Class: LogStash::Outputs::AmazonES
- Inherits:
-
Base
- Object
- Base
- LogStash::Outputs::AmazonES
- Includes:
- Stud::Buffer
- Defined in:
- lib/logstash/outputs/amazon_es.rb
Overview
This output plugin emits data to Amazon Elasticsearch with support for signing requests using AWS V4 Signatures
The configuration and experience is similar to logstash-output-elasticsearch plugin and we have added Signature V4 support for the same Some of the default configurations like connection timeouts have been tuned for optimal performance with Amazon Elasticsearch
Retry Policy
This plugin uses the same retry policy as logstash-output-elasticsearch, It uses bulk API to optimize its imports into Elasticsearch.. These requests may experience either partial or total failures. Events are retried if they fail due to either a network error or the status codes 429 (the server is busy), 409 (Version Conflict), or 503 (temporary overloading/maintenance).
The retry policy’s logic can be described as follows:
-
Block and retry all events in the bulk response that experience transient network exceptions until a successful submission is received by Elasticsearch.
-
Retry the subset of sent events which resulted in ES errors of a retryable nature.
-
Events which returned retryable error codes will be pushed onto a separate queue for retrying events. Events in this queue will be retried a maximum of 5 times by default (configurable through :max_retries). The size of this queue is capped by the value set in :retry_max_items.
-
Events from the retry queue are submitted again when the queue reaches its max size or when the max interval time is reached. The max interval time is configurable via :retry_max_interval.
-
Events which are not retryable or have reached their max retry count are logged to stderr.
Constant Summary collapse
- RETRYABLE_CODES =
[409, 429, 503]
- SUCCESS_CODES =
[200, 201]
Instance Attribute Summary collapse
-
#client ⇒ Object
readonly
Returns the value of attribute client.
Instance Method Summary collapse
- #flush(actions, teardown = false) ⇒ Object
- #get_template ⇒ Object
- #receive(event) ⇒ Object
- #register ⇒ Object
-
#submit(actions) ⇒ Object
The submit method can be called from both the Stud::Buffer flush thread and from our own retry thread.
- #teardown ⇒ Object
Instance Attribute Details
#client ⇒ Object (readonly)
Returns the value of attribute client.
39 40 41 |
# File 'lib/logstash/outputs/amazon_es.rb', line 39 def client @client end |
Instance Method Details
#flush(actions, teardown = false) ⇒ Object
349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 |
# File 'lib/logstash/outputs/amazon_es.rb', line 349 def flush(actions, teardown = false) begin submit(actions) rescue Manticore::SocketException => e # If we can't even connect to the server let's just print out the URL (:hosts is actually a URL) # and let the user sort it out from there @logger.error( "Attempted to send a bulk request to Elasticsearch configured at '#{@client.[:hosts]}',"+ " but Elasticsearch appears to be unreachable or down!", :client_config => @client., :error_message => e. ) @logger.debug("Failed actions for last bad bulk request!", :actions => actions) rescue => e # For all other errors print out full connection issues @logger.error( "Attempted to send a bulk request to Elasticsearch configured at '#{@client.[:hosts]}'," + " but an error occurred and it failed! Are you sure you can reach elasticsearch from this machine using " + "the configuration provided?", :client_config => @client., :error_message => e., :error_class => e.class.name, :backtrace => e.backtrace ) @logger.debug("Failed actions for last bad bulk request!", :actions => actions) raise e end end |
#get_template ⇒ Object
267 268 269 270 271 272 273 274 275 276 277 278 |
# File 'lib/logstash/outputs/amazon_es.rb', line 267 def get_template if @template.nil? @template = ::File.('amazon_es/elasticsearch-template.json', ::File.dirname(__FILE__)) if !File.exists?(@template) raise "You must specify 'template => ...' in your elasticsearch output (I looked for '#{@template}')" end end template_json = IO.read(@template).gsub(/\n/,'') template = LogStash::Json.load(template_json) @logger.info("Using mapping template", :template => template) return template end |
#receive(event) ⇒ Object
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 |
# File 'lib/logstash/outputs/amazon_es.rb', line 281 def receive(event) return unless output?(event) # block until we have not maxed out our # retry queue. This is applying back-pressure # to slow down the receive-rate @retry_flush_mutex.synchronize { @retry_queue_not_full.wait(@retry_flush_mutex) while @retry_queue.size > @retry_max_items } event.set('[@metadata][retry_count]', 0) # Set the 'type' value for the index. type = if @document_type event.sprintf(@document_type) elsif @index_type # deprecated event.sprintf(@index_type) else event.get('type') || 'logs' end params = { :_id => @document_id ? event.sprintf(@document_id) : nil, :_index => event.sprintf(@index), :_type => type, :_routing => @routing ? event.sprintf(@routing) : nil } params[:_upsert] = LogStash::Json.load(event.sprintf(@upsert)) if @action == 'update' && @upsert != "" buffer_receive([event.sprintf(@action), params, event]) end |
#register ⇒ Object
194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 |
# File 'lib/logstash/outputs/amazon_es.rb', line 194 def register @hosts = Array(@hosts) # retry-specific variables @retry_flush_mutex = Mutex.new @retry_teardown_requested = Concurrent::AtomicBoolean.new(false) # needs flushing when interval @retry_queue_needs_flushing = ConditionVariable.new @retry_queue_not_full = ConditionVariable.new @retry_queue = Queue.new @submit_mutex = Mutex.new client_settings = {} = { :client_settings => client_settings } client_settings[:path] = "/#{@path}/".gsub(/\/+/, "/") # Normalize slashes @logger.debug? && @logger.debug("Normalizing http path", :path => @path, :normalized => client_settings[:path]) if @hosts.nil? || @hosts.empty? @logger.info("No 'host' set in elasticsearch output. Defaulting to localhost") @hosts = ["localhost"] end client_settings.merge! setup_proxy() .merge! setup_basic_auth() # Update API setup = { :upsert => @upsert, :doc_as_upsert => @doc_as_upsert } .merge! if @action == 'update' @client = LogStash::Outputs::AES::HttpClient.new( .merge(:hosts => @hosts, :port => @port, :region => @region, :aws_access_key_id => @aws_access_key_id, :aws_secret_access_key => @aws_secret_access_key,:protocol => @protocol) ) if @manage_template begin @logger.info("Automatic template management enabled", :manage_template => @manage_template.to_s) @client.template_install(@template_name, get_template, @template_overwrite) rescue => e @logger.error("Failed to install template: #{e.}") end end @logger.info("New Elasticsearch output", :hosts => @hosts, :port => @port) @client_idx = 0 buffer_initialize( :max_items => @flush_size, :max_interval => @idle_flush_time, :logger => @logger ) @retry_timer_thread = Thread.new do loop do sleep(@retry_max_interval) @retry_flush_mutex.synchronize { @retry_queue_needs_flushing.signal } end end @retry_thread = Thread.new do while @retry_teardown_requested.false? @retry_flush_mutex.synchronize { @retry_queue_needs_flushing.wait(@retry_flush_mutex) } retry_flush end end end |
#submit(actions) ⇒ Object
The submit method can be called from both the Stud::Buffer flush thread and from our own retry thread.
317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 |
# File 'lib/logstash/outputs/amazon_es.rb', line 317 def submit(actions) @submit_mutex.synchronize do es_actions = actions.map { |a, doc, event| [a, doc, event.to_hash] } bulk_response = @client.bulk(es_actions) if bulk_response["errors"] && bulk_response["items"] actions_to_retry = [] bulk_response['items'].each_with_index do |item,idx| action = es_actions[idx] action_type, props = item.first # These are all hashes with one value, so we destructure them here status = props['status'] error = props['error'] if RETRYABLE_CODES.include?(status) @logger.warn "retrying failed action with response code: #{status}" actions_to_retry << action elsif not SUCCESS_CODES.include?(status) @logger.warn "failed action", status: status, error: error, action: action end end retry_push(actions_to_retry) unless actions_to_retry.empty? end end end |
#teardown ⇒ Object
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 |
# File 'lib/logstash/outputs/amazon_es.rb', line 381 def teardown @retry_teardown_requested.make_true # First, make sure retry_timer_thread is stopped # to ensure we do not signal a retry based on # the retry interval. Thread.kill(@retry_timer_thread) @retry_timer_thread.join # Signal flushing in the case that #retry_flush is in # the process of waiting for a signal. @retry_flush_mutex.synchronize { @retry_queue_needs_flushing.signal } # Now, #retry_flush is ensured to not be in a state of # waiting and can be safely joined into the main thread # for further final execution of an in-process remaining call. @retry_thread.join # execute any final actions along with a proceeding retry for any # final actions that did not succeed. buffer_flush(:final => true) retry_flush end |