dox-jaeger-client 2.0.0

data/crossdock/server

@@ -0,0 +1,175 @@
+#!/usr/bin/env ruby
+
+$stdout.sync = true
+
+require 'sinatra/base'
+require 'webrick'
+require 'jaeger/client'
+require 'net/http'
+require 'uri'
+
+class HealthServer < Sinatra::Application
+  get '/' do
+    status 200
+  end
+end
+
+class HttpServer < Sinatra::Application
+  post '/start_trace' do
+    puts "Got request to start trace: #{trace_request}"
+
+    parent_context = tracer.extract(OpenTracing::FORMAT_RACK, request.env)
+    server_span = tracer.start_span('/start_trace', child_of: parent_context)
+
+    server_span.set_baggage_item('crossdock-baggage-key', trace_request['baggage'])
+    if trace_request.key?('sampled')
+      server_span.set_tag('sampling.priority', trace_request['sampled'] ? 1 : 0)
+    end
+
+    response = {
+      span: observe_span(server_span),
+      notImplementedError: ''
+    }
+
+    if trace_request['downstream']
+      downstream = trace_request['downstream']
+      transport = downstream['transport']
+
+      response[:downstream] =
+        case transport
+        when 'HTTP'
+          call_downstream_http(downstream, server_span)
+        when 'DUMMY'
+          { notImplementedError: 'Dummy has not been implemented' }
+        else
+          { notImplementedError: "Unrecognized transport received: #{transport}" }
+        end
+    end
+
+    puts "Response: #{response}"
+
+    server_span.finish
+    body JSON.dump(response)
+  end
+
+  post '/join_trace' do
+    puts 'Got request to join trace' \
+         "\n  Params: #{trace_request}" \
+         "\n  Headers: #{request_headers(request)}"
+
+    parent_context = tracer.extract(OpenTracing::FORMAT_RACK, request.env)
+    server_span = tracer.start_span('/join_trace', child_of: parent_context)
+
+    response = {
+      span: observe_span(server_span),
+      notImplementedError: ''
+    }
+
+    if trace_request['downstream']
+      downstream = trace_request['downstream']
+      transport = downstream['transport']
+
+      response[:downstream] =
+        case transport
+        when 'HTTP'
+          call_downstream_http(downstream, server_span)
+        when 'DUMMY'
+          { notImplementedError: 'Dummy has not been implemented' }
+        else
+          { notImplementedError: "Unrecognized transport received: #{transport}" }
+        end
+    end
+
+    puts "Response: #{response}"
+
+    server_span.finish
+    body JSON.dump(response)
+  end
+
+  post '/create_traces' do
+    puts "Got request to create traces: #{trace_request}"
+
+    trace_request['count'].times do
+      span = tracer.start_span(trace_request['operation'], tags: trace_request['tags'])
+      span.finish
+    end
+
+    status 200
+  end
+
+  private
+
+  def tracer
+    @tracer ||= Jaeger::Client.build(
+      service_name: 'crossdock-ruby',
+      host: 'jaeger-agent',
+      port: 6831,
+      flush_interval: 1,
+      sampler: Jaeger::Samplers::Const.new(true)
+    )
+  end
+
+  def trace_request
+    @trace_request ||= begin
+      request.body.rewind
+      JSON.parse(request.body.read)
+    end
+  end
+
+  def observe_span(span)
+    if span
+      {
+        traceId: span.context.to_trace_id,
+        sampled: span.context.sampled?,
+        baggage: span.get_baggage_item('crossdock-baggage-key')
+      }
+    else
+      {
+        traceId: 'no span found',
+        sampled: false,
+        baggage: 'no span found'
+      }
+    end
+  end
+
+  def call_downstream_http(downstream, server_span)
+    downstream_url = "http://#{downstream['host']}:#{downstream['port']}/join_trace"
+
+    client_span = tracer.start_span('client-span', child_of: server_span)
+
+    headers = { 'Content-Type' => 'application/json' }
+    tracer.inject(client_span.context, OpenTracing::FORMAT_RACK, headers)
+
+    response = Net::HTTP.post(
+      URI(downstream_url),
+      JSON.dump(
+        serverRole: downstream['serverRole'],
+        downstream: downstream['downstream']
+      ),
+      headers
+    )
+
+    client_span.finish
+
+    if response.is_a?(Net::HTTPSuccess)
+      JSON.parse(response.body)
+    else
+      { error: response.body }
+    end
+  end
+
+  def request_headers(request)
+    request.env.select do |key, _value|
+      key.start_with?('HTTP_')
+    end
+  end
+end
+
+threads = []
+threads << Thread.new do
+  Rack::Handler::WEBrick.run(HealthServer, Port: 8080, Host: '0.0.0.0')
+end
+threads << Thread.new do
+  Rack::Handler::WEBrick.run(HttpServer, Port: 8081, Host: '0.0.0.0')
+end
+threads.each(&:join)

data/jaeger-client.gemspec

@@ -0,0 +1,37 @@
+lib = File.expand_path('lib', __dir__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+
+require 'jaeger/client/version'
+
+Gem::Specification.new do |spec|
+  spec.name          = 'dox-jaeger-client'
+  spec.version       = Jaeger::Client::VERSION
+  spec.authors       = ['Jeremiah Hemphill']
+  spec.email         = ['jhemphill@doximity.com']
+  spec.metadata['rubygems_mfa_required'] = 'true'
+
+  spec.summary       = 'OpenTracing Tracer implementation for Jaeger in Ruby with ruby 3.2 support'
+  spec.description   = 'OpenTracing Tracer implementation for Jaeger in Ruby with ruby 3.2 support'
+  spec.homepage      = 'https://github.com/doximity/jaeger-client-ruby'
+  spec.license       = 'MIT'
+
+  spec.files         = `git ls-files -z`.split("\x0").reject do |f|
+    f.match(%r{^(test|spec|features)/})
+  end
+  spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
+  spec.require_paths = ['lib']
+
+  spec.required_ruby_version = ['>= 2.7', '< 3.3']
+
+  spec.add_development_dependency 'bundler'
+  spec.add_development_dependency 'rake', '~> 13.0'
+  spec.add_development_dependency 'rspec', '~> 3.10'
+  spec.add_development_dependency 'rubocop', '~> 1.25'
+  spec.add_development_dependency 'rubocop-rake', '~> 0.6'
+  spec.add_development_dependency 'rubocop-rspec', '~> 2.8'
+  spec.add_development_dependency 'timecop', '~> 0.9'
+  spec.add_development_dependency 'webmock', '~> 3.14'
+
+  spec.add_dependency 'opentracing', '~> 0.3'
+  spec.add_dependency 'thrift'
+end

data/lib/jaeger/client/version.rb

@@ -0,0 +1,7 @@
+# frozen_string_literal: true
+
+module Jaeger
+  module Client
+    VERSION = '2.0.0'
+  end
+end

data/lib/jaeger/client.rb

@@ -0,0 +1,77 @@
+# frozen_string_literal: true
+
+$LOAD_PATH.push("#{File.dirname(__FILE__)}/../../thrift/gen-rb")
+
+require 'opentracing'
+require 'jaeger/thrift/agent'
+require 'logger'
+require 'time'
+require 'net/http'
+require 'cgi'
+require 'json'
+
+require_relative 'tracer'
+require_relative 'span'
+require_relative 'span_context'
+require_relative 'scope'
+require_relative 'scope_manager'
+require_relative 'trace_id'
+require_relative 'udp_sender'
+require_relative 'http_sender'
+require_relative 'reporters'
+require_relative 'client/version'
+require_relative 'samplers'
+require_relative 'encoders/thrift_encoder'
+require_relative 'injectors'
+require_relative 'extractors'
+require_relative 'rate_limiter'
+require_relative 'thrift_tag_builder'
+require_relative 'recurring_executor'
+
+module Jaeger
+  module Client
+    # We initially had everything under Jaeger::Client namespace. This however
+    # was not very useful and was removed. These assignments are here for
+    # backwards compatibility. Fine to remove in the next major version.
+    UdpSender = Jaeger::UdpSender
+    HttpSender = Jaeger::HttpSender
+    Encoders = Jaeger::Encoders
+    Samplers = Jaeger::Samplers
+    Reporters = Jaeger::Reporters
+    Injectors = Jaeger::Injectors
+    Extractors = Jaeger::Extractors
+
+    DEFAULT_FLUSH_INTERVAL = 10
+
+    def self.build(service_name:,
+                   host: '127.0.0.1',
+                   port: 6831,
+                   flush_interval: DEFAULT_FLUSH_INTERVAL,
+                   sampler: Samplers::Const.new(true),
+                   logger: Logger.new($stdout),
+                   sender: nil,
+                   reporter: nil,
+                   injectors: {},
+                   extractors: {},
+                   tags: {})
+      encoder = Encoders::ThriftEncoder.new(service_name: service_name, tags: tags, logger: logger)
+
+      if sender
+        warn '[DEPRECATION] Passing `sender` directly to Jaeger::Client.build is deprecated.' \
+             'Please use `reporter` instead.'
+      end
+
+      reporter ||= Reporters::RemoteReporter.new(
+        sender: sender || UdpSender.new(host: host, port: port, encoder: encoder, logger: logger),
+        flush_interval: flush_interval
+      )
+
+      Tracer.new(
+        reporter: reporter,
+        sampler: sampler,
+        injectors: Injectors.prepare(injectors),
+        extractors: Extractors.prepare(extractors)
+      )
+    end
+  end
+end

data/lib/jaeger/encoders/thrift_encoder.rb

@@ -0,0 +1,173 @@
+# frozen_string_literal: true
+
+module Jaeger
+  module Encoders
+    class ThriftEncoder
+      def initialize(service_name:, logger:, tags: {})
+        @service_name = service_name
+        @logger = logger
+        @tags = prepare_tags(tags)
+        @process = Jaeger::Thrift::Process.new('serviceName' => @service_name, 'tags' => @tags)
+      end
+
+      def encode(spans)
+        encode_batch(spans.map(&method(:encode_span)))
+      end
+
+      def encode_limited_size(spans, protocol_class, max_message_length)
+        batches = []
+        current_batch = []
+        current_batch_size = 0
+
+        max_spans_length = calculate_max_spans_length(protocol_class, max_message_length)
+
+        spans.each do |span|
+          encoded_span = encode_span(span)
+          span_size = message_size(encoded_span, protocol_class)
+
+          if span_size > max_spans_length
+            @logger.warn("Skip span #{span.operation_name} with size #{span_size}")
+            next
+          end
+
+          if (current_batch_size + span_size) > max_spans_length
+            batches << encode_batch(current_batch)
+            current_batch = []
+            current_batch_size = 0
+          end
+
+          current_batch << encoded_span
+          current_batch_size += span_size
+        end
+        batches << encode_batch(current_batch) unless current_batch.empty?
+        batches
+      end
+
+      private
+
+      def encode_batch(encoded_spans)
+        Jaeger::Thrift::Batch.new('process' => @process, 'spans' => encoded_spans)
+      end
+
+      def encode_span(span)
+        context = span.context
+        start_ts, duration = build_timestamps(span)
+
+        Jaeger::Thrift::Span.new(
+          'traceIdLow' => TraceId.uint64_id_to_int64(trace_id_to_low(context.trace_id)),
+          'traceIdHigh' => TraceId.uint64_id_to_int64(trace_id_to_high(context.trace_id)),
+          'spanId' => TraceId.uint64_id_to_int64(context.span_id),
+          'parentSpanId' => TraceId.uint64_id_to_int64(context.parent_id),
+          'operationName' => span.operation_name,
+          'references' => build_references(span.references || []),
+          'flags' => context.flags,
+          'startTime' => start_ts,
+          'duration' => duration,
+          'tags' => span.tags,
+          'logs' => span.logs
+        )
+      end
+
+      def build_references(references)
+        references.map do |ref|
+          Jaeger::Thrift::SpanRef.new(
+            'refType' => span_ref_type(ref.type),
+            'traceIdLow' => TraceId.uint64_id_to_int64(trace_id_to_low(ref.context.trace_id)),
+            'traceIdHigh' => TraceId.uint64_id_to_int64(trace_id_to_high(ref.context.trace_id)),
+            'spanId' => TraceId.uint64_id_to_int64(ref.context.span_id)
+          )
+        end
+      end
+
+      def build_timestamps(span)
+        start_ts = (span.start_time.to_f * 1_000_000).to_i
+        end_ts = (span.end_time.to_f * 1_000_000).to_i
+        duration = end_ts - start_ts
+        [start_ts, duration]
+      end
+
+      def span_ref_type(type)
+        case type
+        when OpenTracing::Reference::CHILD_OF
+          Jaeger::Thrift::SpanRefType::CHILD_OF
+        when OpenTracing::Reference::FOLLOWS_FROM
+          Jaeger::Thrift::SpanRefType::FOLLOWS_FROM
+        else
+          warn "Jaeger::Client with format #{type} is not supported yet"
+          nil
+        end
+      end
+
+      def prepare_tags(tags)
+        with_default_tags = tags.dup
+        with_default_tags['jaeger.version'] = "Ruby-#{Jaeger::Client::VERSION}"
+        with_default_tags['hostname'] ||= Socket.gethostname
+
+        unless with_default_tags['ip']
+          ipv4 = Socket.ip_address_list.find { |ai| ai.ipv4? && !ai.ipv4_loopback? }
+          with_default_tags['ip'] = ipv4.ip_address unless ipv4.nil?
+        end
+
+        with_default_tags.map do |key, value|
+          ThriftTagBuilder.build(key, value)
+        end
+      end
+
+      # Returns the right most 64 bits of trace id
+      def trace_id_to_low(trace_id)
+        trace_id & TraceId::MAX_64BIT_UNSIGNED_INT
+      end
+
+      # Returns the left most 64 bits of trace id
+      def trace_id_to_high(trace_id)
+        (trace_id >> 64) & TraceId::MAX_64BIT_UNSIGNED_INT
+      end
+
+      class DummyTransport
+        attr_accessor :size
+
+        def initialize
+          @size = 0
+        end
+
+        def write(buf)
+          @size += buf.size
+        end
+
+        def flush
+          @size = 0
+        end
+
+        def close; end
+      end
+
+      def message_size(message, protocol_class)
+        transport = DummyTransport.new
+        protocol = protocol_class.new(transport)
+        message.write(protocol)
+        transport.size
+      end
+
+      # Compact protocol have dynamic size of list header
+      # https://erikvanoosten.github.io/thrift-missing-specification/#_list_and_set_2
+      BATCH_SPANS_SIZE_WINDOW = 4
+
+      def empty_batch_size_cache
+        @empty_batch_size_cache ||= {}
+      end
+
+      def caclulate_empty_batch_size(protocol_class)
+        empty_batch_size_cache[protocol_class] ||=
+          message_size(encode_batch([]), protocol_class) + BATCH_SPANS_SIZE_WINDOW
+      end
+
+      def calculate_max_spans_length(protocol_class, max_message_length)
+        empty_batch_size = caclulate_empty_batch_size(protocol_class)
+        max_spans_length = max_message_length - empty_batch_size
+        return max_spans_length if max_spans_length.positive?
+
+        raise "Batch header have size #{empty_batch_size}, but limit #{max_message_length}"
+      end
+    end
+  end
+end

data/lib/jaeger/extractors.rb

@@ -0,0 +1,173 @@
+# frozen_string_literal: true
+
+module Jaeger
+  module Extractors
+    class SerializedJaegerTrace
+      def self.parse(trace)
+        return nil if !trace || trace == ''
+
+        trace_arguments = trace.split(':')
+        return nil if trace_arguments.size != 4
+
+        trace_id = TraceId.base16_hex_id_to_uint128(trace_arguments[0])
+        span_id, parent_id, flags = trace_arguments[1..3].map(&TraceId.method(:base16_hex_id_to_uint64))
+
+        return nil if trace_id.zero? || span_id.zero?
+
+        SpanContext.new(
+          trace_id: trace_id,
+          parent_id: parent_id,
+          span_id: span_id,
+          flags: flags
+        )
+      end
+    end
+
+    class JaegerTextMapCodec
+      def self.extract(carrier)
+        context = SerializedJaegerTrace.parse(carrier['uber-trace-id'])
+        return nil unless context
+
+        carrier.each do |key, value|
+          baggage_match = key.match(/\Auberctx-([\w-]+)\Z/)
+          if baggage_match
+            context.set_baggage_item(baggage_match[1], value)
+          end
+        end
+
+        context
+      end
+    end
+
+    class JaegerRackCodec
+      def self.extract(carrier)
+        serialized_trace = carrier['HTTP_UBER_TRACE_ID']
+        serialized_trace = CGI.unescape(serialized_trace) if serialized_trace
+        context = SerializedJaegerTrace.parse(serialized_trace)
+        return nil unless context
+
+        carrier.each do |key, value|
+          baggage_match = key.match(/\AHTTP_UBERCTX_(\w+)\Z/)
+          if baggage_match
+            key = baggage_match[1].downcase.tr('_', '-')
+            context.set_baggage_item(key, CGI.unescape(value))
+          end
+        end
+
+        context
+      end
+    end
+
+    class JaegerBinaryCodec
+      def self.extract(_carrier)
+        warn 'Jaeger::Client with binary format is not supported yet'
+      end
+    end
+
+    class B3RackCodec
+      class Keys
+        TRACE_ID = 'HTTP_X_B3_TRACEID'
+        SPAN_ID = 'HTTP_X_B3_SPANID'
+        PARENT_SPAN_ID = 'HTTP_X_B3_PARENTSPANID'
+        FLAGS = 'HTTP_X_B3_FLAGS'
+        SAMPLED = 'HTTP_X_B3_SAMPLED'
+      end.freeze
+
+      def self.extract(carrier)
+        B3CodecCommon.extract(carrier, Keys)
+      end
+    end
+
+    class B3TextMapCodec
+      class Keys
+        TRACE_ID = 'x-b3-traceid'
+        SPAN_ID = 'x-b3-spanid'
+        PARENT_SPAN_ID = 'x-b3-parentspanid'
+        FLAGS = 'x-b3-flags'
+        SAMPLED = 'x-b3-sampled'
+      end.freeze
+
+      def self.extract(carrier)
+        B3CodecCommon.extract(carrier, Keys)
+      end
+    end
+
+    class B3CodecCommon
+      def self.extract(carrier, keys)
+        return nil if carrier[keys::TRACE_ID].nil? || carrier[keys::SPAN_ID].nil?
+
+        trace_id = if carrier[keys::TRACE_ID].length <= 16
+                     TraceId.base16_hex_id_to_uint64(carrier[keys::TRACE_ID])
+                   else
+                     TraceId.base16_hex_id_to_uint128(carrier[keys::TRACE_ID])
+                   end
+
+        span_id = TraceId.base16_hex_id_to_uint64(carrier[keys::SPAN_ID])
+        parent_id = TraceId.base16_hex_id_to_uint64(carrier[keys::PARENT_SPAN_ID])
+        flags = parse_flags(carrier[keys::FLAGS], carrier[keys::SAMPLED])
+
+        return nil if span_id.zero? || trace_id.zero?
+
+        SpanContext.new(
+          trace_id: trace_id,
+          parent_id: parent_id,
+          span_id: span_id,
+          flags: flags
+        )
+      end
+
+      # if the flags header is '1' then the sampled header should not be present
+      def self.parse_flags(flags_header, sampled_header)
+        if flags_header == '1'
+          Jaeger::SpanContext::Flags::DEBUG
+        else
+          TraceId.base16_hex_id_to_uint64(sampled_header)
+        end
+      end
+      private_class_method :parse_flags
+    end
+
+    class TraceContextRackCodec
+      # Internal regex used to identify the TraceContext version
+      VERSION_PATTERN = /^([0-9a-fA-F]{2})-(.+)$/.freeze
+
+      # Internal regex used to parse fields in version 0
+      HEADER_V0_PATTERN = /^([0-9a-fA-F]{32})-([0-9a-fA-F]{16})(-([0-9a-fA-F]{2}))?$/.freeze
+
+      def self.extract(carrier)
+        header_value = carrier['HTTP_TRACEPARENT']
+
+        version_match = VERSION_PATTERN.match(header_value)
+        return nil unless version_match
+
+        # We currently only support version 0
+        return nil if version_match[1].to_i(16) != 0
+
+        match = HEADER_V0_PATTERN.match(version_match[2])
+        return nil unless match
+
+        trace_id = TraceId.base16_hex_id_to_uint128(match[1])
+        span_id = TraceId.base16_hex_id_to_uint64(match[2])
+        flags = TraceId.base16_hex_id_to_uint64(match[4])
+        return nil if trace_id.zero? || span_id.zero?
+
+        SpanContext.new(trace_id: trace_id, span_id: span_id, flags: flags)
+      end
+    end
+
+    DEFAULT_EXTRACTORS = {
+      OpenTracing::FORMAT_TEXT_MAP => JaegerTextMapCodec,
+      OpenTracing::FORMAT_BINARY => JaegerBinaryCodec,
+      OpenTracing::FORMAT_RACK => JaegerRackCodec
+    }.freeze
+
+    def self.prepare(extractors)
+      DEFAULT_EXTRACTORS.reduce(extractors) do |acc, (format, default)|
+        provided_extractors = Array(extractors[format])
+        provided_extractors += [default] if provided_extractors.empty?
+
+        acc.merge(format => provided_extractors)
+      end
+    end
+  end
+end

data/lib/jaeger/http_sender.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+require 'logger'
+
+module Jaeger
+  class HttpSender
+    def initialize(url:, encoder:, headers: {}, logger: Logger.new($stdout))
+      @encoder = encoder
+      @logger = logger
+
+      @uri = URI(url)
+      @uri.query = 'format=jaeger.thrift'
+
+      @transport = ::Thrift::HTTPClientTransport.new(@uri.to_s)
+      @transport.add_headers(headers)
+
+      @serializer = ::Thrift::Serializer.new
+    end
+
+    def send_spans(spans)
+      batch = @encoder.encode(spans)
+      @transport.write(@serializer.serialize(batch))
+      @transport.flush
+    rescue StandardError => e
+      @logger.error("Failure while sending a batch of spans: #{e}")
+    end
+  end
+end