# Copyright 2016 Google LLC # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # https://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. require "monitor" require "concurrent" require "google/cloud/logging/errors" module Google module Cloud module Logging ## # # AsyncWriter # # AsyncWriter buffers, batches, and transmits log entries efficiently. # Writing log entries is asynchronous and will not block. # # Batches that cannot be delivered immediately are queued. When the queue # is full new batch requests will raise errors that can be consumed using # the {#on_error} callback. This provides back pressure in case the writer # cannot keep up with requests. # # This object is thread-safe; it may accept write requests from # multiple threads simultaneously, and will serialize them when # executing in the background thread. # # @example # require "google/cloud/logging" # # logging = Google::Cloud::Logging.new # # async = logging.async_writer # # entry1 = logging.entry payload: "Job started." # entry2 = logging.entry payload: "Job completed." # # labels = { job_size: "large", job_code: "red" } # resource = logging.resource "gae_app", # "module_id" => "1", # "version_id" => "20150925t173233" # # async.write_entries [entry1, entry2], # log_name: "my_app_log", # resource: resource, # labels: labels # class AsyncWriter include MonitorMixin ## # @private Implementation accessors attr_reader :logging, :max_bytes, :max_count, :interval, :threads, :max_queue, :partial_success ## # @private Creates a new AsyncWriter instance. def initialize logging, max_count: 10000, max_bytes: 10000000, max_queue: 100, interval: 5, threads: 10, partial_success: false # init MonitorMixin super() @logging = logging @max_count = max_count @max_bytes = max_bytes @max_queue = max_queue @interval = interval @threads = threads @partial_success = partial_success @error_callbacks = [] @cond = new_cond # Make sure all buffered messages are sent when process exits. at_exit { stop } end ## # Asynchronously write one or more log entries to the Stackdriver # Logging service. # # Unlike the main write_entries method, this method usually does not # block. The actual write RPCs will happen in the background, and may # be batched with related calls. However, if the queue is full, this # method will block until enough space has cleared out. # # @param [Google::Cloud::Logging::Entry, # Array] entries One or more entry # objects to write. The log entries must have values for all required # fields. # @param [String] log_name A default log ID for those log entries in # `entries` that do not specify their own `log_name`. See also # {Entry#log_name=}. # @param [Resource] resource A default monitored resource for those log # entries in entries that do not specify their own resource. See also # {Entry#resource}. # @param [Hash{Symbol,String => String}] labels User-defined `key:value` # items that are added to the `labels` field of each log entry in # `entries`, except when a log entry specifies its own `key:value` # item with the same key. See also {Entry#labels=}. # # @return [Google::Cloud::Logging::AsyncWriter] Returns self. # # @example # require "google/cloud/logging" # # logging = Google::Cloud::Logging.new # async = logging.async_writer # # entry = logging.entry payload: "Job started.", # log_name: "my_app_log" # entry.resource.type = "gae_app" # entry.resource.labels[:module_id] = "1" # entry.resource.labels[:version_id] = "20150925t173233" # # async.write_entries entry # def write_entries entries, log_name: nil, resource: nil, labels: nil synchronize do raise "AsyncWriter has been stopped" if @stopped Array(entries).each do |entry| # Update the entry to have all the data directly on it entry.log_name ||= log_name if entry.resource.nil? || entry.resource.empty? entry.resource = resource end entry.labels = labels if entry.labels.nil? || entry.labels.empty? # Add the entry to the batch @batch ||= Batch.new self next if @batch.try_add entry # If we can't add to the batch, publish and create a new batch publish_batch! @batch = Batch.new self @batch.add entry end @thread_pool ||= Concurrent::ThreadPoolExecutor.new \ max_threads: @threads, max_queue: @max_queue @thread ||= Thread.new { run_background } publish_batch! if @batch&.ready? @cond.broadcast end self end ## # Creates a logger instance that is API-compatible with Ruby's standard # library [Logger](http://ruby-doc.org/stdlib/libdoc/logger/rdoc). # # The logger will use AsyncWriter to transmit log entries on a # background thread. # # @param [String] log_name A log resource name to be associated with the # written log entries. # @param [Google::Cloud::Logging::Resource] resource The monitored # resource to be associated with written log entries. # @param [Hash] labels A set of user-defined data to be associated with # written log entries. # # @return [Google::Cloud::Logging::Logger] a Logger object that can be # used in place of a ruby standard library logger object. # # @example # require "google/cloud/logging" # # logging = Google::Cloud::Logging.new # # resource = logging.resource "gae_app", # module_id: "1", # version_id: "20150925t173233" # # async = logging.async_writer # logger = async.logger "my_app_log", resource, env: :production # logger.info "Job started." # def logger log_name, resource, labels = {} Logger.new self, log_name, resource, labels end ## # Begins the process of stopping the writer. Entries already in the # queue will be published, but no new entries can be added. Use {#wait!} # to block until the writer is fully stopped and all pending entries # have been published. # # @return [AsyncWriter] returns self so calls can be chained. def stop synchronize do break if @stopped @stopped = true publish_batch! @cond.broadcast @thread_pool.shutdown if @thread_pool end self end ## # Blocks until the writer is fully stopped, all pending entries have # been published, and all callbacks have completed. Does not stop the # writer. To stop the writer, first call {#stop} and then call {#wait!} # to block until the writer is stopped. # # @param [Number, nil] timeout The maximum number of seconds to wait for # shutdown to complete. Will wait forever when the value is `nil`. The # default value is `nil`. # # @return [AsyncWriter] returns self so calls can be chained. def wait! timeout = nil synchronize do if @thread_pool @thread_pool.shutdown @thread_pool.wait_for_termination timeout end end self end ## # Stop this asynchronous writer and block until it has been stopped. # # @param [Number, nil] timeout The maximum number of seconds to wait for # shutdown to complete. Will wait forever when the value is `nil`. The # default value is `nil`. # @param [Boolean] force If set to true, and the writer hasn't stopped # within the given timeout, kill it forcibly by terminating the # thread. This should be used with extreme caution, as it can # leave RPCs unfinished. Default is false. # # @return [Symbol] Returns `:new` if {#write_entries} has never been # called on the AsyncWriter, `:stopped` if it was already stopped # at the time of invocation, `:waited` if it stopped during the # timeout period, `:timeout` if it is still running after the # timeout, or `:forced` if it was forcibly killed. # def stop! timeout = nil, force: nil return :new unless @thread_pool return :stopped if stopped? stop wait! timeout if synchronize { @thread_pool.shutdown? } return :waited if timeout elsif force @thread_pool.kill return :forced end :timeout end alias async_stop! stop! ## # Forces all entries in the current batch to be published # immediately. # # @return [AsyncWriter] returns self so calls can be chained. def flush synchronize do publish_batch! @cond.broadcast end self end ## # Whether the writer has been started. # # @return [boolean] `true` when started, `false` otherwise. def started? !stopped? end ## # Whether the writer has been stopped. # # @return [boolean] `true` when stopped, `false` otherwise. def stopped? synchronize { @stopped } end ## # Register to be notified of errors when raised. # # If an unhandled error has occurred the writer will attempt to # recover from the error and resume buffering, batching, and # transmitting log entries # # Multiple error handlers can be added. # # @yield [callback] The block to be called when an error is raised. # @yieldparam [Exception] error The error raised. # # @example # require "google/cloud/logging" # require "google/cloud/error_reporting" # # logging = Google::Cloud::Logging.new # # resource = logging.resource "gae_app", # module_id: "1", # version_id: "20150925t173233" # # async = logging.async_writer # # # Register to be notified when unhandled errors occur. # async.on_error do |error| # # error can be a AsyncWriterError or AsyncWriteEntriesError # Google::Cloud::ErrorReporting.report error # end # # logger = async.logger "my_app_log", resource, env: :production # logger.info "Job started." # def on_error &block synchronize do @error_callbacks << block end end ## # The most recent unhandled error to occur while transmitting log # entries. # # If an unhandled error has occurred the subscriber will attempt to # recover from the error and resume buffering, batching, and # transmitting log entries. # # @return [Exception, nil] error The most recent error raised. # # @example # require "google/cloud/logging" # # logging = Google::Cloud::Logging.new # # resource = logging.resource "gae_app", # module_id: "1", # version_id: "20150925t173233" # # async = logging.async_writer # # logger = async.logger "my_app_log", resource, env: :production # logger.info "Job started." # # # If an error was raised, it can be retrieved here: # async.last_error #=> nil # def last_error synchronize { @last_error } end alias last_exception last_error protected def run_background synchronize do until @stopped if @batch.nil? @cond.wait next end if @batch.ready? # interval met, publish the batch... publish_batch! @cond.wait else # still waiting for the interval to publish the batch... @cond.wait @batch.publish_wait end end end end def publish_batch! return unless @batch batch_to_be_published = @batch @batch = nil publish_batch_async batch_to_be_published end # Sets the last_error and calls all error callbacks. def error! error error_callbacks = synchronize do @last_error = error @error_callbacks end error_callbacks = default_error_callbacks if error_callbacks.empty? error_callbacks.each { |error_callback| error_callback.call error } end def default_error_callbacks # This is memoized to reduce calls to the configuration. @default_error_callbacks ||= begin error_callback = Google::Cloud::Logging.configure.on_error error_callback ||= Google::Cloud.configure.on_error if error_callback [error_callback] else [] end end end def publish_batch_async batch Concurrent::Promises.future_on( @thread_pool, batch.entries ) do |entries| write_entries_with entries end rescue Concurrent::RejectedExecutionError => e async_error = AsyncWriterError.new( "Error writing entries: #{e.message}", batch.entries ) # Manually set backtrace so we don't have to raise async_error.set_backtrace caller error! async_error end def write_entries_with entries logging.write_entries entries, partial_success: partial_success rescue StandardError => e write_error = AsyncWriteEntriesError.new( "Error writing entries: #{e.message}", entries ) # Manually set backtrace so we don't have to raise write_error.set_backtrace caller error! write_error end ## # @private class Batch attr_reader :created_at, :entries def initialize writer @writer = writer @entries = [] @entries_bytes = 2 # initial size w/ partial_success @created_at = nil end def add entry, addl_bytes: nil addl_bytes ||= addl_bytes_for entry @entries << entry @entries_bytes += addl_bytes @created_at ||= Time.now nil end def try_add entry addl_bytes = addl_bytes_for entry new_message_count = @entries.count + 1 new_message_bytes = @entries_bytes + addl_bytes if new_message_count > @writer.max_count || new_message_bytes >= @writer.max_bytes return false end add entry, addl_bytes: addl_bytes true end def ready? @entries.count >= @writer.max_count || @entries_bytes >= @writer.max_bytes || (@created_at.nil? || (publish_at < Time.now)) end def publish_at return nil if @created_at.nil? @created_at + @writer.interval end def publish_wait publish_wait = publish_at - Time.now return 0 if publish_wait < 0 publish_wait end def addl_bytes_for entry entry.to_grpc.to_proto.bytesize + 2 end end end end end end