fluent-plugin-kafka-xst 0.19.1

data/examples/out_kafka2/protobuf-formatter.conf

@@ -0,0 +1,23 @@
+<source>
+    @type sample
+    sample {"hello": "world", "some_record":{"event":"message"}}
+    rate 7000
+    tag sample.hello.world
+</source>
+
+<match sample.**>
+    @type kafka2
+
+    brokers "broker:29092"
+
+    record_key "some_record"
+    default_topic "events"
+
+    <format>
+        # requires the fluent-plugin-formatter-protobuf gem
+        # see its docs for full usage
+        @type protobuf
+        class_name SomeRecord
+        include_paths ["/opt/fluent-plugin-formatter-protobuf/some_record_pb.rb"]
+    </format>
+</match>

data/examples/out_kafka2/record_key.conf

@@ -0,0 +1,31 @@
+<source>
+    @type sample
+    sample {"hello": "world", "some_record":{"event":"message"}}
+    rate 7000
+    tag sample.hello.world
+</source>
+
+<match sample.**>
+    @type kafka2
+
+    brokers "broker:29092"
+
+    # {"event": "message"} will be formatted and sent to Kafka
+    record_key "some_record"
+
+    default_topic "events"
+
+    <format>
+        @type json
+    </format>
+
+    <buffer>
+      flush_at_shutdown true
+      flush_mode interval
+      flush_interval 1s
+      chunk_limit_size 3MB
+      chunk_full_threshold 1
+      total_limit_size 1024MB
+      overflow_action block
+    </buffer>
+</match>

data/fluent-plugin-kafka.gemspec

@@ -0,0 +1,27 @@
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |gem|
+  gem.authors       = ["Hidemasa Togashi", "Masahiro Nakagawa"]
+  gem.email         = ["togachiro@gmail.com", "repeatedly@gmail.com"]
+  gem.description   = %q{Fluentd plugin for Apache Kafka > 0.8}
+  gem.summary       = %q{Fluentd plugin for Apache Kafka > 0.8}
+  gem.homepage      = "https://github.com/fluent/fluent-plugin-kafka"
+  gem.license       = "Apache-2.0"
+
+  gem.files         = `git ls-files`.split($\)
+  gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.name          = "fluent-plugin-kafka-xst"
+  gem.require_paths = ["lib"]
+  gem.version       = '0.19.1'
+  gem.required_ruby_version = ">= 2.1.0"
+
+  gem.add_dependency "fluentd", [">= 0.10.58", "< 2"]
+  gem.add_dependency 'ltsv'
+  gem.add_dependency 'ruby-kafka', '>= 1.5.0', '< 2'
+  gem.add_development_dependency "rake", ">= 0.9.2"
+  gem.add_development_dependency "test-unit", ">= 3.0.8"
+  gem.add_development_dependency "test-unit-rr", "~> 1.0"
+  gem.add_development_dependency "webrick"
+  gem.add_development_dependency "digest-murmurhash"
+end

data/lib/fluent/plugin/in_kafka.rb

@@ -0,0 +1,388 @@
+require 'fluent/input'
+require 'fluent/time'
+require 'fluent/plugin/kafka_plugin_util'
+
+class Fluent::KafkaInput < Fluent::Input
+  Fluent::Plugin.register_input('kafka', self)
+
+  config_param :format, :string, :default => 'json',
+               :desc => "Supported format: (json|text|ltsv|msgpack)"
+  config_param :message_key, :string, :default => 'message',
+               :desc => "For 'text' format only."
+  config_param :host, :string, :default => nil,
+               :desc => "Broker host"
+  config_param :port, :integer, :default => nil,
+               :desc => "Broker port"
+  config_param :brokers, :string, :default => 'localhost:9092',
+               :desc => "List of broker-host:port, separate with comma, must set."
+  config_param :interval, :integer, :default => 1, # seconds
+               :desc => "Interval (Unit: seconds)"
+  config_param :topics, :string, :default => nil,
+               :desc => "Listening topics(separate with comma',')"
+  config_param :client_id, :string, :default => 'kafka'
+  config_param :sasl_over_ssl, :bool, :default => true,
+               :desc => "Set to false to prevent SSL strict mode when using SASL authentication"
+  config_param :partition, :integer, :default => 0,
+               :desc => "Listening partition"
+  config_param :offset, :integer, :default => -1,
+               :desc => "Listening start offset"
+  config_param :add_prefix, :string, :default => nil,
+               :desc => "Tag prefix"
+  config_param :add_suffix, :string, :default => nil,
+               :desc => "tag suffix"
+  config_param :add_offset_in_record, :bool, :default => false
+  config_param :tag_source, :enum, :list => [:topic, :record], :default => :topic,
+               :desc => "Source for the fluentd event tag"
+  config_param :record_tag_key, :string, :default => 'tag',
+               :desc => "Tag field when tag_source is 'record'"
+
+  config_param :offset_zookeeper, :string, :default => nil
+  config_param :offset_zk_root_node, :string, :default => '/fluent-plugin-kafka'
+  config_param :use_record_time, :bool, :default => false,
+               :desc => "Replace message timestamp with contents of 'time' field.",
+               :deprecated => "Use 'time_source record' instead."
+  config_param :time_source, :enum, :list => [:now, :kafka, :record], :default => :now,
+               :desc => "Source for message timestamp."
+  config_param :record_time_key, :string, :default => 'time',
+               :desc => "Time field when time_source is 'record'"
+  config_param :get_kafka_client_log, :bool, :default => false
+  config_param :time_format, :string, :default => nil,
+               :desc => "Time format to be used to parse 'time' field."
+  config_param :kafka_message_key, :string, :default => nil,
+               :desc => "Set kafka's message key to this field"
+
+  # Kafka#fetch_messages options
+  config_param :max_bytes, :integer, :default => nil,
+               :desc => "Maximum number of bytes to fetch."
+  config_param :max_wait_time, :integer, :default => nil,
+               :desc => "How long to block until the server sends us data."
+  config_param :min_bytes, :integer, :default => nil,
+               :desc => "Smallest amount of data the server should send us."
+
+  include Fluent::KafkaPluginUtil::SSLSettings
+  include Fluent::KafkaPluginUtil::SaslSettings
+
+  unless method_defined?(:router)
+    define_method("router") { Fluent::Engine }
+  end
+
+  def initialize
+    super
+    require 'kafka'
+
+    @time_parser = nil
+    @zookeeper = nil
+  end
+
+  def configure(conf)
+    super
+
+    @topic_list = []
+    if @topics
+      @topic_list = @topics.split(',').map { |topic|
+        TopicEntry.new(topic.strip, @partition, @offset)
+      }
+    else
+      conf.elements.select { |element| element.name == 'topic' }.each do |element|
+        unless element.has_key?('topic')
+          raise Fluent::ConfigError, "kafka: 'topic' is a require parameter in 'topic element'."
+        end
+        partition = element.has_key?('partition') ? element['partition'].to_i : 0
+        offset = element.has_key?('offset') ? element['offset'].to_i : -1
+        @topic_list.push(TopicEntry.new(element['topic'], partition, offset))
+      end
+    end
+
+    if @topic_list.empty?
+      raise Fluent::ConfigError, "kafka: 'topics' or 'topic element' is a require parameter"
+    end
+
+    # For backward compatibility
+    @brokers = case
+               when @host && @port
+                 ["#{@host}:#{@port}"]
+               when @host
+                 ["#{@host}:9092"]
+               when @port
+                 ["localhost:#{@port}"]
+               else
+                 @brokers
+               end
+
+    if conf['max_wait_ms']
+      log.warn "'max_wait_ms' parameter is deprecated. Use second unit 'max_wait_time' instead"
+      @max_wait_time = conf['max_wait_ms'].to_i / 1000
+    end
+
+    @max_wait_time = @interval if @max_wait_time.nil?
+
+    require 'zookeeper' if @offset_zookeeper
+
+    @parser_proc = setup_parser(conf)
+
+    @time_source = :record if @use_record_time
+
+    if @time_source == :record and @time_format
+      if defined?(Fluent::TimeParser)
+        @time_parser = Fluent::TimeParser.new(@time_format)
+      else
+        @time_parser = Fluent::TextParser::TimeParser.new(@time_format)
+      end
+    end
+  end
+
+  def setup_parser(conf)
+    case @format
+    when 'json'
+      begin
+        require 'oj'
+        Oj.default_options = Fluent::DEFAULT_OJ_OPTIONS
+        Proc.new { |msg, te|
+          r = Oj.load(msg.value)
+          add_offset_in_hash(r, te, msg.offset) if @add_offset_in_record
+          r
+        }
+      rescue LoadError
+        require 'yajl'
+        Proc.new { |msg, te|
+          r = Yajl::Parser.parse(msg.value)
+          add_offset_in_hash(r, te, msg.offset) if @add_offset_in_record
+          r
+        }
+      end
+    when 'ltsv'
+      require 'ltsv'
+      Proc.new { |msg, te|
+        r = LTSV.parse(msg.value, {:symbolize_keys => false}).first
+        add_offset_in_hash(r, te, msg.offset) if @add_offset_in_record
+        r
+      }
+    when 'msgpack'
+      require 'msgpack'
+      Proc.new { |msg, te|
+        r = MessagePack.unpack(msg.value)
+        add_offset_in_hash(r, te, msg.offset) if @add_offset_in_record
+        r
+      }
+    when 'text'
+      Proc.new { |msg, te|
+        r = {@message_key => msg.value}
+        add_offset_in_hash(r, te, msg.offset) if @add_offset_in_record
+        r
+      }
+    else
+      @custom_parser = Fluent::Plugin.new_parser(conf['format'])
+      @custom_parser.configure(conf)
+      Proc.new { |msg|
+        @custom_parser.parse(msg.value) {|_time, record|
+          record
+        }
+      }
+    end
+  end
+
+  def add_offset_in_hash(hash, te, offset)
+    hash['kafka_topic'.freeze] = te.topic
+    hash['kafka_partition'.freeze] = te.partition
+    hash['kafka_offset'.freeze] = offset
+  end
+
+  def start
+    super
+
+    @loop = Coolio::Loop.new
+    opt = {}
+    opt[:max_bytes] = @max_bytes if @max_bytes
+    opt[:max_wait_time] = @max_wait_time if @max_wait_time
+    opt[:min_bytes] = @min_bytes if @min_bytes
+
+    logger = @get_kafka_client_log ? log : nil
+    if @scram_mechanism != nil && @username != nil && @password != nil
+      @kafka = Kafka.new(seed_brokers: @brokers, client_id: @client_id, logger: logger, ssl_ca_cert_file_path: @ssl_ca_cert,
+                         ssl_client_cert: read_ssl_file(@ssl_client_cert), ssl_client_cert_key: read_ssl_file(@ssl_client_cert_key),
+                         ssl_client_cert_key_password: @ssl_client_cert_key_password,
+                         ssl_ca_certs_from_system: @ssl_ca_certs_from_system, sasl_scram_username: @username, sasl_scram_password: @password,
+                         sasl_scram_mechanism: @scram_mechanism, sasl_over_ssl: @sasl_over_ssl, ssl_verify_hostname: @ssl_verify_hostname)
+    elsif @username != nil && @password != nil
+      @kafka = Kafka.new(seed_brokers: @brokers, client_id: @client_id, logger: logger, ssl_ca_cert_file_path: @ssl_ca_cert,
+                         ssl_client_cert: read_ssl_file(@ssl_client_cert), ssl_client_cert_key: read_ssl_file(@ssl_client_cert_key),
+                         ssl_client_cert_key_password: @ssl_client_cert_key_password,
+                         ssl_ca_certs_from_system: @ssl_ca_certs_from_system,sasl_plain_username: @username, sasl_plain_password: @password,
+                         sasl_over_ssl: @sasl_over_ssl, ssl_verify_hostname: @ssl_verify_hostname)
+    else
+      @kafka = Kafka.new(seed_brokers: @brokers, client_id: @client_id, logger: logger, ssl_ca_cert_file_path: @ssl_ca_cert,
+                         ssl_client_cert: read_ssl_file(@ssl_client_cert), ssl_client_cert_key: read_ssl_file(@ssl_client_cert_key),
+                         ssl_client_cert_key_password: @ssl_client_cert_key_password,
+                         ssl_ca_certs_from_system: @ssl_ca_certs_from_system, sasl_gssapi_principal: @principal, sasl_gssapi_keytab: @keytab,
+                         ssl_verify_hostname: @ssl_verify_hostname)
+    end
+
+    @zookeeper = Zookeeper.new(@offset_zookeeper) if @offset_zookeeper
+
+    @topic_watchers = @topic_list.map {|topic_entry|
+      offset_manager = OffsetManager.new(topic_entry, @zookeeper, @offset_zk_root_node) if @offset_zookeeper
+      TopicWatcher.new(
+        topic_entry,
+        @kafka,
+        interval,
+        @parser_proc,
+        @add_prefix,
+        @add_suffix,
+        offset_manager,
+        router,
+        @kafka_message_key,
+        @time_source,
+        @record_time_key,
+        @tag_source,
+        @record_tag_key,
+        opt)
+    }
+    @topic_watchers.each {|tw|
+      tw.attach(@loop)
+    }
+    @thread = Thread.new(&method(:run))
+  end
+
+  def shutdown
+    @loop.stop
+    @zookeeper.close! if @zookeeper
+    @thread.join
+    @kafka.close
+    super
+  end
+
+  def run
+    @loop.run
+  rescue => e
+    $log.error "unexpected error", :error => e.to_s
+    $log.error_backtrace
+  end
+
+  class TopicWatcher < Coolio::TimerWatcher
+    def initialize(topic_entry, kafka, interval, parser, add_prefix, add_suffix, offset_manager, router, kafka_message_key, time_source, record_time_key, tag_source, record_tag_key, options={})
+      @topic_entry = topic_entry
+      @kafka = kafka
+      @callback = method(:consume)
+      @parser = parser
+      @add_prefix = add_prefix
+      @add_suffix = add_suffix
+      @options = options
+      @offset_manager = offset_manager
+      @router = router
+      @kafka_message_key = kafka_message_key
+      @time_source = time_source
+      @record_time_key = record_time_key
+      @tag_source = tag_source
+      @record_tag_key = record_tag_key
+
+      @next_offset = @topic_entry.offset
+      if @topic_entry.offset == -1 && offset_manager
+        @next_offset = offset_manager.next_offset
+      end
+      @fetch_args = {
+        topic: @topic_entry.topic,
+        partition: @topic_entry.partition,
+      }.merge(@options)
+
+      super(interval, true)
+    end
+
+    def on_timer
+      @callback.call
+    rescue => e
+      # TODO log?
+      $log.error e.to_s
+      $log.error_backtrace
+    end
+
+    def consume
+      offset = @next_offset
+      @fetch_args[:offset] = offset
+      messages = @kafka.fetch_messages(**@fetch_args)
+
+      return if messages.size.zero?
+
+      es = Fluent::MultiEventStream.new
+      tag = @topic_entry.topic
+      tag = @add_prefix + "." + tag if @add_prefix
+      tag = tag + "." + @add_suffix if @add_suffix
+
+      messages.each { |msg|
+        begin
+          record = @parser.call(msg, @topic_entry)
+          if @tag_source == :record
+            tag = record[@record_tag_key]
+            tag = @add_prefix + "." + tag if @add_prefix
+            tag = tag + "." + @add_suffix if @add_suffix
+          end
+          case @time_source
+          when :kafka
+            record_time = Fluent::EventTime.from_time(msg.create_time)
+          when :now
+            record_time = Fluent::Engine.now
+          when :record
+            if @time_format
+              record_time = @time_parser.parse(record[@record_time_key])
+            else
+              record_time = record[@record_time_key]
+            end
+          else
+            $log.fatal "BUG: invalid time_source: #{@time_source}"
+          end
+          if @kafka_message_key
+            record[@kafka_message_key] = msg.key
+          end
+          es.add(record_time, record)
+        rescue => e
+          $log.warn "parser error in #{@topic_entry.topic}/#{@topic_entry.partition}", :error => e.to_s, :value => msg.value, :offset => msg.offset
+          $log.debug_backtrace
+        end
+      }
+      offset = messages.last.offset + 1
+
+      unless es.empty?
+        @router.emit_stream(tag, es)
+
+        if @offset_manager
+          @offset_manager.save_offset(offset)
+        end
+        @next_offset = offset
+      end
+    end
+  end
+
+  class TopicEntry
+    def initialize(topic, partition, offset)
+      @topic = topic
+      @partition = partition
+      @offset = offset
+    end
+    attr_reader :topic, :partition, :offset
+  end
+
+  class OffsetManager
+    def initialize(topic_entry, zookeeper, zk_root_node)
+      @zookeeper = zookeeper
+      @zk_path = "#{zk_root_node}/#{topic_entry.topic}/#{topic_entry.partition}/next_offset"
+      create_node(@zk_path, topic_entry.topic, topic_entry.partition)
+    end
+
+    def create_node(zk_path, topic, partition)
+      path = ""
+      zk_path.split(/(\/[^\/]+)/).reject(&:empty?).each { |dir|
+        path = path + dir
+        @zookeeper.create(:path => "#{path}")
+      }
+      $log.trace "use zk offset node : #{path}"
+    end
+
+    def next_offset
+      @zookeeper.get(:path => @zk_path)[:data].to_i
+    end
+
+    def save_offset(offset)
+      @zookeeper.set(:path => @zk_path, :data => offset.to_s)
+      $log.trace "update zk offset node : #{offset.to_s}"
+    end
+  end
+end