datadog 2.3.0 → 2.5.0

data/lib/datadog/tracing/distributed/propagation.rb

@@ -3,6 +3,7 @@
 require_relative '../configuration/ext'
 require_relative '../trace_digest'
 require_relative '../trace_operation'
+require_relative '../../core/telemetry/logger'
 
 module Datadog
   module Tracing
@@ -43,6 +44,7 @@ module Datadog
         # DEV-2.0: if needed.
         # DEV-2.0: Ideally, we'd have a separate stream to report tracer errors and never
         # DEV-2.0: touch the active span.
+        # DEV-3.0: Sample trace here instead of when generating digest.
         #
         # @param digest [TraceDigest]
         # @param data [Hash]
@@ -72,6 +74,10 @@ module Datadog
             ::Datadog.logger.error(
               "Error injecting distributed trace data. Cause: #{e} Location: #{Array(e.backtrace).first}"
             )
+            ::Datadog::Core::Telemetry::Logger.report(
+              e,
+              description: "Error injecting distributed trace data with #{propagator.class.name}"
+            )
           end
 
           result
@@ -127,6 +133,7 @@ module Datadog
               )
             end
           rescue => e
+            # TODO: Not to report Telemetry logs for now
             ::Datadog.logger.error(
               "Error extracting distributed trace data. Cause: #{e} Location: #{Array(e.backtrace).first}"
             )

data/lib/datadog/tracing/metadata/ext.rb

@@ -87,6 +87,8 @@ module Datadog
           TAG_STATUS_CODE = 'http.status_code'
           TAG_USER_AGENT = 'http.useragent'
           TAG_URL = 'http.url'
+          TAG_ROUTE = 'http.route'
+          TAG_ROUTE_PATH = 'http.route.path'
           TYPE_INBOUND = AppTypes::TYPE_WEB.freeze
           TYPE_OUTBOUND = 'http'
           TYPE_PROXY = 'proxy'

data/lib/datadog/tracing/remote.rb

@@ -13,7 +13,10 @@ module Datadog
         PRODUCT = 'APM_TRACING'
 
         CAPABILITIES = [
-          1 << 29 # APM_TRACING_SAMPLE_RULES: Dynamic trace sampling rules configuration
+          1 << 12, # APM_TRACING_SAMPLE_RATE: Dynamic trace sampling rate configuration
+          1 << 13, # APM_TRACING_LOGS_INJECTION: Dynamic trace logs injection configuration
+          1 << 14, # APM_TRACING_HTTP_HEADER_TAGS: Dynamic trace HTTP header tags configuration
+          1 << 29, # APM_TRACING_SAMPLE_RULES: Dynamic trace sampling rules configuration
         ].freeze
 
         def products
@@ -45,7 +48,7 @@ module Datadog
           content.errored("#{e.class.name} #{e.message}: #{Array(e.backtrace).join("\n")}")
         end
 
-        def receivers
+        def receivers(_telemetry)
           receiver do |repository, _changes|
             # DEV: Filter our by product. Given it will be very common
             # DEV: we can filter this out before we receive the data in this method.

data/lib/datadog/tracing/sampling/matcher.rb

@@ -28,7 +28,7 @@ module Datadog
         # @return [#match?(String)]
         def self.glob_to_regex(glob)
           # Optimization for match-all case
-          return MATCH_ALL if glob == MATCH_ALL_PATTERN
+          return MATCH_ALL if /\A\*+\z/.match?(glob)
 
           # Ensure no undesired characters are treated as regex.
           glob = Regexp.quote(glob)
@@ -100,6 +100,11 @@ module Datadog
           @tags.all? do |name, matcher|
             tag = trace.get_tag(name)
 
+            # Floats: Matching floating point values with a non-zero decimal part is not supported.
+            # For floating point values with a non-zero decimal part, any all * pattern always returns true.
+            # Other patterns always return false.
+            return false if tag.is_a?(Float) && tag.truncate != tag && matcher != MATCH_ALL
+
             # Format metrics as strings, to allow for partial number matching (/4.*/ matching '400', '404', etc.).
             # Because metrics are floats, we use the '%g' format specifier to avoid trailing zeros, which
             # can affect exact string matching (e.g. '400' matching '400.0').

data/lib/datadog/tracing/sampling/rate_sampler.rb

@@ -20,7 +20,7 @@ module Datadog
           super()
 
           unless sample_rate >= 0.0 && sample_rate <= 1.0
-            Datadog.logger.error('sample rate is not between 0 and 1, falling back to 1')
+            Datadog.logger.warn('sample rate is not between 0 and 1, falling back to 1')
             sample_rate = 1.0
           end
 

data/lib/datadog/tracing/sampling/rule.rb

@@ -2,6 +2,7 @@
 
 require_relative 'matcher'
 require_relative 'rate_sampler'
+require_relative '../../core/telemetry/logger'
 
 module Datadog
   module Tracing
@@ -35,6 +36,7 @@ module Datadog
           Datadog.logger.error(
             "Matcher failed. Cause: #{e.class.name} #{e.message} Source: #{Array(e.backtrace).first}"
           )
+          Datadog::Core::Telemetry::Logger.report(e, description: 'Matcher failed')
           nil
         end
 

data/lib/datadog/tracing/sampling/rule_sampler.rb

@@ -1,8 +1,9 @@
 # frozen_string_literal: true
 
 require_relative 'ext'
-require_relative 'rate_limiter'
+require_relative '../../core/rate_limiter'
 require_relative 'rule'
+require_relative '../../core/telemetry/logger'
 
 module Datadog
   module Tracing
@@ -28,20 +29,22 @@ module Datadog
           default_sample_rate: Datadog.configuration.tracing.sampling.default_rate,
           default_sampler: nil
         )
-          @rules = rules
+          @rules = if default_sample_rate && !default_sampler
+                     # Add to the end of the rule list a rule always matches any trace
+                     rules << SimpleRule.new(sample_rate: default_sample_rate)
+                   else
+                     rules
+                   end
           @rate_limiter = if rate_limiter
                             rate_limiter
                           elsif rate_limit
-                            TokenBucket.new(rate_limit)
+                            Core::TokenBucket.new(rate_limit)
                           else
-                            UnlimitedLimiter.new
+                            Core::UnlimitedLimiter.new
                           end
-
           @default_sampler = if default_sampler
                                default_sampler
                              elsif default_sample_rate
-                               # Add to the end of the rule list a rule always matches any trace
-                               @rules << SimpleRule.new(sample_rate: default_sample_rate)
                                nil
                              else
                                # TODO: Simplify .tags access, as `Tracer#tags` can't be arbitrarily changed anymore
@@ -80,9 +83,10 @@ module Datadog
 
           new(parsed_rules, rate_limit: rate_limit, default_sample_rate: default_sample_rate)
         rescue => e
-          Datadog.logger.error do
+          Datadog.logger.warn do
             "Could not parse trace sampling rules '#{rules}': #{e.class.name} #{e.message} at #{Array(e.backtrace).first}"
           end
+
           nil
         end
 
@@ -121,7 +125,7 @@ module Datadog
 
           return false unless sampled
 
-          rate_limiter.allow?(1).tap do |allowed|
+          rate_limiter.allow?.tap do |allowed|
             set_priority(trace, allowed)
             set_limiter_metrics(trace, rate_limiter.effective_rate)
 
@@ -140,6 +144,8 @@ module Datadog
           Datadog.logger.error(
             "Rule sampling failed. Cause: #{e.class.name} #{e.message} Source: #{Array(e.backtrace).first}"
           )
+          Datadog::Core::Telemetry::Logger.report(e, description: 'Rule sampling failed')
+
           yield(trace)
         end
 

data/lib/datadog/tracing/sampling/span/ext.rb

@@ -9,7 +9,7 @@ module Datadog
           # Accept all spans (100% retention).
           DEFAULT_SAMPLE_RATE = 1.0
           # Unlimited.
-          # @see Datadog::Tracing::Sampling::TokenBucket
+          # @see Datadog::Core::TokenBucket
           DEFAULT_MAX_PER_SECOND = -1
 
           # Sampling decision method used to come to the sampling decision for this span

data/lib/datadog/tracing/sampling/span/rule.rb

@@ -30,7 +30,7 @@ module Datadog
             @rate_limit = rate_limit
 
             @sampler = Sampling::RateSampler.new(sample_rate)
-            @rate_limiter = Sampling::TokenBucket.new(rate_limit)
+            @rate_limiter = Core::TokenBucket.new(rate_limit)
           end
 
           # This method should only be invoked for spans that are part
@@ -54,7 +54,7 @@ module Datadog
           def sample!(trace_op, span_op)
             return :not_matched unless @matcher.match?(span_op)
 
-            if @rate_limiter.allow?(1) && @sampler.sample!(trace_op)
+            if @rate_limiter.allow? && @sampler.sample!(trace_op)
               span_op.set_metric(Span::Ext::TAG_MECHANISM, Sampling::Ext::Mechanism::SPAN_SAMPLING_RATE)
               span_op.set_metric(Span::Ext::TAG_RULE_RATE, @sample_rate)
               span_op.set_metric(Span::Ext::TAG_MAX_PER_SECOND, @rate_limit)

data/lib/datadog/tracing/trace_operation.rb

@@ -2,7 +2,7 @@
 
 require_relative '../core/environment/identity'
 require_relative '../core/utils'
-
+require_relative 'tracer'
 require_relative 'event'
 require_relative 'metadata/tagging'
 require_relative 'sampling/ext'
@@ -75,7 +75,9 @@ module Datadog
         metrics: nil,
         trace_state: nil,
         trace_state_unknown_fields: nil,
-        remote_parent: false
+        remote_parent: false,
+        tracer: nil
+
       )
         # Attributes
         @id = id || Tracing::Utils::TraceId.next_id
@@ -98,6 +100,7 @@ module Datadog
         @profiling_enabled = profiling_enabled
         @trace_state = trace_state
         @trace_state_unknown_fields = trace_state_unknown_fields
+        @tracer = tracer
 
         # Generic tags
         set_tags(tags) if tags
@@ -161,6 +164,23 @@ module Datadog
         @resource || (root_span && root_span.resource)
       end
 
+      # When retrieving tags or metrics we need to include root span tags for sampling purposes
+      def get_tag(key)
+        super || (root_span && root_span.get_tag(key))
+      end
+
+      def get_metric(key)
+        super || (root_span && root_span.get_metric(key))
+      end
+
+      def tags
+        all_tags = {}
+        all_tags.merge!(root_span&.tags || {}) if root_span
+        all_tags.merge!(super)
+
+        all_tags
+      end
+
       # Returns true if the resource has been explicitly set
       #
       # @return [Boolean]
@@ -284,10 +304,14 @@ module Datadog
       # Returns a set of trace headers used for continuing traces.
       # Used for propagation across execution contexts.
       # Data should reflect the active state of the trace.
+      # DEV-3.0: Sampling is a side effect of generating the digest.
+      # We should move the sample call to inject and right before moving to new contexts(threads, forking etc.)
       def to_digest
         # Resolve current span ID
         span_id = @active_span && @active_span.id
         span_id ||= @parent_span_id unless finished?
+        # sample the trace_operation with the tracer
+        @tracer&.sample_trace(self) unless sampling_priority
 
         TraceDigest.new(
           span_id: span_id,

data/lib/datadog/tracing/tracer.rb

@@ -262,6 +262,17 @@ module Datadog
         context.activate!(trace, &block)
       end
 
+      # Sample a span, tagging the trace as appropriate.
+      def sample_trace(trace_op)
+        begin
+          @sampler.sample!(trace_op)
+        rescue StandardError => 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
+
       # @!visibility private
       # TODO: make this private
       def trace_completed
@@ -331,12 +342,14 @@ module Datadog
             trace_state: digest.trace_state,
             trace_state_unknown_fields: digest.trace_state_unknown_fields,
             remote_parent: digest.span_remote,
+            tracer: self
           )
         else
           TraceOperation.new(
             hostname: hostname,
             profiling_enabled: profiling_enabled,
             remote_parent: false,
+            tracer: self
           )
         end
       end
@@ -347,7 +360,6 @@ module Datadog
         events.span_before_start.subscribe do |event_span_op, event_trace_op|
           event_trace_op.service ||= @default_service
           event_span_op.service ||= @default_service
-          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|
@@ -394,7 +406,7 @@ module Datadog
             on_error: on_error,
             resource: resource,
             service: service,
-            tags: resolve_tags(tags),
+            tags: resolve_tags(tags, service),
             type: type,
             id: id,
             &block
@@ -408,7 +420,7 @@ module Datadog
             resource: resource,
             service: service,
             start_time: start_time,
-            tags: resolve_tags(tags),
+            tags: resolve_tags(tags, service),
             type: type,
             id: id
           )
@@ -420,15 +432,20 @@ module Datadog
       end
       # rubocop:enable Lint/UnderscorePrefixedVariableName
 
-      def resolve_tags(tags)
-        if @tags.any? && tags
-          # Combine default tags with provided tags,
-          # preferring provided tags.
-          @tags.merge(tags)
-        else
-          # Use provided tags or default tags if none.
-          tags || @tags.dup
+      def resolve_tags(tags, service)
+        merged_tags = if @tags.any? && tags
+                        # Combine default tags with provided tags,
+                        # preferring provided tags.
+                        @tags.merge(tags)
+                      else
+                        # Use provided tags or default tags if none.
+                        tags || @tags.dup
+                      end
+        # Remove version tag if service is not the default service
+        if merged_tags.key?(Core::Environment::Ext::TAG_VERSION) && service != @default_service
+          merged_tags.delete(Core::Environment::Ext::TAG_VERSION)
         end
+        merged_tags
       end
 
       # Manually activate and deactivate the trace, when the span completes.
@@ -463,17 +480,6 @@ module Datadog
         end
       end
 
-      # Sample a span, tagging the trace as appropriate.
-      def sample_trace(trace_op)
-        begin
-          @sampler.sample!(trace_op)
-        rescue StandardError => 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
 
@@ -492,6 +498,7 @@ module Datadog
 
       # Flush finished spans from the trace buffer, send them to writer.
       def flush_trace(trace_op)
+        sample_trace(trace_op) unless trace_op.sampling_priority
         begin
           trace = @trace_flush.consume!(trace_op)
           write(trace) if trace && !trace.empty?

data/lib/datadog/tracing/transport/http/client.rb

@@ -38,6 +38,7 @@ module Datadog
             if stats.consecutive_errors > 0
               Datadog.logger.debug(message)
             else
+              # Not to report telemetry logs
               Datadog.logger.error(message)
             end
 

data/lib/datadog/tracing/transport/http.rb

@@ -74,6 +74,10 @@ module Datadog
             # Add container ID, if present.
             container_id = Datadog::Core::Environment::Container.container_id
             headers[Datadog::Core::Transport::Ext::HTTP::HEADER_CONTAINER_ID] = container_id unless container_id.nil?
+            # Pretend that stats computation are already done by the client
+            if Datadog.configuration.appsec.standalone.enabled
+              headers[Datadog::Core::Transport::Ext::HTTP::HEADER_CLIENT_COMPUTED_STATS] = 'yes'
+            end
           end
         end
 

data/lib/datadog/tracing/transport/io/client.rb

@@ -49,6 +49,7 @@ module Datadog
             if stats.consecutive_errors > 0
               Datadog.logger.debug(message)
             else
+              # Not to report telemetry logs
               Datadog.logger.error(message)
             end
 

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

@@ -43,7 +43,7 @@ module Datadog
           traces = process_traces(traces)
           flush_traces(traces)
         rescue StandardError => e
-          Datadog.logger.error(
+          Datadog.logger.warn(
             "Error while writing traces: dropped #{traces.length} items. Cause: #{e} Location: #{Array(e.backtrace).first}"
           )
         end

data/lib/datadog/tracing/workers.rb

@@ -56,7 +56,7 @@ module Datadog
             # ensures that the thread will not die because of an exception.
             # TODO[manu]: findout the reason and reschedule the send if it's not
             # a fatal exception
-            Datadog.logger.error(
+            Datadog.logger.warn(
               "Error during traces flush: dropped #{traces.length} items. Cause: #{e} Location: #{Array(e.backtrace).first}"
             )
           end
@@ -70,7 +70,7 @@ module Datadog
             @run = true
             Datadog.logger.debug { "Starting thread for: #{self}" }
             @worker = Thread.new { perform }
-            @worker.name = self.class.name unless Gem::Version.new(RUBY_VERSION) < Gem::Version.new('2.3')
+            @worker.name = self.class.name
             @worker.thread_variable_set(:fork_safe, true)
 
             nil

data/lib/datadog/tracing/writer.rb

@@ -68,21 +68,6 @@ module Datadog
         end
       end
 
-      # spawns a worker for spans; they share the same transport which is thread-safe
-      # @!visibility private
-      def start_worker
-        @trace_handler = ->(items, transport) { send_spans(items, transport) }
-        @worker = Workers::AsyncTransport.new(
-          transport: @transport,
-          buffer_size: @buff_size,
-          on_trace: @trace_handler,
-          interval: @flush_interval,
-          shutdown_timeout: @shutdown_timeout
-        )
-
-        @worker.start
-      end
-
       # Gracefully shuts down this writer.
       #
       # Once stopped methods calls won't fail, but
@@ -93,19 +78,6 @@ module Datadog
         @mutex_after_fork.synchronize { stop_worker }
       end
 
-      def stop_worker
-        @stopped = true
-
-        return if @worker.nil?
-
-        @worker.stop
-        @worker = nil
-
-        true
-      end
-
-      private :start_worker, :stop_worker
-
       # flush spans to the trace-agent, handles spans only
       # @!visibility private
       def send_spans(traces, transport)
@@ -179,6 +151,32 @@ module Datadog
 
       private
 
+      # spawns a worker for spans; they share the same transport which is thread-safe
+      # @!visibility private
+      def start_worker
+        @trace_handler = ->(items, transport) { send_spans(items, transport) }
+        @worker = Workers::AsyncTransport.new(
+          transport: @transport,
+          buffer_size: @buff_size,
+          on_trace: @trace_handler,
+          interval: @flush_interval,
+          shutdown_timeout: @shutdown_timeout
+        )
+
+        @worker.start
+      end
+
+      def stop_worker
+        @stopped = true
+
+        return if @worker.nil?
+
+        @worker.stop
+        @worker = nil
+
+        true
+      end
+
       def reset_stats!
         @traces_flushed = 0
         @transport.stats.reset!

data/lib/datadog/version.rb

@@ -3,7 +3,7 @@
 module Datadog
   module VERSION
     MAJOR = 2
-    MINOR = 3
+    MINOR = 5
     PATCH = 0
     PRE = nil
     BUILD = nil