ddtrace 1.12.1 → 1.23.2

data/lib/datadog/opentelemetry/sdk/trace/span.rb

@@ -0,0 +1,167 @@
+# frozen_string_literal: true
+
+module Datadog
+  module OpenTelemetry
+    module Trace
+      # Stores associated Datadog entities to the OpenTelemetry Span.
+      module Span
+        # Attributes are equivalent to span tags and metrics.
+        def set_attribute(key, value)
+          res = super
+          # Attributes can get dropped or their values truncated by `super`
+          datadog_set_attribute(key)
+          res
+        end
+
+        # `alias` performed to match {OpenTelemetry::SDK::Trace::Span} aliasing upstream
+        alias []= set_attribute
+
+        # Attributes are equivalent to span tags and metrics.
+        def add_attributes(attributes)
+          res = super
+          # Attributes can get dropped or their values truncated by `super`
+          attributes.each { |key, _| datadog_set_attribute(key) }
+          res
+        end
+
+        # Captures changes to span error state.
+        def status=(s)
+          super
+
+          return unless status # Return if status are currently disabled by OpenTelemetry.
+          return unless (span = datadog_span)
+
+          # Status code can only change into an error state.
+          # Other change operations should be ignored.
+          span.set_error(status.description) if status && status.code == ::OpenTelemetry::Trace::Status::ERROR
+        end
+
+        # Serialize values into Datadog span tags and metrics.
+        # Notably, arrays are exploded into many keys, each with
+        # a numeric suffix representing the array index, for example:
+        # `'foo' => ['a','b']` becomes `'foo.0' => 'a', 'foo.1' => 'b'`
+        def self.serialize_attribute(key, value)
+          if value.is_a?(Array)
+            value.flat_map.with_index do |v, idx|
+              serialize_attribute("#{key}.#{idx}", v)
+            end
+          elsif value.is_a?(TrueClass) || value.is_a?(FalseClass)
+            [[key, value.to_s]]
+          else
+            [[key, value]]
+          end
+        end
+
+        # Create a meaningful Datadog operation name from the OpenTelemetry
+        # semantic convention for span kind and span attributes.
+        # @see https://opentelemetry.io/docs/specs/semconv/general/trace/
+
+        # rubocop:disable Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+        def self.enrich_name(kind, attrs)
+          if attrs.key?('http.request.method')
+            return 'http.server.request' if kind == :server
+            return 'http.client.request' if kind == :client
+          end
+
+          return "#{attrs['db.system']}.query" if attrs.key?('db.system') && kind == :client
+
+          if (attrs.key?('messaging.system') || attrs.key?('messaging.operation')) &&
+              [:consumer, :producer, :server, :client].include?(kind)
+
+            return "#{attrs['messaging.system']}.#{attrs['messaging.operation']}"
+          end
+
+          if attrs.key?('rpc.system')
+            if attrs['rpc.system'] == 'aws-api' && kind == :client
+              service = attrs['rpc.service']
+              return "aws.#{service || 'client'}.request"
+            end
+
+            if kind == :client
+              return "#{attrs['rpc.system']}.client.request"
+            elsif kind == :server
+              return "#{attrs['rpc.system']}.server.request"
+            end
+          end
+
+          if attrs.key?('faas.invoked_provider') && attrs.key?('faas.invoked_name') && kind == :client
+            provider = attrs['faas.invoked_provider']
+            name = attrs['faas.invoked_name']
+            return "#{provider}.#{name}.invoke"
+          end
+
+          return "#{attrs['faas.trigger']}.invoke" if attrs.key?('faas.trigger') && kind == :server
+
+          return 'graphql.server.request' if attrs.key?('graphql.operation.type')
+
+          if kind == :server
+            protocol = attrs['network.protocol.name']
+            return protocol ? "#{protocol}.server.request" : 'server.request'
+          end
+
+          if kind == :client
+            protocol = attrs['network.protocol.name']
+            return protocol ? "#{protocol}.client.request" : 'client.request'
+          end
+
+          kind.to_s
+        end
+        # rubocop:enable Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+
+        private
+
+        def datadog_set_attribute(key)
+          # Return if attributes are currently disabled by OpenTelemetry.
+          return unless defined?(@attributes) && @attributes
+          return unless (span = datadog_span)
+
+          # DEV: Accesses `@attributes` directly, since using `#attributes`
+          # DEV: clones the hash, causing unnecessary overhead.
+          if @attributes.key?(key)
+            # Try to find a richer operation name, unless an explicit override was provided.
+            if !@attributes.key?('operation.name') && (rich_name = Span.enrich_name(kind, @attributes))
+              span.name = rich_name.downcase
+            end
+
+            Span.serialize_attribute(key, @attributes[key]).each do |new_key, value|
+              override_datadog_values(span, new_key, value)
+
+              # When an attribute is used to override a Datadog Span property,
+              # it should NOT be set as a Datadog Span tag.
+              span.set_tag(new_key, value) unless DATADOG_SPAN_ATTRIBUTE_OVERRIDES.include?(new_key)
+            end
+          else
+            span.clear_tag(key)
+
+            if key == 'service.name'
+              # By removing the service name, we set it to the fallback default,
+              # effectively removing the `service` attribute from OpenTelemetry's perspective.
+              span.service = Datadog.send(:components).tracer.default_service
+            end
+          end
+        end
+
+        # Some special attributes can override Datadog Span fields beyond tags and metrics.
+        # @return [Boolean] true if the key is a Datadog Span override attribute, false otherwise
+        def override_datadog_values(span, key, value)
+          span.name = value if key == 'operation.name'
+          span.resource = value if key == 'resource.name'
+          span.service = value if key == 'service.name'
+          span.type = value if key == 'span.type'
+
+          if key == 'analytics.event' && value.respond_to?(:casecmp)
+            Datadog::Tracing::Analytics.set_sample_rate(
+              span,
+              value.casecmp('true') == 0 ? 1 : 0
+            )
+          end
+        end
+
+        DATADOG_SPAN_ATTRIBUTE_OVERRIDES = ['analytics.event', 'operation.name', 'resource.name', 'service.name',
+                                            'span.type'].freeze
+
+        ::OpenTelemetry::SDK::Trace::Span.prepend(self)
+      end
+    end
+  end
+end

data/lib/datadog/opentelemetry/trace.rb

@@ -0,0 +1,58 @@
+# frozen_string_literal: true
+
+require_relative '../tracing/trace_digest'
+
+module Datadog
+  module OpenTelemetry
+    # OpenTelemetry utilities related to the respective Datadog trace.
+    module Trace
+      class << self
+        # Creates a new TraceOperation object that can be attached to a new
+        # OpenTelemetry span.
+        # If `parent_span` is provided, then that span is set as the currently
+        # active parent span.
+        #
+        # @return [TraceOperation]
+        def start_trace_copy(trace, parent_span: nil)
+          digest = if parent_span
+                     digest_with_parent_span(trace, parent_span)
+                   else
+                     trace.to_digest
+                   end
+
+          # Create a new TraceOperation, attached to the current Datadog Tracer.
+          Datadog::Tracing.continue_trace!(digest)
+        end
+
+        private
+
+        # Creates a TraceDigest with the active span modified.
+        # This supports the implementation of `OpenTelemetry::Trace.context_with_span`,
+        # which allows you to specific any span as the arbitrary parent of a new span.
+        def digest_with_parent_span(trace, parent_span)
+          digest = trace.to_digest
+
+          Tracing::TraceDigest.new(
+            span_id: parent_span.id,
+            span_name: parent_span.name,
+            span_resource: parent_span.resource,
+            span_service: parent_span.service,
+            span_type: parent_span.type,
+            trace_distributed_tags: digest.trace_distributed_tags,
+            trace_hostname: digest.trace_hostname,
+            trace_id: digest.trace_id,
+            trace_name: digest.trace_name,
+            trace_origin: digest.trace_origin,
+            trace_process_id: digest.trace_process_id,
+            trace_resource: digest.trace_resource,
+            trace_runtime_id: digest.trace_runtime_id,
+            trace_sampling_priority: digest.trace_sampling_priority,
+            trace_service: digest.trace_service,
+            trace_state: digest.trace_state,
+            trace_state_unknown_fields: digest.trace_state_unknown_fields,
+          ).freeze
+        end
+      end
+    end
+  end
+end

data/lib/datadog/opentelemetry.rb

@@ -8,7 +8,10 @@
 # This file activates the integrations of all OpenTelemetry
 # components supported by Datadog.
 
+# Load Tracing
 require_relative 'tracing'
+require_relative 'tracing/contrib'
+
 require_relative 'opentelemetry/api/context'
 
 # DEV: Should this be a Contrib integration, that depends on the `opentelemetry-sdk`
@@ -16,6 +19,7 @@ require_relative 'opentelemetry/api/context'
 # DEV: This is different from our existing OpenTracer API, but there are many safety
 # DEV: features built into Contrib instrumentation today.
 require_relative 'opentelemetry/sdk/configurator' if defined?(OpenTelemetry::SDK)
+require_relative 'opentelemetry/sdk/trace/span' if defined?(OpenTelemetry::SDK)
 
 module Datadog
   # Datadog OpenTelemetry integration.

data/lib/datadog/opentracer/text_map_propagator.rb

@@ -33,7 +33,8 @@ module Datadog
           carrier[Tracing::Distributed::Datadog::ORIGIN_KEY] = digest.trace_origin
           carrier[Tracing::Distributed::Datadog::PARENT_ID_KEY] = digest.span_id
           carrier[Tracing::Distributed::Datadog::SAMPLING_PRIORITY_KEY] = digest.trace_sampling_priority
-          carrier[Tracing::Distributed::Datadog::TRACE_ID_KEY] = digest.trace_id
+          carrier[Tracing::Distributed::Datadog::TRACE_ID_KEY] =
+            Datadog::Tracing::Utils::TraceId.to_low_order(digest.trace_id)
 
           nil
         end

data/lib/datadog/opentracer.rb

@@ -22,3 +22,12 @@ require_relative 'opentracer/global_tracer'
 
 # Modify the OpenTracing module functions
 ::OpenTracing.singleton_class.prepend(Datadog::OpenTracer::GlobalTracer)
+
+module Datadog
+  # Datadog OpenTracing integration.
+  # DEV: This module should be named `Datadog::OpenTracing` to match the gem (`opentracing`).
+  module OpenTracer
+    # Used by Telemetry to decide if OpenTracing instrumentation is enabled
+    LOADED = true
+  end
+end

data/lib/datadog/profiling/collectors/code_provenance.rb

@@ -97,14 +97,20 @@ module Datadog
           end
         end
 
-        Library = Struct.new(:kind, :name, :version, :path) do
+        # Represents metadata we have for a ruby gem
+        class Library
+          attr_reader :kind, :name, :version, :path
+
           def initialize(kind:, name:, version:, path:)
-            super(kind.freeze, name.dup.freeze, version.to_s.dup.freeze, path.dup.freeze)
+            @kind = kind.freeze
+            @name = name.dup.freeze
+            @version = version.to_s.dup.freeze
+            @path = path.dup.freeze
             freeze
           end
 
-          def to_json(*args)
-            { kind: kind, name: name, version: version, paths: [path] }.to_json(*args)
+          def to_json(arg = nil)
+            { kind: @kind, name: @name, version: @version, paths: [@path] }.to_json(arg)
           end
         end
       end

data/lib/datadog/profiling/collectors/cpu_and_wall_time_worker.rb

@@ -14,23 +14,16 @@ module Datadog
         public
 
         def initialize(
-          recorder:,
-          max_frames:,
-          tracer:,
-          endpoint_collection_enabled:,
           gc_profiling_enabled:,
-          allocation_counting_enabled:,
           no_signals_workaround_enabled:,
-          thread_context_collector: ThreadContext.new(
-            recorder: recorder,
-            max_frames: max_frames,
-            tracer: tracer,
-            endpoint_collection_enabled: endpoint_collection_enabled,
-          ),
-          idle_sampling_helper: IdleSamplingHelper.new,
+          thread_context_collector:,
+          dynamic_sampling_rate_overhead_target_percentage:,
+          allocation_profiling_enabled:,
           # **NOTE**: This should only be used for testing; disabling the dynamic sampling rate will increase the
           # profiler overhead!
-          dynamic_sampling_rate_enabled: true
+          dynamic_sampling_rate_enabled: true,
+          skip_idle_samples_for_testing: false,
+          idle_sampling_helper: IdleSamplingHelper.new
         )
           unless dynamic_sampling_rate_enabled
             Datadog.logger.warn(
@@ -43,17 +36,21 @@ module Datadog
             thread_context_collector,
             gc_profiling_enabled,
             idle_sampling_helper,
-            allocation_counting_enabled,
             no_signals_workaround_enabled,
             dynamic_sampling_rate_enabled,
+            dynamic_sampling_rate_overhead_target_percentage,
+            allocation_profiling_enabled,
+            skip_idle_samples_for_testing,
           )
           @worker_thread = nil
           @failure_exception = nil
           @start_stop_mutex = Mutex.new
           @idle_sampling_helper = idle_sampling_helper
+          @wait_until_running_mutex = Mutex.new
+          @wait_until_running_condition = ConditionVariable.new
         end
 
-        def start
+        def start(on_failure_proc: nil)
           @start_stop_mutex.synchronize do
             return if @worker_thread && @worker_thread.alive?
 
@@ -74,18 +71,17 @@ module Datadog
                   'CpuAndWallTimeWorker thread error. ' \
                   "Cause: #{e.class.name} #{e.message} Location: #{Array(e.backtrace).first}"
                 )
+                on_failure_proc&.call
               end
             end
+            @worker_thread.name = self.class.name # Repeated from above to make sure thread gets named asap
+            @worker_thread.thread_variable_set(:fork_safe, true)
           end
 
           true
         end
 
-        # TODO: Provided only for compatibility with the API for Collectors::OldStack used in the Profiler class.
-        # Can be removed once we remove OldStack.
-        def enabled=(_); end
-
-        def stop(*_)
+        def stop
           @start_stop_mutex.synchronize do
             Datadog.logger.debug('Requesting CpuAndWallTimeWorker thread shut down')
 
@@ -108,6 +104,33 @@ module Datadog
         def stats
           self.class._native_stats(self)
         end
+
+        def stats_and_reset_not_thread_safe
+          stats = self.stats
+          self.class._native_stats_reset_not_thread_safe(self)
+          stats
+        end
+
+        # Useful for testing, to e.g. make sure the profiler is running before we start running some code we want to observe
+        def wait_until_running(timeout_seconds: 5)
+          @wait_until_running_mutex.synchronize do
+            return true if self.class._native_is_running?(self)
+
+            @wait_until_running_condition.wait(@wait_until_running_mutex, timeout_seconds)
+
+            if self.class._native_is_running?(self)
+              true
+            else
+              raise "Timeout waiting for #{self.class.name} to start (waited for #{timeout_seconds} seconds)"
+            end
+          end
+        end
+
+        private
+
+        def signal_running
+          @wait_until_running_mutex.synchronize { @wait_until_running_condition.broadcast }
+        end
       end
     end
   end

data/lib/datadog/profiling/collectors/idle_sampling_helper.rb

@@ -42,12 +42,14 @@ module Datadog
                 )
               end
             end
+            @worker_thread.name = self.class.name # Repeated from above to make sure thread gets named asap
+            @worker_thread.thread_variable_set(:fork_safe, true)
           end
 
           true
         end
 
-        def stop(*_unused)
+        def stop
           @start_stop_mutex.synchronize do
             Datadog.logger.debug('Requesting IdleSamplingHelper thread shut down')
 

data/lib/datadog/profiling/collectors/info.rb

@@ -0,0 +1,101 @@
+# frozen_string_literal: true
+
+require 'set'
+require 'time'
+
+module Datadog
+  module Profiling
+    module Collectors
+      # Collects information of relevance for profiler. This will get sent alongside
+      # the profile and show up in the UI or potentially influence processing in some way.
+      #
+      # Information is currently collected and frozen at construction time. A full collector
+      # could be seen as overkill for this case but it allows us to centralize information
+      # gathering and easily support more flexible/dynamic info collection in the future.
+      class Info
+        def initialize(settings)
+          @profiler_info = nil
+          @info = {
+            platform: collect_platform_info,
+            runtime: collect_runtime_info,
+            application: collect_application_info(settings),
+            profiler: collect_profiler_info(settings),
+          }.freeze
+        end
+
+        attr_reader :info
+
+        private
+
+        # Instead of trying to figure out real process start time by checking
+        # /proc or some other complex/non-portable way, approximate start time
+        # by time of requirement of this file.
+        START_TIME = Time.now.utc.freeze
+
+        def collect_platform_info
+          @platform_info ||= {
+            container_id: Datadog::Core::Environment::Container.container_id,
+            hostname: Datadog::Core::Environment::Platform.hostname,
+            kernel_name: Datadog::Core::Environment::Platform.kernel_name,
+            kernel_release: Datadog::Core::Environment::Platform.kernel_release,
+            kernel_version: Datadog::Core::Environment::Platform.kernel_version
+          }.freeze
+        end
+
+        def collect_runtime_info
+          @runtime_info ||= {
+            engine: Datadog::Core::Environment::Identity.lang_engine,
+            version: Datadog::Core::Environment::Identity.lang_version,
+            platform: Datadog::Core::Environment::Identity.lang_platform,
+          }.freeze
+        end
+
+        def collect_application_info(settings)
+          @application_info ||= {
+            start_time: START_TIME.iso8601,
+            env: settings.env,
+            service: settings.service,
+            version: settings.version,
+          }.freeze
+        end
+
+        def collect_profiler_info(settings)
+          unless @profiler_info
+            lib_datadog_gem = ::Gem.loaded_specs['libdatadog']
+            @profiler_info = {
+              version: Datadog::Core::Environment::Identity.tracer_version,
+              libdatadog: "#{lib_datadog_gem.version}-#{lib_datadog_gem.platform}",
+              settings: collect_settings_recursively(settings.profiling),
+            }.freeze
+          end
+          @profiler_info
+        end
+
+        # The settings/option model isn't directly serializable because
+        # of subsettings and options that link to full blown custom object
+        # instances without proper serialization.
+        # This method navigates a settings object recursively, converting
+        # it into more basic types that are trivially convertible to JSON.
+        def collect_settings_recursively(v)
+          v = v.options_hash if v.respond_to?(:options_hash)
+
+          if v.nil? || v.is_a?(Symbol) || v.is_a?(Numeric) || v.is_a?(String) || v.equal?(true) || v.equal?(false)
+            Core::Utils::SafeDup.frozen_or_dup(v)
+          elsif v.is_a?(Hash)
+            collected_hash = v.each_with_object({}) do |(key, value), hash|
+              collected_value = collect_settings_recursively(value)
+              hash[key] = collected_value
+            end
+            collected_hash.freeze
+          elsif v.is_a?(Enumerable)
+            collected_list = v
+              .map { |value| collect_settings_recursively(value) }
+            collected_list.freeze
+          else
+            v.inspect
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/collectors/thread_context.rb

@@ -14,9 +14,24 @@ module Datadog
       #
       # Methods prefixed with _native_ are implemented in `collectors_thread_context.c`
       class ThreadContext
-        def initialize(recorder:, max_frames:, tracer:, endpoint_collection_enabled:)
+        def initialize(
+          recorder:,
+          max_frames:,
+          tracer:,
+          endpoint_collection_enabled:,
+          timeline_enabled:,
+          allocation_type_enabled: true
+        )
           tracer_context_key = safely_extract_context_key_from(tracer)
-          self.class._native_initialize(self, recorder, max_frames, tracer_context_key, endpoint_collection_enabled)
+          self.class._native_initialize(
+            self,
+            recorder,
+            max_frames,
+            tracer_context_key,
+            endpoint_collection_enabled,
+            timeline_enabled,
+            allocation_type_enabled,
+          )
         end
 
         def inspect