ddtrace 1.1.0 → 1.12.1

data/lib/datadog/tracing/trace_digest.rb

@@ -1,4 +1,4 @@
-# typed: true
+# frozen_string_literal: true
 
 module Datadog
   module Tracing
@@ -6,12 +6,84 @@ module Datadog
     # Used to propagate context and continue traces across execution boundaries.
     # @public_api
     class TraceDigest
+      # @!attribute [r] span_id
+      #   Datadog id for the currently active span.
+      #   @return [Integer]
+      # @!attribute [r] span_name
+      #   The operation name of the currently active span.
+      #   @return [String]
+      # @!attribute [r] span_resource
+      #   The resource name of the currently active span.
+      #   @return [String]
+      # @!attribute [r] span_service
+      #   The service of the currently active span.
+      #   @return [String]
+      # @!attribute [r] span_type
+      #   The type of the currently active span.
+      #   @return [String]
+      # @!attribute [r] trace_distributed_tags
+      #   Datadog-specific tags that support richer distributed tracing association.
+      #   @return [Hash<String,String>]
+      # @!attribute [r] trace_hostname
+      #   The hostname of the currently active trace. Use to attribute traces to hosts.
+      #   @return [String]
+      # @!attribute [r] trace_id
+      #   Datadog id for the currently active trace.
+      #   @return [Integer]
+      # @!attribute [r] trace_name
+      #   Operation name for the currently active trace.
+      #   @return [Integer]
+      # @!attribute [r] trace_origin
+      #   Datadog-specific attribution of this trace's creation.
+      #   @return [String]
+      # @!attribute [r] trace_process_id
+      #   The OS-specific process id.
+      #   @return [Integer]
+      # @!attribute [r] trace_resource
+      #   The resource name of the currently active trace.
+      #   @return [String]
+      # @!attribute [r] trace_runtime_id
+      #   Unique id to this Ruby process. Used to differentiate traces coming from
+      #   child processes forked from same parent process.
+      #   @return [String]
+      # @!attribute [r] trace_sampling_priority
+      #   Datadog-specific sampling decision for the currently active trace.
+      #   @return [Integer]
+      # @!attribute [r] trace_service
+      #   The service of the currently active trace.
+      #   @return [String]
+      # @!attribute [r] trace_distributed_id
+      #   The trace id extracted from a distributed context, if different from `trace_id`.
+      #
+      #   The current use case is when the distributed context has a trace id integer larger than 64-bit:
+      #   This attribute will preserve the original id, while `trace_id` will only contain the lower 64 bits.
+      #   @return [Integer]
+      #   @see https://www.w3.org/TR/trace-context/#trace-id
+      # @!attribute [r] trace_flags
+      #   The W3C "trace-flags" extracted from a distributed context. This field is an 8-bit unsigned integer.
+      #   @return [Integer]
+      #   @see https://www.w3.org/TR/trace-context/#trace-flags
+      # @!attribute [r] trace_state
+      #   The W3C "tracestate" extracted from a distributed context.
+      #   This field is a string representing vendor-specific distribution data.
+      #   The `dd=` entry is removed from `trace_state` as its value is dynamically calculated
+      #   on every propagation injection.
+      #   @return [String]
+      #   @see https://www.w3.org/TR/trace-context/#tracestate-header
+      # @!attribute [r] trace_state_unknown_fields
+      #   From W3C "tracestate"'s `dd=` entry, when keys are not recognized they are stored here long with their values.
+      #   This allows later propagation to include those unknown fields, as they can represent future versions of the spec
+      #   sending data through this service. This value ends in a trailing `;` to facilitate serialization.
+      #   @return [String]
+      # TODO: The documentation for the last attribute above won't be rendered.
+      # TODO: This might be a YARD bug as adding an attribute, making it now second-last attribute, renders correctly.
       attr_reader \
         :span_id,
         :span_name,
         :span_resource,
         :span_service,
         :span_type,
+        :trace_distributed_tags,
         :trace_hostname,
         :trace_id,
         :trace_name,
@@ -20,7 +92,11 @@ module Datadog
         :trace_resource,
         :trace_runtime_id,
         :trace_sampling_priority,
-        :trace_service
+        :trace_service,
+        :trace_distributed_id,
+        :trace_flags,
+        :trace_state,
+        :trace_state_unknown_fields
 
       def initialize(
         span_id: nil,
@@ -28,6 +104,7 @@ module Datadog
         span_resource: nil,
         span_service: nil,
         span_type: nil,
+        trace_distributed_tags: nil,
         trace_hostname: nil,
         trace_id: nil,
         trace_name: nil,
@@ -36,13 +113,18 @@ module Datadog
         trace_resource: nil,
         trace_runtime_id: nil,
         trace_sampling_priority: nil,
-        trace_service: nil
+        trace_service: nil,
+        trace_distributed_id: nil,
+        trace_flags: nil,
+        trace_state: nil,
+        trace_state_unknown_fields: nil
       )
         @span_id = span_id
         @span_name = span_name && span_name.dup.freeze
         @span_resource = span_resource && span_resource.dup.freeze
         @span_service = span_service && span_service.dup.freeze
         @span_type = span_type && span_type.dup.freeze
+        @trace_distributed_tags = trace_distributed_tags && trace_distributed_tags.dup.freeze
         @trace_hostname = trace_hostname && trace_hostname.dup.freeze
         @trace_id = trace_id
         @trace_name = trace_name && trace_name.dup.freeze
@@ -52,6 +134,10 @@ module Datadog
         @trace_runtime_id = trace_runtime_id && trace_runtime_id.dup.freeze
         @trace_sampling_priority = trace_sampling_priority
         @trace_service = trace_service && trace_service.dup.freeze
+        @trace_distributed_id = trace_distributed_id
+        @trace_flags = trace_flags
+        @trace_state = trace_state && trace_state.dup.freeze
+        @trace_state_unknown_fields = trace_state_unknown_fields && trace_state_unknown_fields.dup.freeze
 
         freeze
       end

data/lib/datadog/tracing/trace_operation.rb

@@ -1,15 +1,13 @@
-# typed: false
+require_relative '../core/environment/identity'
+require_relative '../core/utils'
 
-require 'datadog/core'
-require 'datadog/core/environment/identity'
-require 'datadog/core/utils'
-
-require 'datadog/tracing/sampling/ext'
-require 'datadog/tracing/event'
-require 'datadog/tracing/span_operation'
-require 'datadog/tracing/trace_segment'
-require 'datadog/tracing/trace_digest'
-require 'datadog/tracing/metadata/tagging'
+require_relative 'event'
+require_relative 'metadata/tagging'
+require_relative 'sampling/ext'
+require_relative 'span_operation'
+require_relative 'trace_digest'
+require_relative 'trace_segment'
+require_relative 'utils'
 
 module Datadog
   module Tracing
@@ -22,7 +20,6 @@ module Datadog
     # For async support, a {Datadog::Tracing::TraceOperation} should be employed
     # per execution context (e.g. Thread, etc.)
     #
-    # rubocop:disable Metrics/ClassLength
     # @public_api
     class TraceOperation
       include Metadata::Tagging
@@ -71,10 +68,10 @@ module Datadog
         metrics: nil
       )
         # Attributes
-        @id = id || Core::Utils.next_id
+        @id = id || Tracing::Utils::TraceId.next_id
         @max_length = max_length || DEFAULT_MAX_LENGTH
         @parent_span_id = parent_span_id
-        @sampled = sampled.nil? ? false : sampled
+        @sampled = sampled.nil? ? true : sampled
 
         # Tags
         @agent_sample_rate = agent_sample_rate
@@ -113,18 +110,34 @@ module Datadog
         @finished == true
       end
 
+      # Will this trace be flushed by the tracer transport?
+      # This includes cases where the span is kept solely due to priority sampling.
+      #
+      # This is not the ultimate Datadog App sampling decision. Downstream systems
+      # can decide to reject this trace, especially for cases where priority
+      # sampling is set to AUTO_KEEP.
+      #
+      # @return [Boolean]
       def sampled?
-        @sampled == true || (!@sampling_priority.nil? && @sampling_priority > 0)
+        @sampled == true || priority_sampled?
+      end
+
+      # Has the priority sampling chosen to keep this span?
+      # @return [Boolean]
+      def priority_sampled?
+        !@sampling_priority.nil? && @sampling_priority > 0
       end
 
       def keep!
         self.sampled = true
         self.sampling_priority = Sampling::Ext::Priority::USER_KEEP
+        set_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER, Tracing::Sampling::Ext::Decision::MANUAL)
       end
 
       def reject!
         self.sampled = false
         self.sampling_priority = Sampling::Ext::Priority::USER_REJECT
+        set_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER, Tracing::Sampling::Ext::Decision::MANUAL)
       end
 
       def name
@@ -135,6 +148,13 @@ module Datadog
         @resource || (root_span && root_span.resource)
       end
 
+      # Returns true if the resource has been explicitly set
+      #
+      # @return [Boolean]
+      def resource_override?
+        !@resource.nil?
+      end
+
       def service
         @service || (root_span && root_span.service)
       end
@@ -226,6 +246,11 @@ module Datadog
         end
       end
 
+      # Returns a {TraceSegment} with all finished spans that can be flushed
+      # at invocation time. All other **finished** spans are discarded.
+      #
+      # @yield [spans] spans that will be returned as part of the trace segment returned
+      # @return [TraceSegment]
       def flush!
         finished = finished?
 
@@ -233,6 +258,8 @@ module Datadog
         spans = @spans.dup
         @spans = []
 
+        spans = yield(spans) if block_given?
+
         # Use them to build a trace
         build_trace(spans, !finished)
       end
@@ -251,6 +278,7 @@ module Datadog
           span_resource: (@active_span && @active_span.resource),
           span_service: (@active_span && @active_span.service),
           span_type: (@active_span && @active_span.type),
+          trace_distributed_tags: distributed_tags,
           trace_hostname: @hostname,
           trace_id: @id,
           trace_name: name,
@@ -425,7 +453,21 @@ module Datadog
           root_span_id: !partial ? root_span && root_span.id : nil
         )
       end
+
+      # Returns tracer tags that will be propagated if this span's context
+      # is exported through {.to_digest}.
+      # @return [Hash] key value pairs of distributed tags
+      def distributed_tags
+        meta.select { |name, _| name.start_with?(Metadata::Ext::Distributed::TAGS_PREFIX) }
+      end
+
+      def reset
+        @root_span = nil
+        @active_span = nil
+        @active_span_count = 0
+        @finished = false
+        @spans = []
+      end
     end
-    # rubocop:enable Metrics/ClassLength
   end
 end

data/lib/datadog/tracing/trace_segment.rb

@@ -1,17 +1,15 @@
-# typed: true
+require_relative '../core/runtime/ext'
+require_relative '../core/utils/safe_dup'
 
-require 'datadog/core/runtime/ext'
-require 'datadog/core/utils/safe_dup'
-
-require 'datadog/tracing/sampling/ext'
-require 'datadog/tracing/metadata/ext'
-require 'datadog/tracing/metadata/tagging'
+require_relative 'sampling/ext'
+require_relative 'metadata/ext'
+require_relative 'metadata/tagging'
+require_relative 'utils'
 
 module Datadog
   module Tracing
     # Serializable construct representing a trace
     # @public_api
-    # rubocop:disable Metrics/ClassLength
     class TraceSegment
       TAG_NAME = 'name'.freeze
       TAG_RESOURCE = 'resource'.freeze
@@ -31,11 +29,13 @@ module Datadog
         :rule_sample_rate,
         :runtime_id,
         :sample_rate,
+        :sampling_decision_maker,
         :sampling_priority,
         :service
 
       # rubocop:disable Metrics/CyclomaticComplexity
       # rubocop:disable Metrics/PerceivedComplexity
+      # @param spans [Array<Datadog::Span>]
       def initialize(
         spans,
         agent_sample_rate: nil,
@@ -77,6 +77,7 @@ module Datadog
         @rule_sample_rate = rule_sample_rate_tag || rule_sample_rate
         @runtime_id = runtime_id || runtime_id_tag
         @sample_rate = sample_rate || sample_rate_tag
+        @sampling_decision_maker = sampling_decision_maker_tag
         @sampling_priority = sampling_priority || sampling_priority_tag
         @service = Core::Utils::SafeDup.frozen_or_dup(service || service_tag)
       end
@@ -126,6 +127,12 @@ module Datadog
           || sampling_priority == Sampling::Ext::Priority::USER_KEEP
       end
 
+      def high_order_tid
+        high_order = Tracing::Utils::TraceId.to_high_order(@id)
+
+        high_order.to_s(16) if high_order != 0
+      end
+
       protected
 
       attr_reader \
@@ -171,7 +178,7 @@ module Datadog
       end
 
       def process_id_tag
-        meta[Core::Runtime::Ext::TAG_PID]
+        meta[Core::Runtime::Ext::TAG_PROCESS_ID]
       end
 
       def rate_limiter_rate_tag
@@ -194,6 +201,10 @@ module Datadog
         metrics[Metadata::Ext::Sampling::TAG_SAMPLE_RATE]
       end
 
+      def sampling_decision_maker_tag
+        meta[Metadata::Ext::Distributed::TAG_DECISION_MAKER]
+      end
+
       def sampling_priority_tag
         meta[Metadata::Ext::Distributed::TAG_SAMPLING_PRIORITY]
       end
@@ -202,6 +213,5 @@ module Datadog
         meta[TAG_SERVICE]
       end
     end
-    # rubocop:enable Metrics/ClassLength
   end
 end

data/lib/datadog/tracing/tracer.rb

@@ -1,20 +1,18 @@
-# typed: true
-
-require 'datadog/core'
-require 'datadog/core/environment/ext'
-require 'datadog/core/environment/socket'
-
-require 'datadog/tracing/correlation'
-require 'datadog/tracing/event'
-require 'datadog/tracing/flush'
-require 'datadog/tracing/context_provider'
-require 'datadog/tracing/sampling/all_sampler'
-require 'datadog/tracing/sampling/rule_sampler'
-require 'datadog/tracing/sampling/priority_sampler'
-require 'datadog/tracing/span_operation'
-require 'datadog/tracing/trace_digest'
-require 'datadog/tracing/trace_operation'
-require 'datadog/tracing/writer'
+require_relative '../core/environment/ext'
+require_relative '../core/environment/socket'
+
+require_relative 'correlation'
+require_relative 'event'
+require_relative 'flush'
+require_relative 'context_provider'
+require_relative 'sampling/all_sampler'
+require_relative 'sampling/rule_sampler'
+require_relative 'sampling/priority_sampler'
+require_relative 'sampling/span/sampler'
+require_relative 'span_operation'
+require_relative 'trace_digest'
+require_relative 'trace_operation'
+require_relative 'writer'
 
 module Datadog
   module Tracing
@@ -22,12 +20,12 @@ module Datadog
     # example, a trace can be used to track the entire time spent processing a complicated web request.
     # Even though the request may require multiple resources and machines to handle the request, all
     # of these function calls and sub-requests would be encapsulated within a single trace.
-    # rubocop:disable Metrics/ClassLength
     class Tracer
       attr_reader \
         :trace_flush,
         :provider,
         :sampler,
+        :span_sampler,
         :tags
 
       attr_accessor \
@@ -56,6 +54,7 @@ module Datadog
           base_sampler: Sampling::AllSampler.new,
           post_sampler: Sampling::RuleSampler.new
         ),
+        span_sampler: Sampling::Span::Sampler.new,
         tags: {},
         writer: Writer.new
       )
@@ -64,6 +63,7 @@ module Datadog
         @enabled = enabled
         @provider = context_provider
         @sampler = sampler
+        @span_sampler = span_sampler
         @tags = tags
         @writer = writer
       end
@@ -323,7 +323,9 @@ module Datadog
             id: digest.trace_id,
             origin: digest.trace_origin,
             parent_span_id: digest.span_id,
-            sampling_priority: digest.trace_sampling_priority
+            sampling_priority: digest.trace_sampling_priority,
+            # Distributed tags are just regular trace tags with special meaning to Datadog
+            tags: digest.trace_distributed_tags,
           )
         else
           TraceOperation.new(
@@ -341,7 +343,8 @@ module Datadog
           sample_trace(event_trace_op) if event_span_op && event_span_op.parent_id == 0
         end
 
-        events.span_finished.subscribe do |_event_span, event_trace_op|
+        events.span_finished.subscribe do |event_span, event_trace_op|
+          sample_span(event_trace_op, event_span)
           flush_trace(event_trace_op)
         end
       end
@@ -461,20 +464,43 @@ module Datadog
         begin
           @sampler.sample!(trace_op)
         rescue StandardError => e
-          Datadog.logger.debug { "Failed to sample trace: #{e}" }
+          SAMPLE_TRACE_LOG_ONLY_ONCE.run do
+            Datadog.logger.warn { "Failed to sample trace: #{e.class.name} #{e} at #{Array(e.backtrace).first}" }
+          end
         end
       end
 
+      SAMPLE_TRACE_LOG_ONLY_ONCE = Core::Utils::OnlyOnce.new
+      private_constant :SAMPLE_TRACE_LOG_ONLY_ONCE
+
+      def sample_span(trace_op, span)
+        begin
+          @span_sampler.sample!(trace_op, span)
+        rescue StandardError => e
+          SAMPLE_SPAN_LOG_ONLY_ONCE.run do
+            Datadog.logger.warn { "Failed to sample span: #{e.class.name} #{e} at #{Array(e.backtrace).first}" }
+          end
+        end
+      end
+
+      SAMPLE_SPAN_LOG_ONLY_ONCE = Core::Utils::OnlyOnce.new
+      private_constant :SAMPLE_SPAN_LOG_ONLY_ONCE
+
       # Flush finished spans from the trace buffer, send them to writer.
       def flush_trace(trace_op)
         begin
           trace = @trace_flush.consume!(trace_op)
           write(trace) if trace && !trace.empty?
         rescue StandardError => e
-          Datadog.logger.debug { "Failed to flush trace: #{e}" }
+          FLUSH_TRACE_LOG_ONLY_ONCE.run do
+            Datadog.logger.warn { "Failed to flush trace: #{e.class.name} #{e} at #{Array(e.backtrace).first}" }
+          end
         end
       end
 
+      FLUSH_TRACE_LOG_ONLY_ONCE = Core::Utils::OnlyOnce.new
+      private_constant :FLUSH_TRACE_LOG_ONLY_ONCE
+
       # Send the trace to the writer to enqueue the spans list in the agent
       # sending queue.
       def write(trace)
@@ -500,6 +526,5 @@ module Datadog
         end
       end
     end
-    # rubocop:enable Metrics/ClassLength
   end
 end

data/lib/datadog/tracing/utils.rb

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+require_relative '../core/utils/forking'
+require_relative '../core/utils/time'
+
+module Datadog
+  module Tracing
+    # Utils contains low-level tracing utility functions.
+    # @public_api
+    module Utils
+      extend Datadog::Core::Utils::Forking
+
+      # The max value for a {Datadog::Tracing::Span} identifier.
+      # Span and trace identifiers should be strictly positive and strictly inferior to this limit.
+      #
+      # Limited to +2<<62-1+ positive integers, as Ruby is able to represent such numbers "inline",
+      # inside a +VALUE+ scalar, thus not requiring memory allocation.
+      #
+      # The range of IDs also has to consider portability across different languages and platforms.
+      RUBY_MAX_ID = (1 << 62) - 1
+
+      # Excludes zero from possible values
+      RUBY_ID_RANGE = (1..RUBY_MAX_ID).freeze
+
+      # While we only generate 63-bit integers due to limitations in other languages, we support
+      # parsing 64-bit integers for distributed tracing since an upstream system may generate one
+      EXTERNAL_MAX_ID = 1 << 64
+
+      # We use a custom random number generator because we want no interference
+      # with the default one. Using the default prng, we could break code that
+      # would rely on srand/rand sequences.
+
+      # Return a randomly generated integer, valid as a Span ID or Trace ID.
+      # This method is thread-safe and fork-safe.
+      def self.next_id
+        after_fork! { reset! }
+        id_rng.rand(RUBY_ID_RANGE)
+      end
+
+      def self.id_rng
+        @id_rng ||= Random.new
+      end
+
+      def self.reset!
+        @id_rng = Random.new
+      end
+
+      private_class_method :id_rng, :reset!
+
+      # The module handles bitwise operation for trace id
+      module TraceId
+        MAX = (1 << 128) - 1
+
+        module_function
+
+        # Format for generating 128 bits trace id =>
+        # - 32-bits : seconds since Epoch
+        # - 32-bits : set to zero,
+        # - 64 bits : random 64-bits
+        def next_id
+          return Utils.next_id unless Datadog.configuration.tracing.trace_id_128_bit_generation_enabled
+
+          concatenate(
+            Core::Utils::Time.now.to_i << 32,
+            Utils.next_id
+          )
+        end
+
+        def to_high_order(trace_id)
+          trace_id >> 64
+        end
+
+        def to_low_order(trace_id)
+          trace_id & 0xFFFFFFFFFFFFFFFF
+        end
+
+        def concatenate(high_order, low_order)
+          high_order << 64 | low_order
+        end
+      end
+    end
+  end
+end

data/lib/datadog/tracing/workers/trace_writer.rb

@@ -1,16 +1,13 @@
-# typed: false
+require_relative '../../core/worker'
+require_relative '../../core/workers/async'
+require_relative '../../core/workers/polling'
+require_relative '../../core/workers/queue'
 
-require 'datadog/core'
-require 'datadog/core/worker'
-require 'datadog/core/workers/async'
-require 'datadog/core/workers/polling'
-require 'datadog/core/workers/queue'
+require_relative '../buffer'
+require_relative '../pipeline'
+require_relative '../event'
 
-require 'datadog/tracing/buffer'
-require 'datadog/tracing/pipeline'
-require 'datadog/tracing/event'
-
-require 'ddtrace/transport/http'
+require_relative '../../../ddtrace/transport/http'
 
 module Datadog
   module Tracing

data/lib/datadog/tracing/workers.rb

@@ -1,9 +1,5 @@
-# typed: true
-
-require 'datadog/core'
-
-require 'datadog/tracing/buffer'
-require 'datadog/tracing/pipeline'
+require_relative 'buffer'
+require_relative 'pipeline'
 
 module Datadog
   module Tracing

data/lib/datadog/tracing/writer.rb

@@ -1,12 +1,8 @@
-# typed: true
+require_relative 'event'
+require_relative 'runtime/metrics'
+require_relative 'workers'
 
-require 'datadog/core'
-
-require 'datadog/tracing/event'
-require 'datadog/tracing/runtime/metrics'
-require 'datadog/tracing/workers'
-
-require 'ddtrace/transport/http'
+require_relative '../../ddtrace/transport/http'
 
 module Datadog
   module Tracing
@@ -175,6 +171,13 @@ module Datadog
           end
         end
       end
+
+      private
+
+      def reset_stats!
+        @traces_flushed = 0
+        @transport.stats.reset!
+      end
     end
   end
 end