fluentd 1.13.3 → 1.14.0.rc

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1603571f755a962fde001cbb33fd24ad8929633bdeb38832e61f45da2c0dfba5
-  data.tar.gz: 5c2d45299617e9e4424758c76eb53d78889387cedc28d1d8bac549402094b6af
+  metadata.gz: eab8e56529b8472ac765f6186687659e56081d0d2b21264c05f1e4550965eff6
+  data.tar.gz: f64f9a36befc145b9ae862f6eead5588670f5c3252ca821608ce2bc6d8649ed7
 SHA512:
-  metadata.gz: 8dd4cc4947766345b5455fbae8043610d1c3d33e0e2ea9ea348a866722c099545301236d507d754e31b063017bb7daefdc8aef0e2a8078b67df83eb3c2489b2c
-  data.tar.gz: f0d126ed7704b519c2b7db5693de9685db7fd3ed1844f770a08e1e3351f3daf0f66e4b88c9435b3e456f40e38480c9af7652241f029a36b7a13184364a595251
+  metadata.gz: 0a3d652b47ce320495511a39c41d6af65825a7455e1a67c7f2c93d7a77c13737acce3c86d79de0c0a95c6a053ba3e4bf3e0dfe19767327f4ea45f489321ef8a6
+  data.tar.gz: 0421fa9b68bf1317ff8e44f33908f7603f4298214daa4a0438e36586eca9829b2701144e2148a4a0ccbfeaaf87dd542be4227918eeab523dd3fdbcca0a08d839

data/lib/fluent/command/fluentd.rb

@@ -134,6 +134,14 @@ op.on('--strict-config-value', "Parse config values strictly", TrueClass) {|b|
   opts[:strict_config_value] = b
 }
 
+op.on('--enable-input-metrics', "Enable input plugin metrics on fluentd", TrueClass) {|b|
+  opts[:enable_input_metrics] = b
+}
+
+op.on('--enable-size-metrics', "Enable plugin record size metrics on fluentd", TrueClass) {|b|
+  opts[:enable_size_metrics] = b
+}
+
 op.on('-v', '--verbose', "increase verbose level (-v: debug, -vv: trace)", TrueClass) {|b|
   if b
     opts[:log_level] = [opts[:log_level] - 1, Fluent::Log::LEVEL_TRACE].max

data/lib/fluent/compat/output.rb

@@ -310,7 +310,7 @@ module Fluent
       # original implementation of v0.12 BufferedOutput
       def emit(tag, es, chain, key="")
         # this method will not be used except for the case that plugin calls super
-        @emit_count += 1
+        @emit_count_metrics.inc
         data = format_stream(tag, es)
         if @buffer.emit(key, data, chain)
           submit_flush
@@ -337,14 +337,14 @@ module Fluent
       # because v0.12 BufferedOutput may overrides #format_stream, but original #handle_stream_simple method doesn't consider about it
       def handle_stream_simple(tag, es, enqueue: false)
         if @overrides_emit
-          current_emit_count = @emit_count
+          current_emit_count = @emit_count_metrics.get
           size = es.size
           key = data = nil
           begin
             emit(tag, es, NULL_OUTPUT_CHAIN)
             key, data = self.last_emit_via_buffer
           ensure
-            @emit_count = current_emit_count
+            @emit_count_metrics.set(current_emit_count)
             self.last_emit_via_buffer = nil
           end
           # on-the-fly key assignment can be done, and it's not configurable if Plugin#emit does it dynamically
@@ -352,7 +352,8 @@ module Fluent
           write_guard do
             @buffer.write({meta => data}, format: ->(_data){ _data }, size: ->(){ size }, enqueue: enqueue)
           end
-          @counter_mutex.synchronize{ @emit_records += size }
+          @emit_records_metrics.add(es.size)
+          @emit_size_metrics.add(es.to_msgpack_stream.bytesize) if @enable_size_metrics
           return [meta]
         end
 
@@ -363,7 +364,8 @@ module Fluent
           write_guard do
             @buffer.write({meta => bulk}, format: ->(_data){ _data }, size: ->(){ size }, enqueue: enqueue)
           end
-          @counter_mutex.synchronize{ @emit_records += size }
+          @emit_records_metrics.add(es.size)
+          @emit_size_metrics.add(es.to_msgpack_stream.bytesize) if @enable_size_metrics
           return [meta]
         end
 
@@ -373,7 +375,8 @@ module Fluent
         write_guard do
           @buffer.write({meta => data}, enqueue: enqueue)
         end
-        @counter_mutex.synchronize{ @emit_records += size }
+        @emit_records_metrics.add(es.size)
+        @emit_size_metrics.add(es.to_msgpack_stream.bytesize) if @enable_size_metrics
         [meta]
       end
 

data/lib/fluent/event_router.rb

@@ -47,6 +47,8 @@ module Fluent
       @match_cache = MatchCache.new
       @default_collector = default_collector
       @emit_error_handler = emit_error_handler
+      @metric_callbacks = {}
+      @caller_plugin_id = nil
     end
 
     attr_accessor :default_collector
@@ -83,6 +85,22 @@ module Fluent
       @match_rules << Rule.new(pattern, collector)
     end
 
+    def add_metric_callbacks(caller_plugin_id, callback)
+      @metric_callbacks[caller_plugin_id] = callback
+    end
+
+    def caller_plugin_id=(caller_plugin_id)
+      @caller_plugin_id = caller_plugin_id
+    end
+
+    def find_callback
+      if @caller_plugin_id
+        @metric_callbacks[@caller_plugin_id]
+      else
+        nil
+      end
+    end
+
     def emit(tag, time, record)
       unless record.nil?
         emit_stream(tag, OneEventStream.new(time, record))
@@ -95,6 +113,9 @@ module Fluent
 
     def emit_stream(tag, es)
       match(tag).emit_events(tag, es)
+      if callback = find_callback
+        callback.call(es)
+      end
     rescue => e
       @emit_error_handler.handle_emits_error(tag, es, e)
     end
@@ -175,7 +196,11 @@ module Fluent
           if optimizable?
             optimized_filter_stream(tag, es)
           else
-            @filters.reduce(es) { |acc, filter| filter.filter_stream(tag, acc) }
+            @filters.reduce(es) { |acc, filter|
+              filtered_es = filter.filter_stream(tag, acc)
+              filter.measure_metrics(filtered_es)
+              filtered_es
+            }
           end
         end
 
@@ -193,6 +218,7 @@ module Fluent
                   begin
                     filtered_time, filtered_record = filter.filter_with_time(tag, filtered_time, filtered_record)
                     throw :break_loop unless filtered_record && filtered_time
+                    filter.measure_metrics(OneEventStream.new(time, record))
                   rescue => e
                     filter.router.emit_error_event(tag, filtered_time, filtered_record, e)
                   end
@@ -200,6 +226,7 @@ module Fluent
                   begin
                     filtered_record = filter.filter(tag, filtered_time, filtered_record)
                     throw :break_loop unless filtered_record
+                    filter.measure_metrics(OneEventStream.new(time, record))
                   rescue => e
                     filter.router.emit_error_event(tag, filtered_time, filtered_record, e)
                   end

data/lib/fluent/plugin.rb

@@ -36,8 +36,9 @@ module Fluent
     FORMATTER_REGISTRY = Registry.new(:formatter, 'fluent/plugin/formatter_',  dir_search_prefix: 'formatter_')
     STORAGE_REGISTRY   = Registry.new(:storage,   'fluent/plugin/storage_',    dir_search_prefix: 'storage_')
     SD_REGISTRY        = Registry.new(:sd,        'fluent/plugin/sd_',         dir_search_prefix: 'sd_')
+    METRICS_REGISTRY   = Registry.new(:metrics,   'fluent/plugin/metrics_',    dir_search_prefix: 'metrics_')
 
-    REGISTRIES = [INPUT_REGISTRY, OUTPUT_REGISTRY, FILTER_REGISTRY, BUFFER_REGISTRY, PARSER_REGISTRY, FORMATTER_REGISTRY, STORAGE_REGISTRY, SD_REGISTRY]
+    REGISTRIES = [INPUT_REGISTRY, OUTPUT_REGISTRY, FILTER_REGISTRY, BUFFER_REGISTRY, PARSER_REGISTRY, FORMATTER_REGISTRY, STORAGE_REGISTRY, SD_REGISTRY, METRICS_REGISTRY]
 
     def self.register_input(type, klass)
       register_impl('input', INPUT_REGISTRY, type, klass)
@@ -59,6 +60,10 @@ module Fluent
       register_impl('sd', SD_REGISTRY, type, klass)
     end
 
+    def self.register_metrics(type, klass)
+      register_impl('metrics', METRICS_REGISTRY, type, klass)
+    end
+
     def self.register_parser(type, klass_or_proc)
       if klass_or_proc.is_a?(Regexp)
         # This usage is not recommended for new API
@@ -121,6 +126,10 @@ module Fluent
       new_impl('sd', SD_REGISTRY, type, parent)
     end
 
+    def self.new_metrics(type, parent: nil)
+      new_impl('metrics', METRICS_REGISTRY, type, parent)
+    end
+
     class << self
       # This should be defined for fluent-plugin-config-formatter type arguments.
       alias_method :new_service_discovery, :new_sd

data/lib/fluent/plugin/bare_output.rb

@@ -23,37 +23,78 @@ require 'fluent/plugin_helper'
 module Fluent
   module Plugin
     class BareOutput < Base
+      include PluginHelper::Mixin # for metrics
+
       # DO NOT USE THIS plugin for normal output plugin. Use Output instead.
       # This output plugin base class is only for meta-output plugins
       # which cannot be implemented on MultiOutput.
       # E.g,: forest, config-expander
 
+      helpers_internal :metrics
+
       include PluginId
       include PluginLoggerMixin
       include PluginHelper::Mixin
 
-      attr_reader :num_errors, :emit_count, :emit_records
-
       def process(tag, es)
         raise NotImplementedError, "BUG: output plugins MUST implement this method"
       end
 
+      def num_errors
+        @num_errors_metrics.get
+      end
+
+      def emit_count
+        @emit_count_metrics.get
+      end
+
+      def emit_size
+        @emit_size_metrics.get
+      end
+
+      def emit_records
+        @emit_records_metrics.get
+      end
+
       def initialize
         super
         @counter_mutex = Mutex.new
         # TODO: well organized counters
-        @num_errors = 0
-        @emit_count = 0
-        @emit_records = 0
+        @num_errors_metrics = nil
+        @emit_count_metrics = nil
+        @emit_records_metrics = nil
+        @emit_size_metrics = nil
+      end
+
+      def configure(conf)
+        super
+
+        @num_errors_metrics = metrics_create(namespace: "fluentd", subsystem: "bare_output", name: "num_errors", help_text: "Number of count num errors")
+        @emit_count_metrics = metrics_create(namespace: "fluentd", subsystem: "bare_output", name: "emit_records", help_text: "Number of count emits")
+        @emit_records_metrics = metrics_create(namespace: "fluentd", subsystem: "bare_output", name: "emit_records", help_text: "Number of emit records")
+        @emit_size_metrics =  metrics_create(namespace: "fluentd", subsystem: "bare_output", name: "emit_size", help_text: "Total size of emit events")
+        @enable_size_metrics = !!system_config.enable_size_metrics
+      end
+
+      def statistics
+        stats = {
+          'num_errors' => @num_errors_metrics.get,
+          'emit_records' => @emit_records_metrics.get,
+          'emit_count' => @emit_count_metrics.get,
+          'emit_size' => @emit_size_metrics.get,
+        }
+
+        { 'bare_output' => stats }
       end
 
       def emit_sync(tag, es)
-        @counter_mutex.synchronize{ @emit_count += 1 }
+        @emit_count_metrics.inc
         begin
           process(tag, es)
-          @counter_mutex.synchronize{ @emit_records += es.size }
+          @emit_records_metrics.add(es.size)
+          @emit_size_metrics.add(es.to_msgpack_stream.bytesize) if @enable_size_metrics
         rescue
-          @counter_mutex.synchronize{ @num_errors += 1 }
+          @num_errors_metrics.inc
           raise
         end
       end

data/lib/fluent/plugin/buffer.rb

@@ -16,6 +16,8 @@
 
 require 'fluent/plugin/base'
 require 'fluent/plugin/owned_by_mixin'
+require 'fluent/plugin_id'
+require 'fluent/plugin_helper'
 require 'fluent/unique_id'
 require 'fluent/ext_monitor_require'
 
@@ -24,7 +26,9 @@ module Fluent
     class Buffer < Base
       include OwnedByMixin
       include UniqueId::Mixin
+      include PluginId
       include MonitorMixin
+      include PluginHelper::Mixin # for metrics
 
       class BufferError < StandardError; end
       class BufferOverflowError < BufferError; end
@@ -39,6 +43,8 @@ module Fluent
 
       configured_in :buffer
 
+      helpers_internal :metrics
+
       # TODO: system total buffer limit size in bytes by SystemConfig
 
       config_param :chunk_limit_size, :size, default: DEFAULT_CHUNK_LIMIT_SIZE
@@ -153,8 +159,11 @@ module Fluent
         end
       end
 
+      # for metrics
+      attr_reader :stage_size_metrics, :stage_length_metrics, :queue_size_metrics, :queue_length_metrics
+      attr_reader :available_buffer_space_ratios_metrics, :total_queued_size_metrics
+      attr_reader :newest_timekey_metrics, :oldest_timekey_metrics
       # for tests
-      attr_accessor :stage_size, :queue_size
       attr_reader :stage, :queue, :dequeued, :queued_num
 
       def initialize
@@ -171,12 +180,35 @@ module Fluent
         @queued_num = {} # metadata => int (number of queued chunks)
         @dequeued_num = {} # metadata => int (number of dequeued chunks)
 
-        @stage_size = @queue_size = 0
+        @stage_length_metrics = nil
+        @stage_size_metrics = nil
+        @queue_length_metrics = nil
+        @queue_size_metrics = nil
+        @available_buffer_space_ratios_metrics = nil
+        @total_queued_size_metrics = nil
+        @newest_timekey_metrics = nil
+        @oldest_timekey_metrics = nil
         @timekeys = Hash.new(0)
         @enable_update_timekeys = false
         @mutex = Mutex.new
       end
 
+      def stage_size
+        @stage_size_metrics.get
+      end
+
+      def stage_size=(value)
+        @stage_size_metrics.set(value)
+      end
+
+      def queue_size
+        @queue_size_metrics.get
+      end
+
+      def queue_size=(value)
+        @queue_size_metrics.set(value)
+      end
+
       def persistent?
         false
       end
@@ -187,6 +219,28 @@ module Fluent
         unless @queue_limit_length.nil?
           @total_limit_size = @chunk_limit_size * @queue_limit_length
         end
+        @stage_length_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "stage_length",
+                                               help_text: 'Length of stage buffers', prefer_gauge: true)
+        @stage_length_metrics.set(0)
+        @stage_size_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "stage_byte_size",
+                                             help_text: 'Total size of stage buffers', prefer_gauge: true)
+        @stage_size_metrics.set(0) # Ensure zero.
+        @queue_length_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "queue_length",
+                                               help_text: 'Length of queue buffers', prefer_gauge: true)
+        @queue_length_metrics.set(0)
+        @queue_size_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "queue_byte_size",
+                                             help_text: 'Total size of queue buffers', prefer_gauge: true)
+        @queue_size_metrics.set(0) # Ensure zero.
+        @available_buffer_space_ratios_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "available_buffer_space_ratios",
+                                                                help_text: 'Ratio of available space in buffer', prefer_gauge: true)
+        @available_buffer_space_ratios_metrics.set(100) # Default is 100%.
+        @total_queued_size_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "total_queued_size",
+                                                    help_text: 'Total size of stage and queue buffers', prefer_gauge: true)
+        @total_queued_size_metrics.set(0)
+        @newest_timekey_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "newest_timekey",
+                                                 help_text: 'Newest timekey in buffer', prefer_gauge: true)
+        @oldest_timekey_metrics = metrics_create(namespace: "fluentd", subsystem: "buffer", name: "oldest_timekey",
+                                                 help_text: 'Oldest timekey in buffer', prefer_gauge: true)
       end
 
       def enable_update_timekeys
@@ -198,15 +252,15 @@ module Fluent
 
         @stage, @queue = resume
         @stage.each_pair do |metadata, chunk|
-          @stage_size += chunk.bytesize
+          @stage_size_metrics.add(chunk.bytesize)
         end
         @queue.each do |chunk|
           @queued_num[chunk.metadata] ||= 0
           @queued_num[chunk.metadata] += 1
-          @queue_size += chunk.bytesize
+          @queue_size_metrics.add(chunk.bytesize)
         end
         update_timekeys
-        log.debug "buffer started", instance: self.object_id, stage_size: @stage_size, queue_size: @queue_size
+        log.debug "buffer started", instance: self.object_id, stage_size: @stage_size_metrics.get, queue_size: @queue_size_metrics.get
       end
 
       def close
@@ -228,17 +282,19 @@ module Fluent
       def terminate
         super
         @dequeued = @stage = @queue = @queued_num = nil
-        @stage_size = @queue_size = 0
+        @stage_length_metrics = @stage_size_metrics = @queue_length_metrics = @queue_size_metrics = nil
+        @available_buffer_space_ratios_metrics = @total_queued_size_metrics = nil
+        @newest_timekey_metrics = @oldest_timekey_metrics = nil
         @timekeys.clear
       end
 
       def storable?
-        @total_limit_size > @stage_size + @queue_size
+        @total_limit_size > @stage_size_metrics.get + @queue_size_metrics.get
       end
 
       ## TODO: for back pressure feature
       # def used?(ratio)
-      #   @total_limit_size * ratio > @stage_size + @queue_size
+      #   @total_limit_size * ratio > @stage_size_metrics.get + @queue_size_metrics.get
       # end
 
       def resume
@@ -344,7 +400,7 @@ module Fluent
           #
           staged_bytesizes_by_chunk.each do |chunk, bytesize|
             chunk.synchronize do
-              synchronize { @stage_size += bytesize }
+              synchronize { @stage_size_metrics.add(bytesize) }
               log.on_trace { log.trace { "chunk #{chunk.path} size_added: #{bytesize} new_size: #{chunk.bytesize}" } }
             end
           end
@@ -361,7 +417,7 @@ module Fluent
                     u.metadata.seq = 0
                     synchronize {
                       @stage[m] = u.staged!
-                      @stage_size += u.bytesize
+                      @stage_size_metrics.add(u.bytesize)
                     }
                   end
                 end
@@ -428,8 +484,8 @@ module Fluent
               chunk.enqueued!
             end
             bytesize = chunk.bytesize
-            @stage_size -= bytesize
-            @queue_size += bytesize
+            @stage_size_metrics.sub(bytesize)
+            @queue_size_metrics.add(bytesize)
           end
         end
         nil
@@ -446,7 +502,7 @@ module Fluent
             @queued_num[metadata] = @queued_num.fetch(metadata, 0) + 1
             chunk.enqueued!
           end
-          @queue_size += chunk.bytesize
+          @queue_size_metrics.add(chunk.bytesize)
         end
       end
 
@@ -531,7 +587,7 @@ module Fluent
           begin
             bytesize = chunk.bytesize
             chunk.purge
-            @queue_size -= bytesize
+            @queue_size_metrics.sub(bytesize)
           rescue => e
             log.error "failed to purge buffer chunk", chunk_id: dump_unique_id_hex(chunk_id), error_class: e.class, error: e
             log.error_backtrace
@@ -562,7 +618,7 @@ module Fluent
               log.error_backtrace
             end
           end
-          @queue_size = 0
+          @queue_size_metrics.set(0)
         end
       end
 
@@ -765,23 +821,29 @@ module Fluent
       ]
 
       def statistics
-        stage_size, queue_size = @stage_size, @queue_size
+        stage_size, queue_size = @stage_size_metrics.get, @queue_size_metrics.get
         buffer_space = 1.0 - ((stage_size + queue_size * 1.0) / @total_limit_size)
+        @stage_length_metrics.set(@stage.size)
+        @queue_length_metrics.set(@queue.size)
+        @available_buffer_space_ratios_metrics.set(buffer_space * 100)
+        @total_queued_size_metrics.set(stage_size + queue_size)
         stats = {
-          'stage_length' => @stage.size,
+          'stage_length' => @stage_length_metrics.get,
           'stage_byte_size' => stage_size,
-          'queue_length' => @queue.size,
+          'queue_length' => @queue_length_metrics.get,
           'queue_byte_size' => queue_size,
-          'available_buffer_space_ratios' => (buffer_space * 100).round(1),
-          'total_queued_size' => stage_size + queue_size,
+          'available_buffer_space_ratios' => @available_buffer_space_ratios_metrics.get.round(1),
+          'total_queued_size' => @total_queued_size_metrics.get,
         }
 
         tkeys = timekeys
         if (m = tkeys.min)
-          stats['oldest_timekey'] = m
+          @oldest_timekey_metrics.set(m)
+          stats['oldest_timekey'] = @oldest_timekey_metrics.get
         end
         if (m = tkeys.max)
-          stats['newest_timekey'] = m
+          @newest_timekey_metrics.set(m)
+          stats['newest_timekey'] = @newest_timekey_metrics.get
         end
 
         { 'buffer' => stats }