ddtrace 0.35.2 → 1.1.0

data/lib/datadog/profiling/stack_recorder.rb

@@ -0,0 +1,33 @@
+# typed: false
+
+module Datadog
+  module Profiling
+    # Used to wrap a ddprof_ffi_Profile in a Ruby object and expose Ruby-level serialization APIs
+    # Methods prefixed with _native_ are implemented in `stack_recorder.c`
+    class StackRecorder
+      def serialize
+        status, result = self.class._native_serialize(self)
+
+        if status == :ok
+          start, finish, encoded_pprof = result
+
+          Datadog.logger.debug { "Encoded profile covering #{start.iso8601} to #{finish.iso8601}" }
+
+          [start, finish, encoded_pprof]
+        else
+          error_message = result
+
+          Datadog.logger.error("Failed to serialize profiling data: #{error_message}")
+
+          nil
+        end
+      end
+
+      # Used only for Ruby 2.2 and below which don't have the native `rb_time_timespec_new` API
+      # Called from native code
+      def self.ruby_time_from(timespec_seconds, timespec_nanoseconds)
+        Time.at(0).utc + timespec_seconds + (timespec_nanoseconds.to_r / 1_000_000_000)
+      end
+    end
+  end
+end

data/lib/datadog/profiling/tag_builder.rb

@@ -0,0 +1,48 @@
+# typed: true
+
+module Datadog
+  module Profiling
+    # Builds a hash of default plus user tags to be included in a profile
+    module TagBuilder
+      include Datadog::Profiling::Ext::Transport::HTTP # Tag name constants
+
+      def self.call(
+        settings:,
+        # Unified service tagging
+        env: settings.env,
+        service: settings.service,
+        version: settings.version,
+        # Other metadata
+        host: Core::Environment::Socket.hostname,
+        language: Core::Environment::Identity.lang,
+        pid: Process.pid.to_s,
+        profiler_version: Core::Environment::Identity.tracer_version,
+        runtime_engine: Core::Environment::Identity.lang_engine,
+        runtime_id: Core::Environment::Identity.id,
+        runtime_platform: Core::Environment::Identity.lang_platform,
+        runtime_version: Core::Environment::Identity.lang_version,
+        # User-provided tags
+        user_tags: settings.tags
+      )
+        tags = {
+          # When changing or adding these, make sure they are kept in sync with
+          # https://docs.google.com/spreadsheets/d/1LOGMf4c4Avbtn36uZ2SWvhIGKRPLM1BoWkUP4JYj7hA/ (Datadog internal link)
+          FORM_FIELD_TAG_HOST => host,
+          FORM_FIELD_TAG_LANGUAGE => language,
+          FORM_FIELD_TAG_PID => pid,
+          FORM_FIELD_TAG_PROFILER_VERSION => profiler_version,
+          FORM_FIELD_TAG_RUNTIME => language, # This is known to be repeated from language, above
+          FORM_FIELD_TAG_RUNTIME_ENGINE => runtime_engine,
+          FORM_FIELD_TAG_RUNTIME_ID => runtime_id,
+          FORM_FIELD_TAG_RUNTIME_PLATFORM => runtime_platform,
+          FORM_FIELD_TAG_RUNTIME_VERSION => runtime_version,
+        }
+        tags[FORM_FIELD_TAG_ENV] = env if env
+        tags[FORM_FIELD_TAG_SERVICE] = service if service
+        tags[FORM_FIELD_TAG_VERSION] = version if version
+
+        user_tags.merge(tags)
+      end
+    end
+  end
+end

data/lib/datadog/profiling/tasks/exec.rb

@@ -0,0 +1,50 @@
+# typed: true
+
+module Datadog
+  module Profiling
+    module Tasks
+      # Wraps command with Datadog tracing
+      class Exec
+        attr_reader :args
+
+        def initialize(args)
+          @args = args
+        end
+
+        def run
+          set_rubyopt!
+          exec_with_error_handling(args)
+        end
+
+        def rubyopts
+          [
+            '-rdatadog/profiling/preload'
+          ]
+        end
+
+        private
+
+        def set_rubyopt!
+          existing_rubyopt = ENV['RUBYOPT']
+
+          ENV['RUBYOPT'] = existing_rubyopt ? "#{existing_rubyopt} #{rubyopts.join(' ')}" : rubyopts.join(' ')
+        end
+
+        # If there's an error here, rather than throwing a cryptic stack trace, let's instead have clearer messages, and
+        # follow the same status codes as the shell uses
+        # See also:
+        # * https://www.gnu.org/software/bash/manual/html_node/Exit-Status.html
+        # * https://github.com/rubygems/rubygems/blob/dd93966cac224532035deda533cba2685dfa30cc/bundler/lib/bundler/cli/exec.rb#L45
+        def exec_with_error_handling(args)
+          Kernel.exec(*args)
+        rescue Errno::ENOENT => e
+          Kernel.warn "ddtracerb exec failed: #{e.class.name} #{e.message} (command was '#{args.join(' ')}')"
+          Kernel.exit 127
+        rescue Errno::EACCES, Errno::ENOEXEC => e
+          Kernel.warn "ddtracerb exec failed: #{e.class.name} #{e.message} (command was '#{args.join(' ')}')"
+          Kernel.exit 126
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/tasks/help.rb

@@ -0,0 +1,18 @@
+# typed: true
+
+module Datadog
+  module Profiling
+    module Tasks
+      # Prints help message for usage of `ddtrace`
+      class Help
+        def run
+          puts %(
+  Usage: ddtracerb [command] [arguments]
+    exec [command]: Executes command with tracing & profiling preloaded.
+    help:           Prints this help message.
+          )
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/tasks/setup.rb

@@ -0,0 +1,93 @@
+# typed: false
+
+require 'datadog/core/utils/only_once'
+require 'datadog/profiling/ext/forking'
+
+module Datadog
+  module Profiling
+    module Tasks
+      # Takes care of loading our extensions/monkey patches to handle fork() and validating if CPU-time profiling is usable
+      class Setup
+        ACTIVATE_EXTENSIONS_ONLY_ONCE = Core::Utils::OnlyOnce.new
+
+        def run
+          ACTIVATE_EXTENSIONS_ONLY_ONCE.run do
+            begin
+              check_if_cpu_time_profiling_is_supported
+              activate_forking_extensions
+              setup_at_fork_hooks
+            rescue StandardError, ScriptError => e
+              Datadog.logger.warn do
+                "Profiler extensions unavailable. Cause: #{e.class.name} #{e.message} " \
+                "Location: #{Array(e.backtrace).first}"
+              end
+            end
+          end
+        end
+
+        private
+
+        def activate_forking_extensions
+          if Ext::Forking.supported?
+            Ext::Forking.apply!
+          elsif Datadog.configuration.profiling.enabled
+            Datadog.logger.debug('Profiler forking extensions skipped; forking not supported.')
+          end
+        rescue StandardError, ScriptError => e
+          Datadog.logger.warn do
+            "Profiler forking extensions unavailable. Cause: #{e.class.name} #{e.message} " \
+            "Location: #{Array(e.backtrace).first}"
+          end
+        end
+
+        def check_if_cpu_time_profiling_is_supported
+          unsupported = cpu_time_profiling_unsupported_reason
+
+          if unsupported
+            Datadog.logger.info do
+              'CPU time profiling skipped because native CPU time is not supported: ' \
+              "#{unsupported}. Profiles containing 'Wall time' data will still be reported."
+            end
+          end
+        end
+
+        def setup_at_fork_hooks
+          if Process.respond_to?(:at_fork)
+            Process.at_fork(:child) do
+              begin
+                # When Ruby forks, clock IDs for each of the threads
+                # will change. We can only update these IDs from the
+                # execution context of the thread that owns it.
+                # This hook will update the IDs for the main thread
+                # after a fork occurs.
+                Thread.current.send(:update_native_ids) if Thread.current.respond_to?(:update_native_ids, true)
+
+                # Restart profiler, if enabled
+                Profiling.start_if_enabled
+              rescue StandardError => e
+                Datadog.logger.warn do
+                  "Error during post-fork hooks. Cause: #{e.class.name} #{e.message} " \
+                  "Location: #{Array(e.backtrace).first}"
+                end
+              end
+            end
+          end
+        end
+
+        def cpu_time_profiling_unsupported_reason
+          # NOTE: Only the first matching reason is returned, so try to keep a nice order on reasons
+
+          if RUBY_ENGINE == 'jruby'
+            'JRuby is not supported'
+          elsif RUBY_PLATFORM.include?('darwin')
+            'Feature requires Linux; macOS is not supported'
+          elsif RUBY_PLATFORM =~ /(mswin|mingw)/
+            'Feature requires Linux; Windows is not supported'
+          elsif !RUBY_PLATFORM.include?('linux')
+            "Feature requires Linux; #{RUBY_PLATFORM} is not supported"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/trace_identifiers/ddtrace.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+# typed: true
+
+require 'datadog/tracing'
+require 'datadog/tracing/metadata/ext'
+
+module Datadog
+  module Profiling
+    module TraceIdentifiers
+      # Used by Datadog::Profiling::TraceIdentifiers::Helper to get the trace identifiers (root span id and span id)
+      # for a given thread, if there is an active trace for that thread in the supplied tracer object.
+      class Ddtrace
+        def initialize(tracer:)
+          @tracer = (tracer if tracer.respond_to?(:active_trace))
+        end
+
+        def trace_identifiers_for(thread)
+          return unless @tracer
+
+          trace = @tracer.active_trace(thread)
+          return unless trace
+
+          root_span = trace.send(:root_span)
+          span = trace.active_span
+          return unless span && root_span
+
+          root_span_id = root_span.id || 0
+          span_id = span.id || 0
+
+          [root_span_id, span_id, maybe_extract_resource(trace, root_span)] if root_span_id != 0 && span_id != 0
+        end
+
+        private
+
+        # NOTE: Currently we're only interested in HTTP service endpoints. Over time, this list may be expanded.
+        # Resources MUST NOT include personal identifiable information (PII); this should not be the case with
+        # ddtrace integrations, but worth mentioning just in case :)
+        def maybe_extract_resource(trace, root_span)
+          trace.resource if root_span.span_type == Tracing::Metadata::Ext::HTTP::TYPE_INBOUND
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/trace_identifiers/helper.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+# typed: true
+
+require 'datadog/profiling/trace_identifiers/ddtrace'
+
+module Datadog
+  module Profiling
+    module TraceIdentifiers
+      # Helper used to retrieve the trace identifiers (root span id and span id) for a given thread,
+      # if there is an active trace for that thread for the supported tracing APIs.
+      #
+      # This data is used to connect profiles to the traces -- samples in a profile will be tagged with this data and
+      # the profile can be filtered down to look at only the samples for a given trace.
+      class Helper
+        DEFAULT_SUPPORTED_APIS = [
+          Profiling::TraceIdentifiers::Ddtrace
+        ].freeze
+        private_constant :DEFAULT_SUPPORTED_APIS
+
+        def initialize(
+          tracer:,
+          # If this is disabled, the helper will strip the optional trace_resource_container even if provided by the api
+          endpoint_collection_enabled:,
+          supported_apis: DEFAULT_SUPPORTED_APIS.map { |api| api.new(tracer: tracer) }
+        )
+          @endpoint_collection_enabled = endpoint_collection_enabled
+          @supported_apis = supported_apis
+        end
+
+        # Expected output of the #trace_identifiers_for
+        # duck type is [root_span_id, span_id, (optional trace_resource_container)]
+        def trace_identifiers_for(thread)
+          @supported_apis.each do |api|
+            trace_identifiers = api.trace_identifiers_for(thread)
+
+            if trace_identifiers
+              return @endpoint_collection_enabled ? trace_identifiers : trace_identifiers[0..1]
+            end
+          end
+
+          nil
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/transport/client.rb

@@ -0,0 +1,16 @@
+# typed: true
+
+module Datadog
+  module Profiling
+    module Transport
+      # Generic interface for profiling transports
+      module Client
+        include Kernel # Ensure that kernel methods are always available (https://sorbet.org/docs/error-reference#7003)
+
+        def send_profiling_flush(flush)
+          raise NotImplementedError
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/transport/http/api/endpoint.rb

@@ -0,0 +1,107 @@
+# typed: true
+
+require 'datadog/core/utils/compression'
+require 'datadog/core/vendor/multipart-post/multipart/post/composite_read_io'
+require 'datadog/profiling/ext'
+require 'datadog/profiling/transport/http/response'
+require 'ddtrace/transport/http/api/endpoint'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        module API
+          # Datadog API endpoint for profiling
+          class Endpoint < Datadog::Transport::HTTP::API::Endpoint
+            include Profiling::Ext::Transport::HTTP
+
+            # These tags are read from the flush object (see below) directly and so we ignore any extra copies that
+            # may come in the tags hash to avoid duplicates.
+            TAGS_TO_IGNORE_IN_TAGS_HASH = %w[service env version].freeze
+            private_constant :TAGS_TO_IGNORE_IN_TAGS_HASH
+
+            attr_reader \
+              :encoder
+
+            def initialize(path, encoder)
+              super(:post, path)
+              @encoder = encoder
+            end
+
+            def call(env, &block)
+              # Build request
+              env.form = build_form(env)
+
+              # Send request
+              http_response = super(env, &block)
+
+              # Build response
+              Profiling::Transport::HTTP::Response.new(http_response)
+            end
+
+            def build_form(env)
+              flush = env.request.parcel.data
+              pprof_file = build_pprof(flush)
+
+              form = {
+                FORM_FIELD_INTAKE_VERSION => '3', # Aka 1.3 intake format
+                FORM_FIELD_RECORDING_START => flush.start.utc.iso8601,
+                FORM_FIELD_RECORDING_END => flush.finish.utc.iso8601,
+                FORM_FIELD_TAGS => [
+                  "#{FORM_FIELD_TAG_RUNTIME}:#{flush.language}",
+                  "#{FORM_FIELD_TAG_RUNTIME_ID}:#{flush.runtime_id}",
+                  "#{FORM_FIELD_TAG_RUNTIME_ENGINE}:#{flush.runtime_engine}",
+                  "#{FORM_FIELD_TAG_RUNTIME_PLATFORM}:#{flush.runtime_platform}",
+                  "#{FORM_FIELD_TAG_RUNTIME_VERSION}:#{flush.runtime_version}",
+                  "#{FORM_FIELD_TAG_PID}:#{Process.pid}",
+                  "#{FORM_FIELD_TAG_PROFILER_VERSION}:#{flush.profiler_version}",
+                  # NOTE: Redundant w/ 'runtime'; may want to remove this later.
+                  "#{FORM_FIELD_TAG_LANGUAGE}:#{flush.language}",
+                  "#{FORM_FIELD_TAG_HOST}:#{flush.host}",
+                  *flush
+                    .tags
+                    .reject { |tag_key| TAGS_TO_IGNORE_IN_TAGS_HASH.include?(tag_key) }
+                    .map { |tag_key, tag_value| "#{tag_key}:#{tag_value}" }
+                ],
+                FORM_FIELD_PPROF_DATA => pprof_file,
+                FORM_FIELD_FAMILY => flush.language,
+              }
+
+              # Optional fields
+              form[FORM_FIELD_TAGS] << "#{FORM_FIELD_TAG_SERVICE}:#{flush.service}" unless flush.service.nil?
+              form[FORM_FIELD_TAGS] << "#{FORM_FIELD_TAG_ENV}:#{flush.env}" unless flush.env.nil?
+              form[FORM_FIELD_TAGS] << "#{FORM_FIELD_TAG_VERSION}:#{flush.version}" unless flush.version.nil?
+
+              # May not be available/enabled
+              form[FORM_FIELD_CODE_PROVENANCE_DATA] = build_code_provenance(flush) if flush.code_provenance
+
+              form
+            end
+
+            def build_pprof(flush)
+              pprof = encoder.encode(flush)
+
+              gzipped_pprof_data = Core::Utils::Compression.gzip(pprof.data)
+
+              Core::Vendor::Multipart::Post::UploadIO.new(
+                StringIO.new(gzipped_pprof_data),
+                HEADER_CONTENT_TYPE_OCTET_STREAM,
+                PPROF_DEFAULT_FILENAME
+              )
+            end
+
+            def build_code_provenance(flush)
+              gzipped_code_provenance = Core::Utils::Compression.gzip(flush.code_provenance)
+
+              Core::Vendor::Multipart::Post::UploadIO.new(
+                StringIO.new(gzipped_code_provenance),
+                HEADER_CONTENT_TYPE_OCTET_STREAM,
+                CODE_PROVENANCE_FILENAME,
+              )
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/transport/http/api/instance.rb

@@ -0,0 +1,38 @@
+# typed: true
+
+require 'ddtrace/transport/http/api/instance'
+require 'datadog/profiling/transport/http/api/spec'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        module API
+          # API instance for profiling
+          class Instance < Datadog::Transport::HTTP::API::Instance
+            def send_profiling_flush(env)
+              raise ProfilesNotSupportedError, spec unless spec.is_a?(Spec)
+
+              spec.send_profiling_flush(env) do |request_env|
+                call(request_env)
+              end
+            end
+
+            # Raised when profiles sent to API that does not support profiles
+            class ProfilesNotSupportedError < StandardError
+              attr_reader :spec
+
+              def initialize(spec)
+                @spec = spec
+              end
+
+              def message
+                'Profiles not supported for this API!'
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,42 @@
+# typed: true
+
+require 'ddtrace/transport/http/api/spec'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        module API
+          # API specification for profiling
+          class Spec < Datadog::Transport::HTTP::API::Spec
+            attr_accessor \
+              :profiles
+
+            def send_profiling_flush(env, &block)
+              raise NoProfilesEndpointDefinedError, self if profiles.nil?
+
+              profiles.call(env, &block)
+            end
+
+            def encoder
+              profiles.encoder
+            end
+
+            # Raised when profiles sent but no profiles endpoint is defined
+            class NoProfilesEndpointDefinedError < StandardError
+              attr_reader :spec
+
+              def initialize(spec)
+                @spec = spec
+              end
+
+              def message
+                'No profiles endpoint is defined for API specification!'
+              end
+            end
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,45 @@
+# typed: true
+
+require 'ddtrace/transport/http/api/map'
+require 'datadog/profiling/encoding/profile'
+require 'datadog/profiling/transport/http/api/spec'
+require 'datadog/profiling/transport/http/api/instance'
+require 'datadog/profiling/transport/http/api/endpoint'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        # Extensions for HTTP API Spec
+        module API
+          # Default API versions
+          V1 = 'v1'.freeze
+
+          module_function
+
+          def agent_defaults
+            @agent_defaults ||= Datadog::Transport::HTTP::API::Map[
+              V1 => Spec.new do |s|
+                s.profiles = Endpoint.new(
+                  '/profiling/v1/input'.freeze,
+                  Profiling::Encoding::Profile::Protobuf
+                )
+              end
+            ]
+          end
+
+          def api_defaults
+            @api_defaults ||= Datadog::Transport::HTTP::API::Map[
+              V1 => Spec.new do |s|
+                s.profiles = Endpoint.new(
+                  '/v1/input'.freeze,
+                  Profiling::Encoding::Profile::Protobuf
+                )
+              end
+            ]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/transport/http/builder.rb

@@ -0,0 +1,30 @@
+# typed: true
+
+require 'ddtrace/transport/http/builder'
+
+require 'datadog/profiling/transport/http/api'
+require 'datadog/profiling/transport/http/client'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        # Builds new instances of Transport::HTTP::Client
+        class Builder < Datadog::Transport::HTTP::Builder
+          def api_instance_class
+            API::Instance
+          end
+
+          def to_transport
+            raise Datadog::Transport::HTTP::Builder::NoDefaultApiError if @default_api.nil?
+
+            # TODO: Profiling doesn't have multiple APIs yet.
+            #       When it does, we should build it out with these APIs.
+            #       Just use :default_api for now.
+            Client.new(to_api_instances[@default_api])
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,35 @@
+# typed: true
+
+require 'ddtrace/transport/http/client'
+require 'datadog/profiling/transport/client'
+
+module Datadog
+  module Profiling
+    module Transport
+      module HTTP
+        # Routes, encodes, and sends tracer data to the trace agent via HTTP.
+        class Client < Datadog::Transport::HTTP::Client
+          include Transport::Client
+
+          def send_profiling_flush(flush)
+            # Build a request
+            request = Profiling::Transport::Request.new(flush)
+            send_payload(request).tap do |response|
+              if response.ok?
+                Datadog.logger.debug('Successfully reported profiling data')
+              else
+                Datadog.logger.debug { "Failed to report profiling data -- #{response.inspect}" }
+              end
+            end
+          end
+
+          def send_payload(request)
+            send_request(request) do |api, env|
+              api.send_profiling_flush(env)
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/datadog/profiling/transport/http/response.rb

@@ -0,0 +1,23 @@
+# typed: true
+
+require 'ddtrace/transport/http/response'
+require 'datadog/profiling/transport/response'
+
+module Datadog
+  module Profiling
+    module Transport
+      # HTTP transport behavior for profiling
+      module HTTP
+        # Response from HTTP transport for profiling
+        class Response
+          include Datadog::Transport::HTTP::Response
+          include Profiling::Transport::Response
+
+          def initialize(http_response, options = {})
+            super(http_response)
+          end
+        end
+      end
+    end
+  end
+end