datadog 2.8.0 → 2.39.0

data/lib/datadog/di/configuration/settings.rb

@@ -1,188 +0,0 @@
-# frozen_string_literal: true
-
-module Datadog
-  module DI
-    module Configuration
-      # Settings
-      module Settings
-        def self.extended(base)
-          base = base.singleton_class unless base.is_a?(Class)
-          add_settings!(base)
-        end
-
-        def self.add_settings!(base)
-          base.class_eval do
-            settings :dynamic_instrumentation do
-              option :enabled do |o|
-                o.type :bool
-                # The environment variable has an "internal" prefix so that
-                # any customers that have the "proper" environment variable
-                # turned on (i.e. DD_DYNAMIC_INSTRUMENTATION_ENABLED)
-                # do not enable Ruby DI until the latter is ready for
-                # customer testing.
-                o.env "DD_DYNAMIC_INSTRUMENTATION_ENABLED"
-                o.default false
-              end
-
-              # An array of variable and key names to redact in addition to
-              # the built-in list of identifiers.
-              #
-              # The names will be normalized by removing the following
-              # symbols: _, -, @, $, and then matched to the complete
-              # variable or key name while ignoring the case.
-              # For example, specifying pass_word will match password and
-              # PASSWORD, and specifying PASSWORD will match pass_word.
-              # Note that, while the at sign (@) is used in Ruby to refer
-              # to instance variables, it does not have any significance
-              # for this setting (and is removed before matching identifiers).
-              option :redacted_identifiers do |o|
-                o.env "DD_DYNAMIC_INSTRUMENTATION_REDACTED_IDENTIFIERS"
-                o.env_parser do |value|
-                  value&.split(",")&.map(&:strip)
-                end
-
-                o.type :array
-                o.default []
-              end
-
-              # An array of class names, values of which will be redacted from
-              # dynamic instrumentation snapshots. Example: FooClass.
-              # If a name is suffixed by '*', it becomes a wildcard and
-              # instances of any class whose name begins with the specified
-              # prefix will be redacted (example: Foo*).
-              #
-              # The names must all be fully-qualified, if any prefix of a
-              # class name is configured to be redacted, the value will be
-              # subject to redaction. For example, if Foo* is in the
-              # redacted class name list, instances of Foo, FooBar,
-              # Foo::Bar are all subject to redaction, but Bar::Foo will
-              # not be subject to redaction.
-              #
-              # Leading double-colon is permitted but has no effect,
-              # because the names are always considered to be fully-qualified.
-              # For example, adding ::Foo to the list will redact instances
-              # of Foo.
-              #
-              # Trailing colons should not be used because they will trigger
-              # exact match behavior but Ruby class names do not have
-              # trailing colons. For example, Foo:: will not cause anything
-              # to be redacted. Use Foo::* to redact all classes under
-              # the Foo module.
-              option :redacted_type_names do |o|
-                o.env "DD_DYNAMIC_INSTRUMENTATION_REDACTED_TYPES"
-                o.env_parser do |value|
-                  value&.split(",")&.map(&:strip)
-                end
-
-                o.type :array
-                o.default []
-              end
-
-              # Maximum number of object or collection traversals that
-              # will be permitted when serializing captured values.
-              option :max_capture_depth do |o|
-                o.type :int
-                o.default 3
-              end
-
-              # Maximum number of collection (Array and Hash) elements
-              # that will be captured. Arrays and hashes that have more
-              # elements will be truncated to this many elements.
-              option :max_capture_collection_size do |o|
-                o.type :int
-                o.default 100
-              end
-
-              # Strings longer than this length will be truncated to this
-              # length in dynamic instrumentation snapshots.
-              #
-              # Note that while all values are stringified during
-              # serialization, only values which are originally instances
-              # of the String class are subject to this length limit.
-              option :max_capture_string_length do |o|
-                o.type :int
-                o.default 255
-              end
-
-              # Maximim number of attributes that will be captured for
-              # a single non-primitive value.
-              option :max_capture_attribute_count do |o|
-                o.type :int
-                o.default 20
-              end
-
-              # Settings in the 'internal' group are for internal Datadog
-              # use only, and are needed to test dynamic instrumentation or
-              # experiment with features not released to customers.
-              settings :internal do
-                # This option instructs dynamic instrumentation to use
-                # untargeted trace points when installing line probes and
-                # code tracking is not active.
-                # WARNING: untargeted trace points carry a massive performance
-                # penalty for the entire file in which a line probe is placed.
-                #
-                # If this option is set to false, which is the default,
-                # dynamic instrumentation will add probes that reference
-                # unknown files to the list of pending probes, and when
-                # the respective files are loaded, the line probes will be
-                # installed using targeted trace points. If the file in
-                # question is already loaded when the probe is received
-                # (for example, it is in a third-party library loaded during
-                # application boot), and code tracking was not active when
-                # the file was loaded, such files will not be instrumentable
-                # via line probes.
-                #
-                # If this option is set to true, dynamic instrumentation will
-                # install untargeted trace points for all line probes,
-                # regardless of whether the referenced file is loaded.
-                # This permits instrumenting code which was loaded prior to
-                # code tracking being activated and instrumenting lines when
-                # code tracking is not activated at all. However, untargeted
-                # trace points are extremely slow and will greatly degrade
-                # performance of *all* code executed while they are installed,
-                # not just the instrumentation target.
-                option :untargeted_trace_points do |o|
-                  o.type :bool
-                  o.default false
-                end
-
-                # If true, all of the catch-all rescue blocks in DI
-                # will propagate the exceptions onward.
-                # WARNING: for internal Datadog use only - this will break
-                # the DI product and potentially the library in general in
-                # a multitude of ways, cause resource leakage, permanent
-                # performance decreases, etc.
-                option :propagate_all_exceptions do |o|
-                  o.type :bool
-                  o.default false
-                end
-
-                # Minimum interval, in seconds, between probe status and
-                # snapshot submissions to the agent. Probe notifier worker will
-                # batch together payloads submitted during each interval.
-                # A longer interval reduces the overhead imposed by dynamic
-                # instrumentation on the application, but also increases the
-                # time when application code cannot run (when the batches are
-                # being sent out by the probe notifier worker) and creates a
-                # possibility of dropping payloads if the queue gets too long.
-                option :min_send_interval do |o|
-                  o.type :int
-                  o.default 3
-                end
-
-                # Enable dynamic instrumentation in development environments.
-                # Currently DI does not fully implement support for code
-                # unloading and reloading, and is not supported in
-                # non-production environments.
-                option :development do |o|
-                  o.type :bool
-                  o.default false
-                end
-              end
-            end
-          end
-        end
-      end
-    end
-  end
-end

data/lib/datadog/di/transport.rb

@@ -1,81 +0,0 @@
-# frozen_string_literal: true
-
-require 'ostruct'
-require_relative 'error'
-
-module Datadog
-  module DI
-    # Transport for sending probe statuses and snapshots to local agent.
-    #
-    # Handles encoding of the payloads into multipart posts if necessary,
-    # body formatting/encoding, setting correct headers, etc.
-    #
-    # The transport does not handle batching of statuses or snapshots -
-    # the batching should be implemented upstream of this class.
-    #
-    # Timeout settings are forwarded from agent settings to the Net adapter.
-    #
-    # The send_* methods raise Error::AgentCommunicationError on errors
-    # (network errors and HTTP protocol errors). It is the responsibility
-    # of upstream code to rescue these exceptions appropriately to prevent them
-    # from being propagated to the application.
-    #
-    # @api private
-    class Transport
-      DIAGNOSTICS_PATH = '/debugger/v1/diagnostics'
-      INPUT_PATH = '/debugger/v1/input'
-
-      def initialize(agent_settings)
-        # Note that this uses host, port, timeout and TLS flag from
-        # agent settings.
-        @client = Core::Transport::HTTP::Adapters::Net.new(agent_settings)
-      end
-
-      def send_diagnostics(payload)
-        event_payload = Core::Vendor::Multipart::Post::UploadIO.new(
-          StringIO.new(JSON.dump(payload)), 'application/json', 'event.json'
-        )
-        payload = {'event' => event_payload}
-        # Core transport unconditionally specifies headers to underlying
-        # Net::HTTP client, ends up passing 'nil' as headers if none are
-        # specified by us, which then causes Net::HTTP to die with an exception.
-        send_request('Probe status submission',
-          path: DIAGNOSTICS_PATH, form: payload, headers: {})
-      end
-
-      def send_input(payload)
-        send_request('Probe snapshot submission',
-          path: INPUT_PATH, body: payload.to_json,
-          headers: {'content-type' => 'application/json'},)
-      end
-
-      # TODO status should use either input or diagnostics endpoints
-      # depending on agent version.
-      alias_method :send_status, :send_diagnostics
-
-      alias_method :send_snapshot, :send_input
-
-      private
-
-      attr_reader :client
-
-      def send_request(desc, **options)
-        # steep:ignore:start
-        env = OpenStruct.new(**options)
-        # steep:ignore:end
-        response = client.post(env)
-        unless response.ok?
-          raise Error::AgentCommunicationError, "#{desc} failed: #{response.code}: #{response.payload}"
-        end
-      # Datadog::Core::Transport does not perform any exception mapping,
-      # therefore we could have any exception here from failure to parse
-      # agent URI for example.
-      # If we ever implement retries for network errors, we should distinguish
-      # actual network errors from non-network errors that are raised by
-      # transport code.
-      rescue => exc
-        raise Error::AgentCommunicationError, "#{desc} failed: #{exc.class}: #{exc}"
-      end
-    end
-  end
-end

data/lib/datadog/tracing/contrib/rails/utils.rb

@@ -1,26 +0,0 @@
-# frozen_string_literal: true
-
-require_relative '../analytics'
-
-module Datadog
-  module Tracing
-    module Contrib
-      module Rails
-        # common utilities for Rails
-        module Utils
-          def self.app_name
-            if ::Rails::VERSION::MAJOR >= 6
-              ::Rails.application.class.module_parent_name.underscore
-            else
-              ::Rails.application.class.parent_name.underscore
-            end
-          end
-
-          def self.railtie_supported?
-            !!defined?(::Rails::Railtie)
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,19 +0,0 @@
-# frozen_string_literal: true
-
-module Datadog
-  module Tracing
-    module Transport
-      module HTTP
-        module API
-          # Specification for an HTTP API
-          # Defines behaviors without specific configuration details.
-          class Spec
-            def initialize
-              yield(self) if block_given?
-            end
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,43 +0,0 @@
-# frozen_string_literal: true
-
-require_relative '../../../core/encoding'
-
-require_relative '../../../core/transport/http/api/map'
-require_relative 'api/spec'
-
-require_relative 'traces'
-
-module Datadog
-  module Tracing
-    module Transport
-      module HTTP
-        # Namespace for API components
-        module API
-          # Default API versions
-          V4 = 'v0.4'
-          V3 = 'v0.3'
-
-          module_function
-
-          def defaults
-            Datadog::Core::Transport::HTTP::API::Map[
-              V4 => Spec.new do |s|
-                s.traces = Traces::API::Endpoint.new(
-                  '/v0.4/traces',
-                  Core::Encoding::MsgpackEncoder,
-                  service_rates: true
-                )
-              end,
-              V3 => Spec.new do |s|
-                s.traces = Traces::API::Endpoint.new(
-                  '/v0.3/traces',
-                  Core::Encoding::MsgpackEncoder
-                )
-              end,
-            ].with_fallbacks(V4 => V3)
-          end
-        end
-      end
-    end
-  end
-end

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

@@ -1,196 +0,0 @@
-# frozen_string_literal: true
-
-require_relative '../../core/worker'
-require_relative '../../core/workers/async'
-require_relative '../../core/workers/polling'
-require_relative '../../core/workers/queue'
-
-require_relative '../buffer'
-require_relative '../pipeline'
-require_relative '../event'
-
-require_relative '../transport/http'
-
-module Datadog
-  module Tracing
-    module Workers
-      # Writes traces to transport synchronously
-      class TraceWriter < Core::Worker
-        attr_reader \
-          :transport
-
-        # rubocop:disable Lint/MissingSuper
-        def initialize(options = {})
-          transport_options = options.fetch(:transport_options, {})
-
-          transport_options[:agent_settings] = options[:agent_settings] if options.key?(:agent_settings)
-
-          @transport = options.fetch(:transport) do
-            Datadog::Tracing::Transport::HTTP.default(**transport_options)
-          end
-        end
-        # rubocop:enable Lint/MissingSuper
-
-        def perform(traces)
-          write_traces(traces)
-        end
-
-        def write(trace)
-          write_traces([trace])
-        end
-
-        def write_traces(traces)
-          traces = process_traces(traces)
-          flush_traces(traces)
-        rescue StandardError => e
-          Datadog.logger.warn(
-            "Error while writing traces: dropped #{traces.length} items. Cause: #{e} Location: #{Array(e.backtrace).first}"
-          )
-        end
-
-        def process_traces(traces)
-          # Run traces through the processing pipeline
-          Pipeline.process!(traces)
-        end
-
-        def flush_traces(traces)
-          transport.send_traces(traces).tap do |response|
-            flush_completed.publish(response)
-          end
-        end
-
-        # TODO: Register `Datadog::Tracing::Diagnostics::EnvironmentLogger.collect_and_log!`
-        # TODO: as a flush_completed subscriber when the `TraceWriter`
-        # TODO: instantiation code is implemented.
-        def flush_completed
-          @flush_completed ||= FlushCompleted.new
-        end
-
-        # Flush completed event for worker
-        class FlushCompleted < Event
-          def initialize
-            super(:flush_completed)
-          end
-        end
-      end
-
-      # Writes traces to transport asynchronously,
-      # using a thread & buffer.
-      class AsyncTraceWriter < TraceWriter
-        include Core::Workers::Queue
-        include Core::Workers::Polling
-
-        DEFAULT_BUFFER_MAX_SIZE = 1000
-        FORK_POLICY_ASYNC = :async
-        FORK_POLICY_SYNC = :sync
-
-        attr_writer \
-          :async
-
-        def initialize(options = {})
-          # Workers::TraceWriter settings
-          super
-
-          # Workers::Polling settings
-          self.enabled = options.fetch(:enabled, true)
-
-          # Workers::Async::Thread settings
-          @async = true
-          self.fork_policy = options.fetch(:fork_policy, FORK_POLICY_ASYNC)
-
-          # Workers::IntervalLoop settings
-          self.loop_base_interval = options[:interval] if options.key?(:interval)
-          self.loop_back_off_ratio = options[:back_off_ratio] if options.key?(:back_off_ratio)
-          self.loop_back_off_max = options[:back_off_max] if options.key?(:back_off_max)
-
-          # Workers::Queue settings
-          @buffer_size = options.fetch(:buffer_size, DEFAULT_BUFFER_MAX_SIZE)
-          self.buffer = TraceBuffer.new(@buffer_size)
-
-          @shutdown_timeout = options.fetch(:shutdown_timeout, Core::Workers::Polling::DEFAULT_SHUTDOWN_TIMEOUT)
-        end
-
-        # NOTE: #perform is wrapped by other modules:
-        #       Polling --> Async --> IntervalLoop --> AsyncTraceWriter --> TraceWriter
-        #
-        # WARNING: This method breaks the Liskov Substitution Principle -- TraceWriter#perform is spec'd to return the
-        # result from the writer, whereas this method always returns nil.
-        def perform(traces)
-          super(traces).tap do |responses|
-            loop_back_off! if responses.find(&:server_error?)
-          end
-
-          nil
-        end
-
-        def stop(force_stop = false, timeout = @shutdown_timeout)
-          buffer.close if running?
-          super
-        end
-
-        def enqueue(trace)
-          buffer.push(trace)
-        end
-
-        def dequeue
-          # Wrap results in Array because they are
-          # splatted as args against TraceWriter#perform.
-          [buffer.pop]
-        end
-
-        # Are there more traces to be processed next?
-        def work_pending?
-          !buffer.empty?
-        end
-
-        def async?
-          @async == true
-        end
-
-        def fork_policy=(policy)
-          # Translate to Workers::Async::Thread policy
-          thread_fork_policy =  case policy
-                                when Core::Workers::Async::Thread::FORK_POLICY_STOP
-                                  policy
-                                when FORK_POLICY_SYNC
-                                  # Stop the async thread because the writer
-                                  # will bypass and run synchronously.
-                                  Core::Workers::Async::Thread::FORK_POLICY_STOP
-                                else
-                                  Core::Workers::Async::Thread::FORK_POLICY_RESTART
-                                end
-
-          # Update thread fork policy
-          super(thread_fork_policy)
-
-          # Update local policy
-          @writer_fork_policy = policy
-        end
-
-        def after_fork
-          # In multiprocess environments, forks will share the same buffer until its written to.
-          # A.K.A. copy-on-write. We don't want forks to write traces generated from another process.
-          # Instead, we reset it after the fork. (Make sure any enqueue operations happen after this.)
-          self.buffer = TraceBuffer.new(@buffer_size)
-
-          # Switch to synchronous mode if configured to do so.
-          # In some cases synchronous writing is preferred because the fork will be short lived.
-          @async = false if @writer_fork_policy == FORK_POLICY_SYNC
-        end
-
-        # WARNING: This method breaks the Liskov Substitution Principle -- TraceWriter#write is spec'd to return the
-        # result from the writer, whereas this method returns something else when running in async mode.
-        def write(trace)
-          # Start worker thread. If the process has forked, it will trigger #after_fork to
-          # reconfigure the worker accordingly.
-          # NOTE: It's important we do this before queuing or it will drop the current trace,
-          #       because #after_fork resets the buffer.
-          perform
-
-          # Queue the trace if running asynchronously, otherwise short-circuit and write it directly.
-          async? ? enqueue(trace) : write_traces([trace])
-        end
-      end
-    end
-  end
-end