datadog 2.8.0 → 2.39.0

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

@@ -10,42 +10,54 @@ module Datadog
       # @public_api
       module Ext
         # Name of package that was instrumented
-        TAG_COMPONENT = 'component'
+        TAG_COMPONENT = "component"
         # Type of operation being performed (e.g. )
-        TAG_OPERATION = 'operation'
+        TAG_OPERATION = "operation"
         # Hostname of external service interacted with
         #
         # This tag also doesn't strictly need to be a “hostname”. It can be a raw IP address and in some cases it
         # can even be a unix domain socket (i.e. postgres client setting host=/var/run/postgres).
         # It should be whatever the client uses to point at the server it’s trying to talk to.
-        TAG_PEER_HOSTNAME = 'peer.hostname'
+        TAG_PEER_HOSTNAME = "peer.hostname"
         # Name of external service that performed the work
-        TAG_PEER_SERVICE = 'peer.service'
+        TAG_PEER_SERVICE = "peer.service"
 
-        TAG_KIND = 'span.kind'
+        TAG_KIND = "span.kind"
 
         # Set this tag to `1.0` if the span is a Service Entry span.
-        TAG_TOP_LEVEL = '_dd.top_level'
+        TAG_TOP_LEVEL = "_dd.top_level"
 
         # Set to `1.0` if profiling is enabled together with tracing, and `0.0` otherwise
         # See Datadog-internal "RFC: Identifying which spans have profiling enabled " for details
-        TAG_PROFILING_ENABLED = '_dd.profiling.enabled'
+        TAG_PROFILING_ENABLED = "_dd.profiling.enabled"
+
+        TAG_APM_ENABLED = "_dd.apm.enabled"
+
+        # Set to the global service name when a span's service is overridden
+        TAG_BASE_SERVICE = "_dd.base_service"
+
+        # Set to the source of a span's service override: the component name for Datadog-instrumented spans
+        # (e.g. 'redis', 'pg'), or {SVC_SRC_MANUAL} for manually-instrumented spans
+        TAG_SVC_SRC = "_dd.svc_src"
+
+        # Value for {TAG_SVC_SRC} indicating that a span was manually instrumented via the Tracing API
+        SVC_SRC_MANUAL = "m"
 
         # Defines constants for trace analytics
         # @public_api
         module Analytics
           DEFAULT_SAMPLE_RATE = 1.0
-          TAG_ENABLED = 'analytics.enabled'
-          TAG_MEASURED = '_dd.measured'
-          TAG_SAMPLE_RATE = '_dd1.sr.eausr'
+          TAG_ENABLED = "analytics.enabled"
+          TAG_MEASURED = "_dd.measured"
+          TAG_SAMPLE_RATE = "_dd1.sr.eausr"
         end
 
         module AppTypes
-          TYPE_WEB = 'web'
-          TYPE_DB = 'db'
-          TYPE_CACHE = 'cache'
-          TYPE_WORKER = 'worker'
-          TYPE_CUSTOM = 'custom'
+          TYPE_WEB = "web"
+          TYPE_DB = "db"
+          TYPE_CACHE = "cache"
+          TYPE_WORKER = "worker"
+          TYPE_CUSTOM = "custom"
         end
 
         # @public_api
@@ -53,48 +65,63 @@ module Datadog
         module Distributed
           # What mechanism was used to make this trace's sampling decision.
           # @see Datadog::Tracing::Sampling::Ext::Mechanism
-          TAG_DECISION_MAKER = '_dd.p.dm'
+          TAG_DECISION_MAKER = "_dd.p.dm"
 
-          TAG_ORIGIN = '_dd.origin'
-          TAG_SAMPLING_PRIORITY = '_sampling_priority_v1'
+          # Bitmask for which product generated an event. E.g.: 2 for an AppSec event.
+          TAG_TRACE_SOURCE = "_dd.p.ts"
 
-          TAG_DD_PARENT_ID = '_dd.parent_id'
-          DD_PARENT_ID_DEFAULT = '0000000000000000'
+          TAG_ORIGIN = "_dd.origin"
+          TAG_SAMPLING_PRIORITY = "_sampling_priority_v1"
+
+          TAG_DD_PARENT_ID = "_dd.parent_id"
+          DD_PARENT_ID_DEFAULT = "0000000000000000"
+
+          # Knuth Sampling Rate: the sampling rate applied by agent-based or rule-based sampling.
+          # This is a propagated tag (prefixed with `_dd.p.`) that is included in `x-datadog-tags`.
+          TAG_KNUTH_SAMPLING_RATE = "_dd.p.ksr"
 
           # Trace tags with this prefix will propagate from a trace through distributed tracing.
           # Distributed headers tags with this prefix will be injected into the active trace.
-          TAGS_PREFIX = '_dd.p.'
+          TAGS_PREFIX = "_dd.p."
 
           # The distributed tag to carry hex encoded high order 64 bits of 127 bits trace id during
           # the context restricted with 64 bits. Such as, Datadog propagation and messagepack encoding
-          TID = 'tid'
+          TID = "tid"
           TAG_TID = TAGS_PREFIX + TID
         end
 
         # @public_api
         module Errors
           STATUS = 1
-          TAG_MSG = 'error.message'
-          TAG_STACK = 'error.stack'
-          TAG_TYPE = 'error.type'
+          TAG_MSG = "error.message"
+          TAG_STACK = "error.stack"
+          TAG_TYPE = "error.type"
+
+          # From https://opentelemetry.io/docs/specs/semconv/registry/attributes/exception/#exception-stacktrace
+          ATTRIBUTE_MESSAGE = "exception.message"
+          ATTRIBUTE_STACKTRACE = "exception.stacktrace"
+          ATTRIBUTE_TYPE = "exception.type"
+
+          # From https://opentelemetry.io/docs/specs/semconv/exceptions/exceptions-spans/#exception-event
+          EVENT_NAME = "exception"
         end
 
         # @public_api
         module HTTP
-          ERROR_RANGE = (500...600).freeze
-          TAG_BASE_URL = 'http.base_url'
-          TAG_METHOD = 'http.method'
-          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'
+          TAG_BASE_URL = "http.base_url"
+          TAG_ENDPOINT = "http.endpoint"
+          TAG_METHOD = "http.method"
+          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'
-          TYPE_TEMPLATE = 'template'
-          TAG_CLIENT_IP = 'http.client_ip'
-          HEADER_USER_AGENT = 'User-Agent'
+          TYPE_OUTBOUND = "http"
+          TYPE_PROXY = "proxy"
+          TYPE_TEMPLATE = "template"
+          TAG_CLIENT_IP = "http.client_ip"
+          HEADER_USER_AGENT = "User-Agent"
 
           # General header functionality
           module Headers
@@ -120,19 +147,19 @@ module Datadog
               # All other characters are replaced with an underscore.
               tag = name.to_s.strip
               tag.downcase!
-              tag.gsub!(INVALID_TAG_CHARACTERS, '_')
+              tag.gsub!(INVALID_TAG_CHARACTERS, "_")
 
               # Additional HTTP header normalization.
               #
               # Periods are replaced with an underscore.
-              tag.tr!('.', '_') unless allow_nested
+              tag.tr!(".", "_") unless allow_nested
               tag
             end
           end
 
           # Request headers
           module RequestHeaders
-            PREFIX = 'http.request.headers'
+            PREFIX = "http.request.headers"
 
             module_function
 
@@ -143,7 +170,7 @@ module Datadog
 
           # Response headers
           module ResponseHeaders
-            PREFIX = 'http.response.headers'
+            PREFIX = "http.response.headers"
 
             module_function
 
@@ -155,42 +182,42 @@ module Datadog
 
         # @public_api
         module NET
-          TAG_HOSTNAME = '_dd.hostname'
-          TAG_TARGET_HOST = 'out.host'
-          TAG_TARGET_PORT = 'out.port'
-          TAG_DESTINATION_NAME = 'network.destination.name'
-          TAG_DESTINATION_PORT = 'network.destination.port'
+          TAG_HOSTNAME = "_dd.hostname"
+          TAG_TARGET_HOST = "out.host"
+          TAG_TARGET_PORT = "out.port"
+          TAG_DESTINATION_NAME = "network.destination.name"
+          TAG_DESTINATION_PORT = "network.destination.port"
         end
 
         # @public_api
         module Sampling
-          TAG_AGENT_RATE = '_dd.agent_psr'
+          TAG_AGENT_RATE = "_dd.agent_psr"
 
           # If rule sampling is applied to a span, set this metric the sample rate configured for that rule.
           # This should be done regardless of sampling outcome.
-          TAG_RULE_SAMPLE_RATE = '_dd.rule_psr'
+          TAG_RULE_SAMPLE_RATE = "_dd.rule_psr"
 
           # If rate limiting is checked on a span, set this metric the effective rate limiting rate applied.
           # This should be done regardless of rate limiting outcome.
-          TAG_RATE_LIMITER_RATE = '_dd.limit_psr'
+          TAG_RATE_LIMITER_RATE = "_dd.limit_psr"
 
-          TAG_SAMPLE_RATE = '_sample_rate'
+          TAG_SAMPLE_RATE = "_sample_rate"
         end
 
         # @public_api
         module SQL
-          TYPE = 'sql'
-          TAG_QUERY = 'sql.query'
+          TYPE = "sql"
+          TAG_QUERY = "sql.query"
         end
 
         # @public_api
         module SpanKind
-          TAG_PROXY = 'proxy'
-          TAG_SERVER = 'server'
-          TAG_CLIENT = 'client'
-          TAG_PRODUCER = 'producer'
-          TAG_CONSUMER = 'consumer'
-          TAG_INTERNAL = 'internal'
+          TAG_PROXY = "proxy"
+          TAG_SERVER = "server"
+          TAG_CLIENT = "client"
+          TAG_PRODUCER = "producer"
+          TAG_CONSUMER = "consumer"
+          TAG_INTERNAL = "internal"
         end
 
         # @public_api

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

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require "forwardable"
+
+module Datadog
+  module Tracing
+    module Metadata
+      # This class is a data structure that is used to store
+      # complex metadata, such as an array of objects.
+      #
+      # It is serialized to MessagePack format when sent to the agent.
+      class Metastruct
+        extend Forwardable
+
+        def_delegators :@metastruct, :[], :[]=, :to_h
+
+        def initialize
+          @metastruct = {}
+        end
+
+        def to_msgpack(packer = nil)
+          # JRuby doesn't pass the packer
+          packer ||= MessagePack::Packer.new
+
+          packer.write(@metastruct.transform_values(&:to_msgpack))
+        end
+
+        def pretty_print(q)
+          q.seplist @metastruct.each do |key, value|
+            q.text "#{key} => #{value}"
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+require_relative "metastruct"
+
+module Datadog
+  module Tracing
+    module Metadata
+      # Adds data storage for the `meta_struct` field.
+      #
+      # This field is used to send more complex data like an array of objects
+      # in MessagePack format to the agent, and has no size limitations.
+      #
+      # The agent fully supports meta_struct from version v7.35.0 (April 2022).
+      #
+      # On versions older than v7.35.0, sending traces containing meta_struct
+      # has no unexpected side-effects; traces are sent to the backend as expected,
+      # while the meta_struct field is stripped.
+      module MetastructTagging
+        # Set the given key / value tag pair on the metastruct.
+        #
+        # A valid example is:
+        #
+        #   span.set_metastruct_tag('_dd.stack', [])
+        def set_metastruct_tag(key, value)
+          metastruct[key] = value
+        end
+
+        # Return the metastruct tag value for the given key,
+        # returns nil if the key doesn't exist.
+        def get_metastruct_tag(key)
+          metastruct[key]
+        end
+
+        private
+
+        def metastruct
+          @metastruct ||= Metastruct.new
+        end
+      end
+    end
+  end
+end

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

@@ -1,8 +1,8 @@
 # frozen_string_literal: true
 
-require_relative '../../core/environment/ext'
+require_relative "../../core/environment/ext"
 
-require_relative 'ext'
+require_relative "ext"
 
 module Datadog
   module Tracing
@@ -51,8 +51,8 @@ module Datadog
             # Encode strings in UTF-8 to facilitate downstream serialization
             meta[Core::Utils.utf8_encode(key)] = Core::Utils.utf8_encode(value)
           end
-        rescue StandardError => e
-          Datadog.logger.debug("Unable to set the tag #{key}, ignoring it. Caused by: #{e}")
+        rescue => e
+          Datadog.logger.debug("Unable to set the tag #{key}, ignoring it. Caused by: #{e.class}: #{e.message}")
         end
 
         # Sets tags from given hash, for each key in hash it sets the tag with that key
@@ -80,10 +80,10 @@ module Datadog
         end
 
         # Convenient interface for setting a single tag.
-        alias []= set_tag
+        alias_method :[]=, :set_tag
 
         # Convenient interface for getting a single tag.
-        alias [] get_tag
+        alias_method :[], :get_tag
 
         # Return the metric with the given key, nil if it doesn't exist.
         def get_metric(key)
@@ -101,8 +101,8 @@ module Datadog
 
           # Encode strings in UTF-8 to facilitate downstream serialization
           metrics[Core::Utils.utf8_encode(key)] = value
-        rescue StandardError => e
-          Datadog.logger.debug("Unable to set the metric #{key}, ignoring it. Caused by: #{e}")
+        rescue => e
+          Datadog.logger.debug("Unable to set the metric #{key}, ignoring it. Caused by: #{e.class}: #{e.message}")
         end
 
         # This method removes a metric for the given key. It acts like {#clear_tag}.

data/lib/datadog/tracing/metadata.rb

@@ -1,8 +1,9 @@
 # frozen_string_literal: true
 
-require_relative 'metadata/analytics'
-require_relative 'metadata/tagging'
-require_relative 'metadata/errors'
+require_relative "metadata/analytics"
+require_relative "metadata/tagging"
+require_relative "metadata/metastruct_tagging"
+require_relative "metadata/errors"
 
 module Datadog
   module Tracing
@@ -10,6 +11,7 @@ module Datadog
     module Metadata
       def self.included(base)
         base.include(Metadata::Tagging)
+        base.include(Metadata::MetastructTagging)
         base.include(Metadata::Errors)
 
         # Additional extensions

data/lib/datadog/tracing/pipeline/span_filter.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require 'set'
-require_relative 'span_processor'
+require "set"
+require_relative "span_processor"
 
 module Datadog
   module Tracing
@@ -38,7 +38,9 @@ module Datadog
         private
 
         def drop_it?(span)
-          @operation.call(span) rescue false
+          @operation.call(span)
+        rescue
+          false
         end
       end
     end

data/lib/datadog/tracing/pipeline/span_processor.rb

@@ -28,7 +28,9 @@ module Datadog
         # @!visibility private
         def call(trace)
           trace.spans.each do |span|
-            @operation.call(span) rescue next
+            @operation.call(span)
+          rescue
+            next
           end
 
           trace

data/lib/datadog/tracing/pipeline.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
-require_relative 'pipeline/span_filter'
-require_relative 'pipeline/span_processor'
+require_relative "pipeline/span_filter"
+require_relative "pipeline/span_processor"
 
 module Datadog
   module Tracing
@@ -47,11 +47,11 @@ module Datadog
           next nil if current_trace.nil? || current_trace.empty?
 
           process_result = processor.call(current_trace)
-          process_result && process_result.empty? ? nil : process_result
+          (process_result && process_result.empty?) ? nil : process_result
         end
       rescue => e
         Datadog.logger.debug(
-          "trace dropped entirely due to `Pipeline.before_flush` error: #{e}"
+          "trace dropped entirely due to `Pipeline.before_flush` error: #{e.class}: #{e.message}"
         )
 
         nil

data/lib/datadog/tracing/remote.rb

@@ -1,22 +1,23 @@
 # frozen_string_literal: true
 
-require_relative '../core/remote/dispatcher'
-require_relative 'configuration/dynamic'
+require_relative "../core/remote/dispatcher"
+require_relative "configuration/dynamic"
 
 module Datadog
   module Tracing
     # Remote configuration declaration
     module Remote
-      class ReadError < StandardError; end
-
       class << self
-        PRODUCT = 'APM_TRACING'
+        PRODUCT = "APM_TRACING"
 
         CAPABILITIES = [
           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
+          # APM_TRACING_ENABLE_DYNAMIC_INSTRUMENTATION (bit 38) is declared in
+          # DI::Remote.capabilities, not here, so it is registered only when DI
+          # is not explicitly disabled and the runtime supports DI.
         ].freeze
 
         def products
@@ -27,8 +28,8 @@ module Datadog
           CAPABILITIES
         end
 
-        def process_config(config, content)
-          lib_config = config['lib_config']
+        def process_config(config, content, repository = nil)
+          lib_config = config["lib_config"]
 
           env_vars = Datadog::Tracing::Configuration::Dynamic::OPTIONS.map do |name, env_var, option|
             value = lib_config[name]
@@ -41,11 +42,46 @@ module Datadog
             [env_var, value]
           end
 
+          if (di_enabled = lib_config["dynamic_instrumentation_enabled"]) != nil # rubocop:disable Style/NonNilCheck
+            # repository is forwarded so that an enable signal can reconcile DI
+            # against probes delivered in an earlier poll while DI was stopped
+            # (see Datadog::DI::Remote.handle_rc_enablement).
+            Datadog::DI::Remote.handle_rc_enablement(di_enabled, repository)
+
+            components = Datadog.send(:components, allow_initialization: false)
+            di_products = Datadog::DI::Remote.products +
+              Datadog::SymbolDatabase::Remote.deferred_products(Datadog.configuration)
+
+            if di_enabled
+              components&.symbol_database&.resume_pending_upload
+              # Advertise the DI products only if the component actually started.
+              # handle_rc_enablement above no-ops when DI cannot run — the
+              # component is nil on an unsupported runtime, or the enable signal
+              # is blocked by DD_DYNAMIC_INSTRUMENTATION_ENABLED=false. Advertising
+              # then would report DI as in use when it is not and invite probe
+              # configs the tracer must refuse; withdraw the products otherwise.
+              if components&.dynamic_instrumentation&.started?
+                components&.remote&.add_products(*di_products)
+              else
+                components&.remote&.remove_products(*di_products)
+              end
+            else
+              components&.symbol_database&.stop_for_di_disable
+              components&.remote&.remove_products(*di_products)
+            end
+          end
+
           content.applied
 
-          Datadog.send(:components).telemetry.client_configuration_change!(env_vars)
+          # allow_initialization: false because process_config runs on the
+          # remote-config worker thread. If components haven't been built yet
+          # (e.g. during a teardown/reset window), the default `true` would
+          # synchronously build the entire component tree from this thread.
+          # The &. chain matches the pattern used by DI::Remote.handle_rc_enablement
+          # in the same dispatch path.
+          Datadog.send(:components, allow_initialization: false)&.telemetry&.client_configuration_change!(env_vars)
         rescue => e
-          content.errored("#{e.class.name} #{e.message}: #{Array(e.backtrace).join("\n")}")
+          content.errored("#{e.class}: #{e.message}: #{Array(e.backtrace).join("\n")}")
         end
 
         def receivers(_telemetry)
@@ -57,7 +93,7 @@ module Datadog
               case content.path.product
               when PRODUCT
                 config = parse_content(content)
-                process_config(config, content)
+                process_config(config, content, repository)
               end
             end
           end
@@ -71,13 +107,7 @@ module Datadog
         private
 
         def parse_content(content)
-          data = content.data.read
-
-          content.data.rewind
-
-          raise ReadError, 'EOF reached' if data.nil?
-
-          JSON.parse(data)
+          JSON.parse(content.data)
         end
       end
     end

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

@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative 'sampler'
+require_relative "sampler"
 
 module Datadog
   module Tracing

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

@@ -37,22 +37,22 @@ module Datadog
         # this tag is currently formatted as `"-sampling_mechanism"`.
         module Decision
           # Used before the tracer receives any rates from agent and there are no rules configured.
-          DEFAULT = '-0'
+          DEFAULT = "-0"
           # The sampling rate received in the agent's http response.
-          AGENT_RATE = '-1'
+          AGENT_RATE = "-1"
           # Locally configured rule.
-          TRACE_SAMPLING_RULE = '-3'
+          TRACE_SAMPLING_RULE = "-3"
           # User directly sets sampling priority via {Tracing.reject!} or {Tracing.keep!},
           # or by a custom sampler implementation.
-          MANUAL = '-4'
+          MANUAL = "-4"
           # Formerly AppSec.
-          ASM = '-5'
-          # Single Span Sampled.
-          SPAN_SAMPLING_RATE = '-8'
+          ASM = "-5"
+          # AI Guard.
+          AI_GUARD = "-13"
           # Dynamically configured rule, explicitly created by the user.
-          REMOTE_USER_RULE = '-11'
+          REMOTE_USER_RULE = "-11"
           # Dynamically configured  rule, automatically generated by Datadog.
-          REMOTE_DYNAMIC_RULE = '-12'
+          REMOTE_DYNAMIC_RULE = "-12"
         end
       end
     end

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

@@ -7,7 +7,7 @@ module Datadog
       # @abstract
       class Matcher
         # Pattern that matches any string
-        MATCH_ALL_PATTERN = '*'
+        MATCH_ALL_PATTERN = "*"
 
         # Returns `true` if the trace should conforms to this rule, `false` otherwise
         #
@@ -35,8 +35,8 @@ module Datadog
 
           # Our valid special characters, `?` and `*`, were just escaped
           # by `Regexp.quote` above. We need to unescape them:
-          glob.gsub!('\?', '.') # Any single character
-          glob.gsub!('\*', '.*') # Any substring
+          glob.gsub!('\?', ".") # Any single character
+          glob.gsub!('\*', ".*") # Any substring
 
           # Patterns have to match the whole input string
           glob = "\\A#{glob}\\z"
@@ -108,7 +108,7 @@ module Datadog
             # 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').
-            tag = format('%g', tag) if tag.is_a?(Numeric)
+            tag = format("%g", tag) if tag.is_a?(Numeric)
 
             matcher.match?(tag)
           end