fluentd 1.10.4 → 1.11.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 578798388cf4aa5c2628456764505f249ee7c5c50b6cf3c45cfd48bfccc5a62e
-  data.tar.gz: 172bf6a9caf7b38ae9274c3a1ada15b828655d79c4186a27c90c5de6882707f6
+  metadata.gz: 269cadf10ed4159e0c5279177cde46de14d563637b38ccadc71da090a098a22e
+  data.tar.gz: 9bf1dca5f46b1f3b245231c0e08b3b709add909840c3241bf29401952e02f27a
 SHA512:
-  metadata.gz: 8056b113c52d327fc314166e0cc4fda3765afe0dd41614e644e423a5504ff164297c01ab03f90c605f43fce84e902b226de7847c9b7f41d40ad8b462b7f86d43
-  data.tar.gz: 4ac3f7fb124ab42e21df67901b7435d4040cea3ad20e58b60515aa3b198755bb763c3415e48760aab7ebe12c895ff827e47487209de26ad06bccd46594742907
+  metadata.gz: 115e9f075e91136de66003ced2eda6076c1cff31c1e8edae5bb063f4c97420e4f832bcf6539a1ac4ee50bbe28d2050a6782840e88c951fa8fcbb52c66a889569
+  data.tar.gz: 25ebff981ebc41c4b6c0de4a6710350a07653be263019a138bb7be62d97dd6593bff67b7dd4d6ca4d37d0f84427b05d4c9b06f56cc9bc344fa46e0cfa82a7d68

data/CHANGELOG.md

@@ -1,3 +1,26 @@
+# v1.11
+
+## Release v1.11.0 - 2020/06/04
+
+### New feature
+
+* in_unix: Use v1 API
+  https://github.com/fluent/fluentd/pull/2992
+
+### Enhancement
+
+* parser_syslog: Support any `time_format` for RFC3164 string parser
+  https://github.com/fluent/fluentd/pull/2988
+* parser_syslog: Add new parser for RFC5424
+  https://github.com/fluent/fluentd/pull/3015
+* Refactoring code
+  https://github.com/fluent/fluentd/pull/3019
+
+### Bug fix
+
+* in_gc_stat: Add `use_symbol_keys` parameter to emit string key record
+  https://github.com/fluent/fluentd/pull/3008
+
 # v1.10
 
 ## Release v1.10.4 - 2020/05/12

data/lib/fluent/plugin/in_gc_stat.rb

@@ -24,13 +24,22 @@ module Fluent::Plugin
 
     def initialize
       super
+      @key_map = nil
     end
 
     config_param :emit_interval, :time, default: 60
+    config_param :use_symbol_keys, :bool, default: true
     config_param :tag, :string
 
     def configure(conf)
       super
+
+      unless @use_symbol_keys
+        @key_map = {}
+        GC.stat.each_key { |key|
+          @key_map[key] = key.to_s
+        }
+      end
     end
 
     def multi_workers_ready?
@@ -50,6 +59,13 @@ module Fluent::Plugin
     def on_timer
       now = Fluent::EventTime.now
       record = GC.stat
+      unless @use_symbol_keys
+        new_record = {}
+        record.each_pair { |k, v|
+          new_record[@key_map[k]] = v
+        }
+        record = new_record
+      end
       router.emit(@tag, now, record)
     end
   end

data/lib/fluent/plugin/in_unix.rb

@@ -14,50 +14,68 @@
 #    limitations under the License.
 #
 
-require 'fileutils'
-require 'socket'
+require 'fluent/env'
+require 'fluent/plugin/input'
+require 'fluent/msgpack_factory'
 
 require 'cool.io'
 require 'yajl'
+require 'fileutils'
+require 'socket'
+
+module Fluent::Plugin
+  # TODO: This plugin will be 3rd party plugin
+  class UnixInput < Input
+    Fluent::Plugin.register_input('unix', self)
+
+    helpers :event_loop
+
+    def initialize
+      super
 
-require 'fluent/input'
-require 'fluent/event'
+      @lsock = nil
+    end
 
-module Fluent
-  # obsolete
-  class StreamInput < Input
-    config_param :blocking_timeout, :time, default: 0.5
+    desc 'The path to your Unix Domain Socket.'
+    config_param :path, :string, default: Fluent::DEFAULT_SOCKET_PATH
+    desc 'The backlog of Unix Domain Socket.'
+    config_param :backlog, :integer, default: nil
+    desc "New tag instead of incoming tag"
+    config_param :tag, :string, default: nil
+
+    def configure(conf)
+      super
+    end
 
     def start
       super
 
-      @loop = Coolio::Loop.new
       @lsock = listen
-      @loop.attach(@lsock)
-      @thread = Thread.new(&method(:run))
+      event_loop_attach(@lsock)
     end
 
     def shutdown
-      @loop.watchers.each {|w| w.detach }
-      @loop.stop
-      @lsock.close
-      @thread.join
+      if @lsock
+        event_loop_detach(@lsock)
+        @lsock.close
+      end
 
       super
     end
 
-    #def listen
-    #end
+    def listen
+      if File.exist?(@path)
+        log.warn "Found existing '#{@path}'. Remove this file for in_unix plugin"
+        File.unlink(@path)
+      end
+      FileUtils.mkdir_p(File.dirname(@path))
 
-    def run
-      @loop.run(@blocking_timeout)
-    rescue
-      log.error "unexpected error", error: $!.to_s
-      log.error_backtrace
+      log.info "listening fluent socket on #{@path}"
+      s = Coolio::UNIXServer.new(@path, Handler, log, method(:on_message))
+      s.listen(@backlog) unless @backlog.nil?
+      s
     end
 
-    private
-
     # message Entry {
     #   1: long time
     #   2: object record
@@ -79,23 +97,27 @@ module Fluent
     #   3: object record
     # }
     def on_message(msg)
-      # TODO format error
-      tag = msg[0].to_s
+      unless msg.is_a?(Array)
+        log.warn "incoming data is broken:", msg: msg
+        return
+      end
+
+      tag = @tag || (msg[0].to_s)
       entries = msg[1]
 
-      if entries.class == String
+      case entries
+      when String
         # PackedForward
-        es = MessagePackEventStream.new(entries)
+        es = Fluent::MessagePackEventStream.new(entries)
         router.emit_stream(tag, es)
 
-      elsif entries.class == Array
+      when Array
         # Forward
-        es = MultiEventStream.new
+        es = Fluent::MultiEventStream.new
         entries.each {|e|
           record = e[1]
           next if record.nil?
-          time = e[0]
-          time = (now ||= EventTime.now) if time.to_i == 0
+          time = convert_time(e[0])
           es.add(time, record)
         }
         router.emit_stream(tag, es)
@@ -105,25 +127,28 @@ module Fluent
         record = msg[2]
         return if record.nil?
 
-        time = msg[1]
-        time = EventTime.now if time.to_i == 0
+        time = convert_time(msg[1])
         router.emit(tag, time, record)
       end
     end
 
+    def convert_time(time)
+      case time
+      when nil, 0
+        Fluent::EventTime.now
+      when Fluent::EventTime
+        time
+      else
+        Fluent::EventTime.from_time(Time.at(time))
+      end
+    end
+
     class Handler < Coolio::Socket
       def initialize(io, log, on_message)
         super(io)
-        if io.is_a?(TCPSocket)
-          opt = [1, @timeout.to_i].pack('I!I!')  # { int l_onoff; int l_linger; }
-          io.setsockopt(Socket::SOL_SOCKET, Socket::SO_LINGER, opt)
-        end
+
         @on_message = on_message
         @log = log
-        @log.trace {
-          remote_port, remote_addr = *Socket.unpack_sockaddr_in(@_io.getpeername) rescue nil
-          "accepted fluent socket from '#{remote_addr}:#{remote_port}': object_id=#{self.object_id}"
-        }
       end
 
       def on_connect
@@ -131,13 +156,13 @@ module Fluent
 
       def on_read(data)
         first = data[0]
-        if first == '{' || first == '['
+        if first == '{'.freeze || first == '['.freeze
           m = method(:on_read_json)
-          @y = Yajl::Parser.new
-          @y.on_parse_complete = @on_message
+          @parser = Yajl::Parser.new
+          @parser.on_parse_complete = @on_message
         else
           m = method(:on_read_msgpack)
-          @u = Fluent::MessagePackFactory.msgpack_unpacker
+          @parser = Fluent::MessagePackFactory.msgpack_unpacker
         end
 
         singleton_class.module_eval do
@@ -147,17 +172,17 @@ module Fluent
       end
 
       def on_read_json(data)
-        @y << data
-      rescue
-        @log.error "unexpected error", error: $!.to_s
+        @parser << data
+      rescue => e
+        @log.error "unexpected error in json payload", error: e.to_s
         @log.error_backtrace
         close
       end
 
       def on_read_msgpack(data)
-        @u.feed_each(data, &@on_message)
-      rescue
-        @log.error "unexpected error", error: $!.to_s
+        @parser.feed_each(data, &@on_message)
+      rescue => e
+        @log.error "unexpected error in msgpack payload", error: e.to_s
         @log.error_backtrace
         close
       end
@@ -167,29 +192,4 @@ module Fluent
       end
     end
   end
-
-  class UnixInput < StreamInput
-    Plugin.register_input('unix', self)
-
-    desc 'The path to your Unix Domain Socket.'
-    config_param :path, :string, default: DEFAULT_SOCKET_PATH
-    desc 'The backlog of Unix Domain Socket.'
-    config_param :backlog, :integer, default: nil
-
-    def configure(conf)
-      super
-      #log.warn "'unix' input is obsoleted and will be removed. Use 'forward' instead."
-    end
-
-    def listen
-      if File.exist?(@path)
-        File.unlink(@path)
-      end
-      FileUtils.mkdir_p File.dirname(@path)
-      log.info "listening fluent socket on #{@path}"
-      s = Coolio::UNIXServer.new(@path, Handler, log, method(:on_message))
-      s.listen(@backlog) unless @backlog.nil?
-      s
-    end
-  end
 end

data/lib/fluent/plugin/parser_syslog.rb

@@ -63,16 +63,22 @@ module Fluent
       def initialize
         super
         @mutex = Mutex.new
-        @space_count = nil
+        @regexp = nil
+        @regexp3164 = nil
+        @regexp5424 = nil
+        @regexp_parser = nil
+        @time_parser_rfc3164 = nil
+        @time_parser_rfc5424 = nil
+        @space_count_rfc3164 = nil
         @space_count_rfc5424 = nil
+        @skip_space_count_rfc3164 = false
+        @skip_space_count_rfc5424 = false
+        @time_parser_rfc5424_without_subseconds = nil
       end
 
       def configure(conf)
         super
 
-        @time_parser_rfc3164 = @time_parser_rfc5424 = nil
-        @time_parser_rfc5424_without_subseconds = nil
-        @support_rfc5424_without_subseconds = false
         @regexp_parser = @parser_type == :regexp
         @regexp = case @message_format
                   when :rfc3164
@@ -85,27 +91,49 @@ module Fluent
                         alias_method :parse, :parse_rfc3164
                       end
                     end
+                    setup_time_parser_3164(@time_format)
                     RFC3164_WITHOUT_TIME_AND_PRI_REGEXP
                   when :rfc5424
-                    class << self
-                      alias_method :parse, :parse_rfc5424_regex
+                    if @regexp_parser
+                      class << self
+                        alias_method :parse, :parse_rfc5424_regex
+                      end
+                    else
+                      class << self
+                        alias_method :parse, :parse_rfc5424
+                      end
                     end
                     @time_format = @rfc5424_time_format unless conf.has_key?('time_format')
-                    @support_rfc5424_without_subseconds = true
+                    setup_time_parser_5424(@time_format)
                     RFC5424_WITHOUT_TIME_AND_PRI_REGEXP
                   when :auto
                     class << self
                       alias_method :parse, :parse_auto
                     end
-                    @time_parser_rfc3164 = time_parser_create(format: @time_format)
-                    @time_parser_rfc5424 = time_parser_create(format: @rfc5424_time_format)
+                    setup_time_parser_3164(@time_format)
+                    setup_time_parser_5424(@rfc5424_time_format)
                     nil
                   end
 
-        @space_count = @time_format.squeeze(' ').count(' ') + 1
-        @space_count_rfc5424 = @rfc5424_time_format.squeeze(' ').count(' ') + 1
-        @time_parser = time_parser_create
+        if @regexp_parser
+          @regexp3164 = RFC3164_WITHOUT_TIME_AND_PRI_REGEXP
+          @regexp5424 = RFC5424_WITHOUT_TIME_AND_PRI_REGEXP
+        end
+      end
+
+      def setup_time_parser_3164(time_fmt)
+        @time_parser_rfc3164 = time_parser_create(format: time_fmt)
+        if ['%b %d %H:%M:%S', '%b %d %H:%M:%S.%N'].include?(time_fmt)
+          @skip_space_count_rfc3164 = true
+        end
+        @space_count_rfc3164 = time_fmt.squeeze(' ').count(' ') + 1
+      end
+
+      def setup_time_parser_5424(time_fmt)
+        @time_parser_rfc5424 = time_parser_create(format: time_fmt)
         @time_parser_rfc5424_without_subseconds = time_parser_create(format: "%Y-%m-%dT%H:%M:%S%z")
+        @skip_space_count_rfc5424 = time_fmt.count(' ').zero?
+        @space_count_rfc5424 = time_fmt.squeeze(' ').count(' ') + 1
       end
 
       # this method is for tests
@@ -118,14 +146,13 @@ module Fluent
       end
 
       def parse_auto(text, &block)
-        if REGEXP_DETECT_RFC5424.match(text)
-          @regexp = RFC5424_WITHOUT_TIME_AND_PRI_REGEXP
-          @time_parser = @time_parser_rfc5424
-          @support_rfc5424_without_subseconds = true
-          parse_rfc5424_regex(text, &block)
+        if REGEXP_DETECT_RFC5424.match?(text)
+          if @regexp_parser
+            parse_rfc5424_regex(text, &block)
+          else
+            parse_rfc5424(text, &block)
+          end
         else
-          @regexp = RFC3164_WITHOUT_TIME_AND_PRI_REGEXP
-          @time_parser = @time_parser_rfc3164
           if @regexp_parser
             parse_rfc3164_regex(text, &block)
           else
@@ -134,6 +161,8 @@ module Fluent
         end
       end
 
+      SPLIT_CHAR = ' '.freeze
+
       def parse_rfc3164_regex(text, &block)
         idx = 0
         record = {}
@@ -151,22 +180,22 @@ module Fluent
 
         i = idx - 1
         sq = false
-        @space_count.times do
-          while text[i + 1] == ' '.freeze
+        @space_count_rfc3164.times do
+          while text[i + 1] == SPLIT_CHAR
             sq = true
             i += 1
           end
 
-          i = text.index(' '.freeze, i + 1)
+          i = text.index(SPLIT_CHAR, i + 1)
         end
 
-        time_str = sq ? text.slice(idx, i - idx).squeeze(' ') : text.slice(idx, i - idx)
-        time = @mutex.synchronize { @time_parser.parse(time_str) }
+        time_str = sq ? text.slice(idx, i - idx).squeeze(SPLIT_CHAR) : text.slice(idx, i - idx)
+        time = @mutex.synchronize { @time_parser_rfc3164.parse(time_str) }
         if @keep_time_key
           record['time'] = time_str
         end
 
-        parse_plain(time, text, i + 1, record, RFC3164_CAPTURES, &block)
+        parse_plain(@regexp3164, time, text, i + 1, record, RFC3164_CAPTURES, &block)
       end
 
       def parse_rfc5424_regex(text, &block)
@@ -186,40 +215,36 @@ module Fluent
         i = idx - 1
         sq = false
         @space_count_rfc5424.times {
-          while text[i + 1] == ' '.freeze
+          while text[i + 1] == SPLIT_CHAR
             sq = true
             i += 1
           end
 
-          i = text.index(' '.freeze, i + 1)
+          i = text.index(SPLIT_CHAR, i + 1)
         }
 
-        time_str = sq ? text.slice(idx, i - idx).squeeze(' '.freeze) : text.slice(idx, i - idx)
+        time_str = sq ? text.slice(idx, i - idx).squeeze(SPLIT_CHAR) : text.slice(idx, i - idx)
         time = @mutex.synchronize do
           begin
-            @time_parser.parse(time_str)
+            @time_parser_rfc5424.parse(time_str)
           rescue Fluent::TimeParser::TimeParseError => e
-            if @support_rfc5424_without_subseconds
-              log.trace(e)
-              @time_parser_rfc5424_without_subseconds.parse(time_str)
-            else
-              raise
-            end
+            log.trace(e)
+            @time_parser_rfc5424_without_subseconds.parse(time_str)
           end
         end
 
         if @keep_time_key
           record['time'] = time_str
         end
-        parse_plain(time, text, i + 1, record, RFC5424_CAPTURES, &block)
+        parse_plain(@regexp5424, time, text, i + 1, record, RFC5424_CAPTURES, &block)
       end
 
       # @param time [EventTime]
       # @param idx [Integer] note: this argument is needed to avoid string creation
       # @param record [Hash]
       # @param capture_list [Array] for performance
-      def parse_plain(time, text, idx, record, capture_list, &block)
-        m = @regexp.match(text, idx)
+      def parse_plain(re, time, text, idx, record, capture_list, &block)
+        m = re.match(text, idx)
         if m.nil?
           yield nil, nil
           return
@@ -244,8 +269,6 @@ module Fluent
         yield time, record
       end
 
-      SPLIT_CHAR = ' '.freeze
-
       def parse_rfc3164(text, &block)
         pri = nil
         cursor = 0
@@ -264,20 +287,35 @@ module Fluent
           end
         end
 
-        # header part
-        time_size = 15 # skip Mmm dd hh:mm:ss
-        time_end = text[cursor + time_size]
-        if time_end == SPLIT_CHAR
-          time_str = text.slice(cursor, time_size)
-          cursor += 16 # time + ' '
-        elsif time_end == '.'.freeze
-          # support subsecond time
-          i = text.index(SPLIT_CHAR, time_size)
-          time_str = text.slice(cursor, i - cursor)
-          cursor = i + 1
+        if @skip_space_count_rfc3164
+          # header part
+          time_size = 15 # skip Mmm dd hh:mm:ss
+          time_end = text[cursor + time_size]
+          if time_end == SPLIT_CHAR
+            time_str = text.slice(cursor, time_size)
+            cursor += 16 # time + ' '
+          elsif time_end == '.'.freeze
+            # support subsecond time
+            i = text.index(SPLIT_CHAR, time_size)
+            time_str = text.slice(cursor, i - cursor)
+            cursor = i + 1
+          else
+            yield nil, nil
+            return
+          end
         else
-          yield nil, nil
-          return
+          i = cursor - 1
+          sq = false
+          @space_count_rfc3164.times do
+            while text[i + 1] == SPLIT_CHAR
+              sq = true
+              i += 1
+            end
+            i = text.index(SPLIT_CHAR, i + 1)
+          end
+
+          time_str = sq ? text.slice(idx, i - cursor).squeeze(SPLIT_CHAR) : text.slice(cursor, i - cursor)
+          cursor = i + 1
         end
 
         i = text.index(SPLIT_CHAR, cursor)
@@ -325,7 +363,130 @@ module Fluent
         msg.chomp!
         record['message'] = msg
 
-        time = @time_parser.parse(time_str)
+        time = @time_parser_rfc3164.parse(time_str)
+        record['time'] = time_str if @keep_time_key
+
+        yield time, record
+      end
+
+      NILVALUE = '-'.freeze
+
+      def parse_rfc5424(text, &block)
+        pri = nil
+        cursor = 0
+        if @with_priority
+          if text.start_with?('<'.freeze)
+            i = text.index('>'.freeze, 1)
+            if i < 2
+              yield nil, nil
+              return
+            end
+            pri = text.slice(1, i - 1).to_i
+            i = text.index(SPLIT_CHAR, i)
+            cursor = i + 1
+          else
+            yield nil, nil
+            return
+          end
+        end
+
+        # timestamp part
+        if @skip_space_count_rfc5424
+          i = text.index(SPLIT_CHAR, cursor)
+          time_str = text.slice(cursor, i - cursor)
+          cursor = i + 1
+        else
+          i = cursor - 1
+          sq = false
+          @space_count_rfc5424.times do
+            while text[i + 1] == SPLIT_CHAR
+              sq = true
+              i += 1
+            end
+            i = text.index(SPLIT_CHAR, i + 1)
+          end
+
+          time_str = sq ? text.slice(idx, i - cursor).squeeze(SPLIT_CHAR) : text.slice(cursor, i - cursor)
+          cursor = i + 1
+        end
+
+        # Repeat same code for the performance
+
+        # host part
+        i = text.index(SPLIT_CHAR, cursor)
+        unless i
+          yield nil, nil
+          return
+        end
+        slice_size = i - cursor
+        host = text.slice(cursor, slice_size)
+        cursor += slice_size + 1
+
+        # ident part
+        i = text.index(SPLIT_CHAR, cursor)
+        unless i
+          yield nil, nil
+          return
+        end
+        slice_size = i - cursor
+        ident = text.slice(cursor, slice_size)
+        cursor += slice_size + 1
+
+        # pid part
+        i = text.index(SPLIT_CHAR, cursor)
+        unless i
+          yield nil, nil
+          return
+        end
+        slice_size = i - cursor
+        pid = text.slice(cursor, slice_size)
+        cursor += slice_size + 1
+
+        # msgid part
+        i = text.index(SPLIT_CHAR, cursor)
+        unless i
+          yield nil, nil
+          return
+        end
+        slice_size = i - cursor
+        msgid = text.slice(cursor, slice_size)
+        cursor += slice_size + 1
+
+        record = {'host' => host, 'ident' => ident, 'pid' => pid, 'msgid' => msgid}
+        record['pri'] = pri if pri
+
+        # extradata part
+        ed_start = text[cursor]
+        if ed_start == NILVALUE
+          record['extradata'] = NILVALUE
+          cursor += 1
+        else
+          start = cursor
+          i = text.index('] '.freeze, cursor)
+          extradata = if i
+                        diff = i + 1 - start # calculate ']' position
+                        cursor += diff
+                        text.slice(start, diff)
+                      else  # No message part case
+                        cursor = text.bytesize
+                        text.slice(start, cursor)
+                      end
+          extradata.tr!("\\".freeze, ''.freeze)
+          record['extradata'] = extradata
+        end
+
+        # message part
+        if cursor != text.bytesize
+          msg = text.slice(cursor + 1, text.bytesize)
+          msg.chomp!
+          record['message'] = msg
+        end
+
+        time = begin
+                 @time_parser_rfc5424.parse(time_str)
+               rescue Fluent::TimeParser::TimeParseError => e
+                 @time_parser_rfc5424_without_subseconds.parse(time_str)
+               end
         record['time'] = time_str if @keep_time_key
 
         yield time, record