2]
+                body = args[3]
+                full_url = URI.parse(path)
+
+                url = full_url.path
+                response = nil
+
+                Tracing.trace(Datadog::Tracing::Contrib::Elasticsearch::Ext::SPAN_QUERY, service: service) do |span|
+                  begin
+                    connection = transport.connections.first
+                    host = connection.host[:host] if connection
+                    port = connection.host[:port] if connection
+
+                    span.span_type = Datadog::Tracing::Contrib::Elasticsearch::Ext::SPAN_TYPE_QUERY
+
+                    span.set_tag(Tracing::Metadata::Ext::TAG_COMPONENT, Ext::TAG_COMPONENT)
+                    span.set_tag(Tracing::Metadata::Ext::TAG_OPERATION, Ext::TAG_OPERATION_QUERY)
+                    span.set_tag(Tracing::Metadata::Ext::TAG_KIND, Tracing::Metadata::Ext::SpanKind::TAG_CLIENT)
+
+                    span.set_tag(Contrib::Ext::DB::TAG_SYSTEM, Ext::TAG_SYSTEM)
+
+                    # load JSON for the following fields unless they're already strings
+                    params = JSON.generate(params) if params && !params.is_a?(String)
+                    body = JSON.generate(body) if body && !body.is_a?(String)
+
+                    # Tag as an external peer service
+                    if Contrib::SpanAttributeSchema.default_span_attribute_schema?
+                      span.set_tag(Tracing::Metadata::Ext::TAG_PEER_SERVICE, span.service)
+                    end
+
+                    span.set_tag(Tracing::Metadata::Ext::TAG_PEER_HOSTNAME, host) if host
+
+                    # Set analytics sample rate
+                    if Contrib::Analytics.enabled?(datadog_configuration[:analytics_enabled])
+                      Contrib::Analytics.set_sample_rate(span, datadog_configuration[:analytics_sample_rate])
+                    end
+
+                    span.set_tag(Datadog::Tracing::Contrib::Elasticsearch::Ext::TAG_METHOD, method)
+                    span.set_tag(Datadog::Tracing::Contrib::Elasticsearch::Ext::TAG_URL, url)
+                    span.set_tag(Datadog::Tracing::Contrib::Elasticsearch::Ext::TAG_PARAMS, params) if params
+                    if body
+                      quantize_options = datadog_configuration[:quantize]
+                      quantized_body = Datadog::Tracing::Contrib::Elasticsearch::Quantize.format_body(
+                        body,
+                        quantize_options
+                      )
+                      span.set_tag(Datadog::Tracing::Contrib::Elasticsearch::Ext::TAG_BODY, quantized_body)
+                    end
+                    span.set_tag(Tracing::Metadata::Ext::NET::TAG_TARGET_HOST, host) if host
+                    span.set_tag(Tracing::Metadata::Ext::NET::TAG_TARGET_PORT, port) if port
+
+                    quantized_url = Datadog::Tracing::Contrib::Elasticsearch::Quantize.format_url(url)
+                    span.resource = "#{method} #{quantized_url}"
+                  rescue StandardError => e
+                    Datadog.logger.error(e.message)
+                  ensure
+                    # the call is still executed
+                    response = perform_request_without_datadog(*args)
+                    span.set_tag(Tracing::Metadata::Ext::HTTP::TAG_STATUS_CODE, response.status)
+                  end
+                end
+                response
+              end
+
+              def datadog_configuration
+                Datadog.configuration.tracing[:elasticsearch]
+              end
+            end
+            # rubocop:enable Metrics/BlockLength
+          end
+          # rubocop:enable Metrics/MethodLength
+          # rubocop:enable Metrics/AbcSize
+          # rubocop:enable Metrics/CyclomaticComplexity
+          # rubocop:enable Metrics/PerceivedComplexity
+
+          # `Elasticsearch` namespace renamed to `Elastic` in version 8.0.0 of the transport gem:
+          # @see https://github.com/elastic/elastic-transport-ruby/commit/ef804cbbd284f2a82d825221f87124f8b5ff823c
+          def transport_module
+            if Integration.version >= Gem::Version.new('8.0.0')
+              ::Elastic::Transport
+            else
+              ::Elasticsearch::Transport
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/contrib/elasticsearch/quantize.rb

@@ -0,0 +1,85 @@
+require_relative '../utils/quantization/hash'
+
+module Datadog
+  module Tracing
+    module Contrib
+      module Elasticsearch
+        # Quantize contains ES-specific resource quantization tools.
+        module Quantize
+          PLACEHOLDER = '?'.freeze
+          ID_PLACEHOLDER = '\1?'.freeze
+          EXCLUDE_KEYS = [].freeze
+          SHOW_KEYS = [:_index, :_type, :_id].freeze
+          DEFAULT_OPTIONS = {
+            exclude: EXCLUDE_KEYS,
+            show: SHOW_KEYS,
+            placeholder: PLACEHOLDER
+          }.freeze
+
+          module_function
+
+          def format_url(url)
+            sanitize_fragment_with_id(url)
+              .gsub(/(?:\d+)/, PLACEHOLDER)
+          end
+
+          def format_body(body, options = {})
+            format_body!(body, options)
+          rescue StandardError
+            options[:placeholder] || PLACEHOLDER
+          end
+
+          def format_body!(body, options = {})
+            options = merge_options(DEFAULT_OPTIONS, options)
+
+            # Determine if bulk query or not, based on content
+            statements = body.end_with?("\n") ? body.split("\n") : [body]
+
+            # Parse each statement and quantize them.
+            statements.collect do |string|
+              reserialize_json(string, options[:placeholder]) do |obj|
+                Contrib::Utils::Quantization::Hash.format(obj, options)
+              end
+            end.join("\n")
+          end
+
+          def merge_options(original, additional)
+            {}.tap do |options|
+              # Show
+              # If either is :all, value becomes :all
+              options[:show] = if original[:show] == :all || additional[:show] == :all
+                                 :all
+                               else
+                                 (original[:show] || []).dup.concat(additional[:show] || []).uniq
+                               end
+
+              # Exclude
+              options[:exclude] = (original[:exclude] || []).dup.concat(additional[:exclude] || []).uniq
+            end
+          end
+
+          # Parses a JSON object from a string, passes its value
+          # to the block provided, and dumps its result back to JSON.
+          # If JSON parsing fails, it prints fail_value.
+          def reserialize_json(string, fail_value = PLACEHOLDER)
+            return string unless block_given?
+
+            begin
+              JSON.dump(yield(JSON.parse(string)))
+            rescue JSON::ParserError
+              # If it can't parse/dump, don't raise an error.
+              fail_value
+            end
+          end
+
+          # Sanitizes URL fragment by changing it to ? whenever a number is detected
+          # This is meant as simple heuristic that attempts to detect if particular fragment
+          # represents document Id. This is meant to reduce the cardinality in most frequent cases.
+          def sanitize_fragment_with_id(url)
+            url.gsub(%r{^(/?[^/]*/[^/]*/)(?:[^?/\d]*\d+[^?/]*)}, ID_PLACEHOLDER)
+          end
+        end
+      end
+    end
+  end
+end