Class: Fluent::Plugin::ElasticsearchOutput

Inherits:
Output
  • Object
show all
Includes:
ElasticsearchIndexTemplate, ElasticsearchConstants
Defined in:
lib/fluent/plugin/out_elasticsearch.rb

Direct Known Subclasses

ElasticsearchOutputDynamic

Defined Under Namespace

Classes: ConnectionFailure, MissingIdFieldError, RetryStreamError

Constant Summary collapse

DEFAULT_BUFFER_TYPE = 
"memory"
DEFAULT_ELASTICSEARCH_VERSION =

For compatibility.

5
DEFAULT_TYPE_NAME_ES_7x = 
"_doc".freeze
DEFAULT_TYPE_NAME = 
"fluentd".freeze

Constants included from ElasticsearchConstants

Fluent::Plugin::ElasticsearchConstants::BODY_DELIMITER, Fluent::Plugin::ElasticsearchConstants::CREATE_OP, Fluent::Plugin::ElasticsearchConstants::ID_FIELD, Fluent::Plugin::ElasticsearchConstants::INDEX_OP, Fluent::Plugin::ElasticsearchConstants::TIMESTAMP_FIELD, Fluent::Plugin::ElasticsearchConstants::UPDATE_OP, Fluent::Plugin::ElasticsearchConstants::UPSERT_OP

Instance Method Summary collapse

Methods included from ElasticsearchIndexTemplate

#get_template, #retry_install, #template_exists?, #template_install, #template_put, #templates_hash_install

Constructor Details

#initialize ⇒ ElasticsearchOutput

Returns a new instance of ElasticsearchOutput.



117
118
119
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 117

def initialize
  super
end

Instance Method Details

#append_record_to_messages(op, meta, header, record, msgs) ⇒ Object

append_record_to_messages adds a record to the bulk message payload to be submitted to Elasticsearch. Records that do not include ‘_id’ field are skipped when ‘write_operation’ is configured for ‘create’ or ‘update’

returns ‘true’ if record was appended to the bulk message

and 'false' otherwise



339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 339

def append_record_to_messages(op, meta, header, record, msgs)
  case op
  when UPDATE_OP, UPSERT_OP
    if meta.has_key?(ID_FIELD)
      header[UPDATE_OP] = meta
      msgs << @dump_proc.call(header) << BODY_DELIMITER
      msgs << @dump_proc.call(update_body(record, op)) << BODY_DELIMITER
      return true
    end
  when CREATE_OP
    if meta.has_key?(ID_FIELD)
      header[CREATE_OP] = meta
      msgs << @dump_proc.call(header) << BODY_DELIMITER
      msgs << @dump_proc.call(record) << BODY_DELIMITER
      return true
    end
  when INDEX_OP
    header[INDEX_OP] = meta
    msgs << @dump_proc.call(header) << BODY_DELIMITER
    msgs << @dump_proc.call(record) << BODY_DELIMITER
    return true
  end
  return false
end

#client ⇒ Object 



245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 245

def client
  @_es ||= begin
    excon_options = { client_key: @client_key, client_cert: @client_cert, client_key_pass: @client_key_pass }
    adapter_conf = lambda {|f| f.adapter :excon, excon_options }
    transport = Elasticsearch::Transport::Transport::HTTP::Faraday.new(get_connection_options.merge(
                                                                        options: {
                                                                          reload_connections: @reload_connections,
                                                                          reload_on_failure: @reload_on_failure,
                                                                          resurrect_after: @resurrect_after,
                                                                          retry_on_failure: 5,
                                                                          logger: @transport_logger,
                                                                          transport_options: {
                                                                            headers: { 'Content-Type' => @content_type.to_s },
                                                                            request: { timeout: @request_timeout },
                                                                            ssl: { verify: @ssl_verify, ca_file: @ca_file, version: @ssl_version }
                                                                          },
                                                                          http: {
                                                                            user: @user,
                                                                            password: @password
                                                                          }
                                                                        }), &adapter_conf)
    es = Elasticsearch::Client.new transport: transport

    begin
      raise ConnectionFailure, "Can not reach Elasticsearch cluster (#{connection_options_description})!" unless es.ping
    rescue *es.transport.host_unreachable_exceptions => e
      raise ConnectionFailure, "Can not reach Elasticsearch cluster (#{connection_options_description})! #{e.message}"
    end

    log.info "Connection opened to Elasticsearch cluster => #{connection_options_description}"
    es
  end
end

#configure(conf) ⇒ Object

Raises:

  • (Fluent::ConfigError) 


121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 121

def configure(conf)
  compat_parameters_convert(conf, :buffer)

  super
  raise Fluent::ConfigError, "'tag' in chunk_keys is required." if not @chunk_key_tag

  @time_parser = create_time_parser

  if @remove_keys
    @remove_keys = @remove_keys.split(/\s*,\s*/)
  end

  if @target_index_key && @target_index_key.is_a?(String)
    @target_index_key = @target_index_key.split '.'
  end

  if @target_type_key && @target_type_key.is_a?(String)
    @target_type_key = @target_type_key.split '.'
  end

  if @remove_keys_on_update && @remove_keys_on_update.is_a?(String)
    @remove_keys_on_update = @remove_keys_on_update.split ','
  end

  raise Fluent::ConfigError, "'max_retry_putting_template' must be positive number." if @max_retry_putting_template < 0
  if @template_name && @template_file
    retry_install(@max_retry_putting_template) do
      template_install(@template_name, @template_file, @template_overwrite)
    end
  elsif @templates
    retry_install(@max_retry_putting_template) do
      templates_hash_install(@templates, @template_overwrite)
    end
  end

  # Consider missing the prefix of "$." in nested key specifiers.
  @id_key = convert_compat_id_key(@id_key) if @id_key
  @parent_key = convert_compat_id_key(@parent_key) if @parent_key
  @routing_key = convert_compat_id_key(@routing_key) if @routing_key

  @meta_config_map = create_meta_config_map

  begin
    require 'oj'
    @dump_proc = Oj.method(:dump)
  rescue LoadError
    @dump_proc = Yajl.method(:dump)
  end

  if @user && m = @user.match(/%{(?<user>.*)}/)
    @user = URI.encode_www_form_component(m["user"])
  end
  if @password && m = @password.match(/%{(?<password>.*)}/)
    @password = URI.encode_www_form_component(m["password"])
  end

  if @hash_config
    raise Fluent::ConfigError, "@hash_config.hash_id_key and id_key must be equal." unless @hash_config.hash_id_key == @id_key
  end
  @transport_logger = nil
  if @with_transporter_log
    @transport_logger = log
    log_level = conf['@log_level'] || conf['log_level']
    log.warn "Consider to specify log_level with @log_level." unless log_level
  end

  @last_seen_major_version = detect_es_major_version rescue DEFAULT_ELASTICSEARCH_VERSION
  if @last_seen_major_version == 6 && @type_name != DEFAULT_TYPE_NAME_ES_7x
    log.info "Detected ES 6.x: ES 7.x will only accept `_doc` in type_name."
  end
  if @last_seen_major_version >= 7 && @type_name != DEFAULT_TYPE_NAME_ES_7x
    log.warn "Detected ES 7.x or above: `_doc` will be used as the document `_type`."
    @type_name = '_doc'.freeze
  end
end

#connection_options_description ⇒ Object 



324
325
326
327
328
329
330
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 324

def connection_options_description
  get_connection_options[:hosts].map do |host_info|
    attributes = host_info.dup
    attributes[:password] = 'obfuscated' if attributes.has_key?(:password)
    attributes.inspect
  end.join(', ')
end

#convert_compat_id_key(key) ⇒ Object 



202
203
204
205
206
207
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 202

def convert_compat_id_key(key)
  if key.include?('.') && !key.start_with?('$[')
    key = "$.#{key}" unless key.start_with?('$.')
  end
  key
end

#create_meta_config_map ⇒ Object 



209
210
211
212
213
214
215
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 209

def create_meta_config_map
  result = []
  result << [record_accessor_create(@id_key), '_id'] if @id_key
  result << [record_accessor_create(@parent_key), '_parent'] if @parent_key
  result << [record_accessor_create(@routing_key), '_routing'] if @routing_key
  result
end

#create_time_parser ⇒ Object

once fluent v0.14 is released we might be able to use Fluent::Parser::TimeParser, but it doesn’t quite do what we want - if gives

sec,nsec

where as we want something we can call ‘strftime` on…



220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 220

def create_time_parser
  if @time_key_format
    begin
      # Strptime doesn't support all formats, but for those it does it's
      # blazingly fast.
      strptime = Strptime.new(@time_key_format)
      Proc.new { |value| strptime.exec(value).to_datetime }
    rescue
      # Can happen if Strptime doesn't recognize the format; or
      # if strptime couldn't be required (because it's not installed -- it's
      # ruby 2 only)
      Proc.new { |value| DateTime.strptime(value, @time_key_format) }
    end
  else
    Proc.new { |value| DateTime.parse(value) }
  end
end

#detect_es_major_version ⇒ Object 



197
198
199
200
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 197

def detect_es_major_version
  @_es_info ||= client.info
  @_es_info["version"]["number"].to_i
end

#expand_placeholders(metadata) ⇒ Object 



401
402
403
404
405
406
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 401

def expand_placeholders(metadata)
  logstash_prefix = extract_placeholders(@logstash_prefix, metadata)
  index_name = extract_placeholders(@index_name, metadata)
  type_name = extract_placeholders(@type_name, metadata)
  return logstash_prefix, index_name, type_name
end

#flatten_record(record, prefix = []) ⇒ Object 



386
387
388
389
390
391
392
393
394
395
396
397
398
399
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 386

def flatten_record(record, prefix=[])
  ret = {}
  if record.is_a? Hash
    record.each { |key, value|
      ret.merge! flatten_record(value, prefix + [key.to_s])
    }
  elsif record.is_a? Array
    # Don't mess with arrays, leave them unprocessed
    ret.merge!({prefix.join(@flatten_hashes_separator) => record})
  else
    return {prefix.join(@flatten_hashes_separator) => record}
  end
  ret
end

#get_connection_options ⇒ Object 



291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 291

def get_connection_options
  raise "`password` must be present if `user` is present" if @user && !@password

  hosts = if @hosts
    @hosts.split(',').map do |host_str|
      # Support legacy hosts format host:port,host:port,host:port...
      if host_str.match(%r{^[^:]+(\:\d+)?$})
        {
          host:   host_str.split(':')[0],
          port:   (host_str.split(':')[1] || @port).to_i,
          scheme: @scheme
        }
      else
        # New hosts format expects URLs such as http://logs.foo.com,https://john:[email protected]/elastic
        uri = URI(get_escaped_userinfo(host_str))
        %w(user password path).inject(host: uri.host, port: uri.port, scheme: uri.scheme) do |hash, key|
          hash[key.to_sym] = uri.public_send(key) unless uri.public_send(key).nil? || uri.public_send(key) == ''
          hash
        end
      end
    end.compact
  else
    [{host: @host, port: @port, scheme: @scheme}]
  end.each do |host|
    host.merge!(user: @user, password: @password) if !host[:user] && @user
    host.merge!(path: @path) if !host[:path] && @path
  end

  {
    hosts: hosts
  }
end

#get_escaped_userinfo(host_str) ⇒ Object 



279
280
281
282
283
284
285
286
287
288
289
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 279

def get_escaped_userinfo(host_str)
  if m = host_str.match(/(?<scheme>.*)%{(?<user>.*)}:%{(?<password>.*)}(?<path>@.*)/)
    m["scheme"] +
      URI.encode_www_form_component(m["user"]) +
      ':' +
      URI.encode_www_form_component(m["password"]) +
      m["path"]
  else
    host_str
  end
end

#get_parent_of(record, path) ⇒ Object

returns [parent, child_key] of child described by path array in record’s tree returns [nil, child_key] if path doesnt exist in record



528
529
530
531
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 528

def get_parent_of(record, path)
  parent_object = path[0..-2].reduce(record) { |a, e| a.is_a?(Hash) ? a[e] : nil }
  [parent_object, path[-1]]
end

#multi_workers_ready? ⇒ Boolean

Returns:

  • (Boolean) 


408
409
410
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 408

def multi_workers_ready?
  true
end

#parse_time(value, event_time, tag) ⇒ Object 



238
239
240
241
242
243
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 238

def parse_time(value, event_time, tag)
  @time_parser.call(value)
rescue => e
  router.emit_error_event(@time_parse_error_tag, Fluent::Engine.now, {'tag' => tag, 'time' => event_time, 'format' => @time_key_format, 'value' => value}, e)
  return Time.at(event_time).to_datetime
end

#process_message(tag, meta, header, time, record, bulk_message, extracted_values) ⇒ Object 



443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 443

def process_message(tag, meta, header, time, record, bulk_message, extracted_values)
  logstash_prefix, index_name, type_name = extracted_values

  if @flatten_hashes
    record = flatten_record(record)
  end

  if @hash_config
    record = generate_hash_id_key(record)
  end

  dt = nil
  if @logstash_format || @include_timestamp
    if record.has_key?(TIMESTAMP_FIELD)
      rts = record[TIMESTAMP_FIELD]
      dt = parse_time(rts, time, tag)
    elsif record.has_key?(@time_key)
      rts = record[@time_key]
      dt = parse_time(rts, time, tag)
      record[TIMESTAMP_FIELD] = dt.iso8601(@time_precision) unless @time_key_exclude_timestamp
    else
      dt = Time.at(time).to_datetime
      record[TIMESTAMP_FIELD] = dt.iso8601(@time_precision)
    end
  end

  target_index_parent, target_index_child_key = @target_index_key ? get_parent_of(record, @target_index_key) : nil
  if target_index_parent && target_index_parent[target_index_child_key]
    target_index = target_index_parent.delete(target_index_child_key)
  elsif @logstash_format
    dt = dt.new_offset(0) if @utc_index
    target_index = "#{logstash_prefix}#{@logstash_prefix_separator}#{dt.strftime(@logstash_dateformat)}"
  else
    target_index = index_name
  end

  # Change target_index to lower-case since Elasticsearch doesn't
  # allow upper-case characters in index names.
  target_index = target_index.downcase
  if @include_tag_key
    record[@tag_key] = tag
  end

  target_type_parent, target_type_child_key = @target_type_key ? get_parent_of(record, @target_type_key) : nil
  if target_type_parent && target_type_parent[target_type_child_key]
    target_type = target_type_parent.delete(target_type_child_key)
    if @last_seen_major_version == 6
      log.warn "Detected ES 6.x: `@type_name` will be used as the document `_type`."
      target_type = type_name
    elsif @last_seen_major_version >= 7
      log.warn "Detected ES 7.x or above: `_doc` will be used as the document `_type`."
      target_type = '_doc'.freeze
    end
  else
    if @last_seen_major_version >= 7 && target_type != DEFAULT_TYPE_NAME_ES_7x
      log.warn "Detected ES 7.x or above: `_doc` will be used as the document `_type`."
      target_type = '_doc'.freeze
    else
      target_type = type_name
    end
  end

  meta.clear
  meta["_index".freeze] = target_index
  meta["_type".freeze] = target_type

  if @pipeline
    meta["pipeline".freeze] = @pipeline
  end

  @meta_config_map.each do |record_accessor, meta_key|
    if raw_value = record_accessor.call(record)
      meta[meta_key] = raw_value
    end
  end

  if @remove_keys
    @remove_keys.each { |key| record.delete(key) }
  end

  append_record_to_messages(@write_operation, meta, header, record, bulk_message)
end

#remove_keys(record) ⇒ Object 



377
378
379
380
381
382
383
384
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 377

def remove_keys(record)
  keys = record[@remove_keys_on_update_key] || @remove_keys_on_update || []
  record.delete(@remove_keys_on_update_key)
  return record unless keys.any?
  record = record.dup
  keys.each { |key| record.delete(key) }
  record
end

#send_bulk(data, tag, chunk, bulk_message_count, extracted_values) ⇒ Object

send_bulk given a specific bulk request, the original tag, chunk, and bulk_message_count



535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 535

def send_bulk(data, tag, chunk, bulk_message_count, extracted_values)
  retries = 0
  begin

    log.on_trace { log.trace "bulk request: #{data}" }
    response = client.bulk body: data
    log.on_trace { log.trace "bulk response: #{response}" }

    if response['errors']
      error = Fluent::Plugin::ElasticsearchErrorHandler.new(self)
      error.handle_error(response, tag, chunk, bulk_message_count, extracted_values)
    end
  rescue RetryStreamError => e
    emit_tag = @retry_tag ? @retry_tag : tag
    router.emit_stream(emit_tag, e.retry_stream)
  rescue *client.transport.host_unreachable_exceptions => e
    if retries < 2
      retries += 1
      @_es = nil
      @_es_info = nil
      log.warn "Could not push logs to Elasticsearch, resetting connection and trying again. #{e.message}"
      sleep 2**retries
      retry
    end
    raise ConnectionFailure, "Could not push logs to Elasticsearch after #{retries} retries. #{e.message}"
  rescue Exception
    @_es = nil if @reconnect_on_error
    @_es_info = nil if @reconnect_on_error
    raise
  end
end

#update_body(record, op) ⇒ Object 



364
365
366
367
368
369
370
371
372
373
374
375
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 364

def update_body(record, op)
  update = remove_keys(record)
  body = {"doc".freeze => update}
  if op == UPSERT_OP
    if update == record
      body["doc_as_upsert".freeze] = true
    else
      body[UPSERT_OP] = record
    end
  end
  body
end

#write(chunk) ⇒ Object 



412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
 
# File 'lib/fluent/plugin/out_elasticsearch.rb', line 412

def write(chunk)
  bulk_message_count = 0
  bulk_message = ''
  header = {}
  meta = {}

  tag = chunk.metadata.tag
  extracted_values = expand_placeholders(chunk.metadata)
  @last_seen_major_version = detect_es_major_version rescue DEFAULT_ELASTICSEARCH_VERSION

  chunk.msgpack_each do |time, record|
    next unless record.is_a? Hash
    begin
      if process_message(tag, meta, header, time, record, bulk_message, extracted_values)
        bulk_message_count += 1
      else
        if @emit_error_for_missing_id
          raise MissingIdFieldError, "Missing '_id' field. Write operation is #{@write_operation}"
        else
          log.on_debug { log.debug("Dropping record because its missing an '_id' field and write_operation is #{@write_operation}: #{record}") }
        end
      end
    rescue => e
      router.emit_error_event(tag, time, record, e)
    end
  end

  send_bulk(bulk_message, tag, chunk, bulk_message_count, extracted_values) unless bulk_message.empty?
  bulk_message.clear
end