ddtrace 0.53.0 → 1.9.0

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

@@ -0,0 +1,72 @@
+# typed: true
+
+require_relative '../../core'
+
+require_relative 'sampler'
+require_relative '../utils'
+
+module Datadog
+  module Tracing
+    module Sampling
+      # {Datadog::Tracing::Sampling::RateSampler} is based on a sample rate.
+      # @public_api
+      class RateSampler < Sampler
+        KNUTH_FACTOR = 1111111111111111111
+
+        # Initialize a {Datadog::Tracing::Sampling::RateSampler}.
+        # This sampler keeps a random subset of the traces. Its main purpose is to
+        # reduce the instrumentation footprint.
+        #
+        # * +sample_rate+: the sample rate as a {Float} between 0.0 and 1.0. 0.0
+        #   means that no trace will be sampled; 1.0 means that all traces will be
+        #   sampled.
+        #
+        # DEV-2.0: Allow for `sample_rate` zero (drop all) to be allowed. This eases
+        # DEV-2.0: usage for all internal users of the {RateSampler} class: both
+        # DEV-2.0: RuleSampler and Single Span Sampling leverage the RateSampler, but want
+        # DEV-2.0: `sample_rate` zero to mean "drop all". They work around this by hard-
+        # DEV-2.0: setting the `sample_rate` to zero like so:
+        # DEV-2.0: ```
+        # DEV-2.0: sampler = RateSampler.new
+        # DEV-2.0: sampler.sample_rate = sample_rate
+        # DEV-2.0: ```
+        def initialize(sample_rate = 1.0, decision: nil)
+          super()
+
+          unless sample_rate > 0.0 && sample_rate <= 1.0
+            Datadog.logger.error('sample rate is not between 0 and 1, disabling the sampler')
+            sample_rate = 1.0
+          end
+
+          self.sample_rate = sample_rate
+
+          @decision = decision
+        end
+
+        def sample_rate(*_)
+          @sample_rate
+        end
+
+        def sample_rate=(sample_rate)
+          @sample_rate = sample_rate
+          @sampling_id_threshold = sample_rate * Tracing::Utils::EXTERNAL_MAX_ID
+        end
+
+        def sample?(trace)
+          ((trace.id * KNUTH_FACTOR) % Tracing::Utils::EXTERNAL_MAX_ID) <= @sampling_id_threshold
+        end
+
+        def sample!(trace)
+          sampled = trace.sampled = sample?(trace)
+
+          return false unless sampled
+
+          trace.sample_rate = @sample_rate
+          trace.set_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER, @decision) if @decision
+
+          true
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,76 @@
+# typed: true
+
+require_relative '../../core'
+
+require_relative 'matcher'
+require_relative 'rate_sampler'
+
+module Datadog
+  module Tracing
+    module Sampling
+      # Sampling rule that dictates if a trace matches
+      # a specific criteria and what sampling strategy to
+      # apply in case of a positive match.
+      # @public_api
+      class Rule
+        attr_reader :matcher, :sampler
+
+        # @param [Matcher] matcher A matcher to verify trace conformity against
+        # @param [Sampler] sampler A sampler to be consulted on a positive match
+        def initialize(matcher, sampler)
+          @matcher = matcher
+          @sampler = sampler
+        end
+
+        # Evaluates if the provided `trace` conforms to the `matcher`.
+        #
+        # @param [TraceOperation] trace
+        # @return [Boolean] whether this rules applies to the trace
+        # @return [NilClass] if the matcher fails errs during evaluation
+        def match?(trace)
+          @matcher.match?(trace)
+        rescue => e
+          Datadog.logger.error(
+            "Matcher failed. Cause: #{e.class.name} #{e.message} Source: #{Array(e.backtrace).first}"
+          )
+          nil
+        end
+
+        # (see Datadog::Tracing::Sampling::Sampler#sample?)
+        def sample?(trace)
+          @sampler.sample?(trace)
+        end
+
+        # (see Datadog::Tracing::Sampling::Sampler#sample_rate)
+        def sample_rate(trace)
+          @sampler.sample_rate(trace)
+        end
+      end
+
+      # A {Datadog::Tracing::Sampling::Rule} that matches a trace based on
+      # trace name and/or service name and
+      # applies a fixed sampling to matching spans.
+      # @public_api
+      class SimpleRule < Rule
+        # @param name [String,Regexp,Proc] Matcher for case equality (===) with the trace name, defaults to always match
+        # @param service [String,Regexp,Proc] Matcher for case equality (===) with the service name,
+        #                defaults to always match
+        # @param sample_rate [Float] Sampling rate between +[0,1]+
+        def initialize(name: SimpleMatcher::MATCH_ALL, service: SimpleMatcher::MATCH_ALL, sample_rate: 1.0)
+          # We want to allow 0.0 to drop all traces, but {Datadog::Tracing::Sampling::RateSampler}
+          # considers 0.0 an invalid rate and falls back to 100% sampling.
+          #
+          # We address that here by not setting the rate in the constructor,
+          # but using the setter method.
+          #
+          # We don't want to make this change directly to {Datadog::Tracing::Sampling::RateSampler}
+          # because it breaks its current contract to existing users.
+          sampler = RateSampler.new
+          sampler.sample_rate = sample_rate
+
+          super(SimpleMatcher.new(name: name, service: service), sampler)
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,132 @@
+# typed: true
+
+require_relative '../../core'
+
+require_relative 'ext'
+require_relative 'rate_limiter'
+require_relative 'rule'
+
+module Datadog
+  module Tracing
+    module Sampling
+      # Span {Sampler} that applies a set of {Rule}s to decide
+      # on sampling outcome. Then, a rate limiter is applied.
+      #
+      # If a trace does not conform to any rules, a default
+      # sampling strategy is applied.
+      # @public_api
+      class RuleSampler
+        attr_reader :rules, :rate_limiter, :default_sampler
+
+        # @param rules [Array<Rule>] ordered list of rules to be applied to a trace
+        # @param rate_limit [Float] number of traces per second, defaults to +100+
+        # @param rate_limiter [RateLimiter] limiter applied after rule matching
+        # @param default_sample_rate [Float] fallback sample rate when no rules apply to a trace,
+        #   between +[0,1]+, defaults to +1+
+        # @param default_sampler [Sample] fallback strategy when no rules apply to a trace
+        def initialize(
+          rules = [],
+          rate_limit: Datadog.configuration.tracing.sampling.rate_limit,
+          rate_limiter: nil,
+          default_sample_rate: Datadog.configuration.tracing.sampling.default_rate,
+          default_sampler: nil
+        )
+          @rules = rules
+          @rate_limiter = if rate_limiter
+                            rate_limiter
+                          elsif rate_limit
+                            TokenBucket.new(rate_limit)
+                          else
+                            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
+                               RateByServiceSampler.new(1.0, env: -> { Tracing.send(:tracer).tags['env'] })
+                             end
+        end
+
+        # /RuleSampler's components (it's rate limiter, for example) are
+        # not be guaranteed to be size-effect free.
+        # It is not possible to guarantee that a call to {#sample?} will
+        # return the same result as a successive call to {#sample!} with the same trace.
+        #
+        # Use {#sample!} instead
+        def sample?(_trace)
+          raise 'RuleSampler cannot be evaluated without side-effects'
+        end
+
+        def sample!(trace)
+          sampled = sample_trace(trace) do |t|
+            @default_sampler.sample!(t).tap do
+              # We want to make sure the trace is tagged with the agent-derived
+              # service rate. Retrieve this from the rate by service sampler.
+              # Only do this if it was set by a RateByServiceSampler.
+              trace.agent_sample_rate = @default_sampler.sample_rate(trace) if @default_sampler.is_a?(RateByServiceSampler)
+            end
+          end
+
+          trace.sampled = sampled
+        end
+
+        # @!visibility private
+        def update(*args, **kwargs)
+          return false unless @default_sampler.respond_to?(:update)
+
+          @default_sampler.update(*args, **kwargs)
+        end
+
+        private
+
+        def sample_trace(trace)
+          rule = @rules.find { |r| r.match?(trace) }
+
+          return yield(trace) if rule.nil?
+
+          sampled = rule.sample?(trace)
+          sample_rate = rule.sample_rate(trace)
+
+          set_priority(trace, sampled)
+          set_rule_metrics(trace, sample_rate)
+
+          return false unless sampled
+
+          rate_limiter.allow?(1).tap do |allowed|
+            set_priority(trace, allowed)
+            set_limiter_metrics(trace, rate_limiter.effective_rate)
+            trace.set_tag(Tracing::Metadata::Ext::Distributed::TAG_DECISION_MAKER, Ext::Decision::TRACE_SAMPLING_RULE)
+          end
+        rescue StandardError => e
+          Datadog.logger.error(
+            "Rule sampling failed. Cause: #{e.class.name} #{e.message} Source: #{Array(e.backtrace).first}"
+          )
+          yield(trace)
+        end
+
+        # Span priority should only be set when the {RuleSampler}
+        # was responsible for the sampling decision.
+        def set_priority(trace, sampled)
+          trace.sampling_priority = if sampled
+                                      Sampling::Ext::Priority::USER_KEEP
+                                    else
+                                      Sampling::Ext::Priority::USER_REJECT
+                                    end
+        end
+
+        def set_rule_metrics(trace, sample_rate)
+          trace.rule_sample_rate = sample_rate
+        end
+
+        def set_limiter_metrics(trace, limiter_rate)
+          trace.rate_limiter_rate = limiter_rate
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,44 @@
+# typed: true
+
+module Datadog
+  module Tracing
+    module Sampling
+      # Interface for client-side trace sampling.
+      # @abstract
+      # @public_api
+      class Sampler
+        # Returns `true` if the provided trace should be kept.
+        # Otherwise, `false`.
+        #
+        # This method *must not* modify the `trace`.
+        #
+        # @param [Datadog::Tracing::TraceOperation] trace
+        # @return [Boolean] should this trace be kept?
+        def sample?(trace)
+          raise NotImplementedError, 'Samplers must implement the #sample? method'
+        end
+
+        # Returns `true` if the provided trace should be kept.
+        # Otherwise, `false`.
+        #
+        # This method *may* modify the `trace`, in case changes are necessary based on the
+        # sampling decision.
+        #
+        # @param [Datadog::Tracing::TraceOperation] trace
+        # @return [Boolean] should this trace be kept?
+        def sample!(trace)
+          raise NotImplementedError, 'Samplers must implement the #sample! method'
+        end
+
+        # The sampling rate, if this sampler has such concept.
+        # Otherwise, `nil`.
+        #
+        # @param [Datadog::Tracing::TraceOperation] trace
+        # @return [Float,nil] sampling ratio between 0.0 and 1.0 (inclusive), or `nil` if not applicable
+        def sample_rate(trace)
+          raise NotImplementedError, 'Samplers must implement the #sample_rate method'
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Tracing
+    module Sampling
+      module Span
+        # Single Span Sampling constants.
+        module Ext
+          # Accept all spans (100% retention).
+          DEFAULT_SAMPLE_RATE = 1.0
+          # Unlimited.
+          # @see Datadog::Tracing::Sampling::TokenBucket
+          DEFAULT_MAX_PER_SECOND = -1
+
+          # Sampling decision method used to come to the sampling decision for this span
+          TAG_MECHANISM = '_dd.span_sampling.mechanism'
+          # Sampling rate applied to this span, if a rule applies
+          TAG_RULE_RATE = '_dd.span_sampling.rule_rate'
+          # Rate limit configured for this span, if a rule applies
+          TAG_MAX_PER_SECOND = '_dd.span_sampling.max_per_second'
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,89 @@
+# frozen_string_literal: true
+
+module Datadog
+  module Tracing
+    module Sampling
+      module Span
+        # Checks if a span conforms to a matching criteria.
+        class Matcher
+          attr_reader :name, :service
+
+          # Pattern that matches any string
+          MATCH_ALL_PATTERN = '*'
+
+          # Matches span name and service to their respective patterns provided.
+          #
+          # The patterns are {String}s with two special characters available:
+          # 1. `?`: matches exactly one of any character.
+          # 2. `*`: matches a substring of any size, including zero.
+          # These patterns can occur any point of the string, any number of times.
+          #
+          # Both {SpanOperation#name} and {SpanOperation#service} must match the provided patterns.
+          #
+          # The whole String has to match the provided patterns: providing a pattern that
+          # matches a portion of the provided String is not considered a match.
+          #
+          # @example web-*
+          #   `'web-*'` will match any string starting with `web-`.
+          # @example cache-?
+          #   `'cache-?'` will match any string starting with `database-` followed by exactly one character.
+          #
+          # @param name_pattern [String] a pattern to be matched against {SpanOperation#name}
+          # @param service_pattern [String] a pattern to be matched against {SpanOperation#service}
+          def initialize(name_pattern: MATCH_ALL_PATTERN, service_pattern: MATCH_ALL_PATTERN)
+            @name = pattern_to_regex(name_pattern)
+            @service = pattern_to_regex(service_pattern)
+          end
+
+          # {Regexp#match?} was added in Ruby 2.4, and it's measurably
+          # the least costly way to get a boolean result for a Regexp match.
+          # @see https://www.ruby-lang.org/en/news/2016/12/25/ruby-2-4-0-released/
+          if Regexp.method_defined?(:match?)
+            # Returns `true` if the span conforms to the configured patterns,
+            # `false` otherwise
+            #
+            # @param [SpanOperation] span
+            # @return [Boolean]
+            def match?(span)
+              # Matching is performed at the end of the lifecycle of a Span,
+              # thus both `name` and `service` are guaranteed to be not `nil`.
+              @name.match?(span.name) && @service.match?(span.service)
+            end
+          else
+            # DEV: Remove when support for Ruby 2.3 and older is removed.
+            def match?(span)
+              @name === span.name && @service === span.service
+            end
+          end
+
+          def ==(other)
+            return super unless other.is_a?(Matcher)
+
+            name == other.name &&
+              service == other.service
+          end
+
+          private
+
+          # @param pattern [String]
+          # @return [Regexp]
+          def pattern_to_regex(pattern)
+            # Ensure no undesired characters are treated as regex.
+            # Our valid special characters, `?` and `*`,
+            # will be escaped so...
+            pattern = Regexp.quote(pattern)
+
+            # ...we account for that here:
+            pattern.gsub!('\?', '.') # Any single character
+            pattern.gsub!('\*', '.*') # Any substring
+
+            # Patterns have to match the whole input string
+            pattern = "\\A#{pattern}\\z"
+
+            Regexp.new(pattern)
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,82 @@
+# frozen_string_literal: true
+
+require_relative 'ext'
+
+module Datadog
+  module Tracing
+    module Sampling
+      module Span
+        # Span sampling rule that applies a sampling rate if the span
+        # matches the provided {Matcher}.
+        # Additionally, a rate limiter is also applied.
+        #
+        # If a span does not conform to the matcher, no changes are made.
+        class Rule
+          attr_reader :matcher, :sample_rate, :rate_limit
+
+          # Creates a new span sampling rule.
+          #
+          # @param [Sampling::Span::Matcher] matcher whether this rule applies to a specific span
+          # @param [Float] sample_rate span sampling ratio, between 0.0 (0%) and 1.0 (100%).
+          # @param [Numeric] rate_limit maximum number of spans sampled per second. Negative numbers mean unlimited spans.
+          def initialize(
+            matcher,
+            sample_rate: Span::Ext::DEFAULT_SAMPLE_RATE,
+            rate_limit: Span::Ext::DEFAULT_MAX_PER_SECOND
+          )
+
+            @matcher = matcher
+            @sample_rate = sample_rate
+            @rate_limit = rate_limit
+
+            @sampler = Sampling::RateSampler.new
+            # Set the sample_rate outside of the initializer to allow for
+            # zero to be a "drop all".
+            # The RateSampler initializer enforces non-zero, falling back to 100% sampling
+            # if zero is provided.
+            @sampler.sample_rate = sample_rate
+            @rate_limiter = Sampling::TokenBucket.new(rate_limit)
+          end
+
+          # This method should only be invoked for spans that are part
+          # of a trace that has been dropped by trace-level sampling.
+          # Invoking it for other spans will cause incorrect sampling
+          # metrics to be reported by the Datadog App.
+          #
+          # Returns `true` if the provided span is sampled.
+          # If the span is dropped due to sampling rate or rate limiting,
+          # it returns `false`.
+          #
+          # Returns `nil` if the span did not meet the matching criteria by the
+          # provided matcher.
+          #
+          # This method modifies the `span` if it matches the provided matcher.
+          #
+          # @param [Datadog::Tracing::SpanOperation] span_op span to be sampled
+          # @return [:kept,:rejected] should this span be sampled?
+          # @return [:not_matched] span did not satisfy the matcher, no changes are made to the span
+          def sample!(span_op)
+            return :not_matched unless @matcher.match?(span_op)
+
+            if @sampler.sample?(span_op) && @rate_limiter.allow?(1)
+              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)
+              :kept
+            else
+              :rejected
+            end
+          end
+
+          def ==(other)
+            return super unless other.is_a?(Rule)
+
+            matcher == other.matcher &&
+              sample_rate == other.sample_rate &&
+              rate_limit == other.rate_limit
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,104 @@
+# frozen_string_literal: true
+
+require 'json'
+
+require_relative 'ext'
+require_relative 'matcher'
+require_relative 'rule'
+
+module Datadog
+  module Tracing
+    module Sampling
+      module Span
+        # Converts user configuration into {Datadog::Tracing::Sampling::Span::Rule} objects,
+        # handling any parsing errors.
+        module RuleParser
+          class << self
+            # Parses the provided JSON string containing the Single Span
+            # Sampling configuration list.
+            # In case of parsing errors, `nil` is returned.
+            #
+            # @param rules [String] the JSON configuration rules to be parsed
+            # @return [Array<Datadog::Tracing::Sampling::Span::Rule>] a list of parsed rules
+            # @return [nil] if parsing failed
+            def parse_json(rules)
+              return nil unless rules
+
+              begin
+                list = JSON.parse(rules)
+              rescue => e
+                Datadog.logger.warn(
+                  "Error parsing Span Sampling Rules `#{rules.inspect}`: "\
+                  "#{e.class.name} #{e.message} at #{Array(e.backtrace).first}"
+                )
+                return nil
+              end
+
+              parse_list(list)
+            end
+
+            # Parses a list of Hashes containing the parsed JSON information
+            # for Single Span Sampling configuration.
+            # In case of parsing errors, `nil` is returned.
+            #
+            # @param rules [Array<String] the JSON configuration rules to be parsed
+            # @return [Array<Datadog::Tracing::Sampling::Span::Rule>] a list of parsed rules
+            # @return [nil] if parsing failed
+            def parse_list(rules)
+              unless rules.is_a?(Array)
+                Datadog.logger.warn("Span Sampling Rules are not an array: #{rules.inspect}")
+                return nil
+              end
+
+              parsed = rules.map do |hash|
+                unless hash.is_a?(Hash)
+                  Datadog.logger.warn("Span Sampling Rule is not a key-value object: #{hash.inspect}")
+                  return nil
+                end
+
+                begin
+                  parse_rule(hash)
+                rescue => e
+                  Datadog.logger.warn(
+                    "Cannot parse Span Sampling Rule #{hash.inspect}: " \
+                    "#{e.class.name} #{e} at #{Array(e.backtrace).first}"
+                  )
+                  return nil
+                end
+              end
+
+              parsed.compact!
+              parsed
+            end
+
+            private
+
+            def parse_rule(hash)
+              matcher_options = {}
+              if (name_pattern = hash['name'])
+                matcher_options[:name_pattern] = name_pattern
+              end
+
+              if (service_pattern = hash['service'])
+                matcher_options[:service_pattern] = service_pattern
+              end
+
+              matcher = Matcher.new(**matcher_options)
+
+              rule_options = {}
+              if (sample_rate = hash['sample_rate'])
+                rule_options[:sample_rate] = sample_rate
+              end
+
+              if (max_per_second = hash['max_per_second'])
+                rule_options[:rate_limit] = max_per_second
+              end
+
+              Rule.new(matcher, **rule_options)
+            end
+          end
+        end
+      end
+    end
+  end
+end