logstash-integration-jdbc 5.2.1 → 5.2.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5826d365fd1f0f71ef3562e40e51d68b9bf76585bbd536eb5868a56c5bb6aacb
-  data.tar.gz: 5ae08858db8043a0840d7a8578eae295e99b6a1da4ca3ec9239ee0730298188e
+  metadata.gz: ad77fc92ab07e1da7e4c6285075b6583f9f293c073199b614e3852cd22bbc6a9
+  data.tar.gz: 9e7b2c0f0c16930903bdee1f22b2651db7cf9f80d01a12f212b7a40155d4ebea
 SHA512:
-  metadata.gz: 685637b889432ff5378e6c9873a90b183a90337a011d9719c17b6928721fbef01e393b111f35d2d4d19e7a8bc62eeffd98d93964e46494f486b93f7a9b45bff3
-  data.tar.gz: 30f804b0e420b7485da98805e80902ae2d8b33417256938706ba79f0f20fe62d8da2bec9a6c716e50a30f3455f0cb86f6757eef9811e631685f3e620c8c6a604
+  metadata.gz: 2928a01bbe1ef418f36d6ddc7847416fc6d43278b20242926037bef02e8a429ba1d4587f1bd88f8d52de2ec1fef26e749aeb358de82a8bf69c36e5c3c9d42f8e
+  data.tar.gz: 8aae596826ec1d9155e0f8bdfad9a882e92fa2be610f89d7c67a7b3f44f0e6937bcbafd78900457d92b20a5a28188163f7b4f25040d8deb229f505ab6faf2c22

data/CHANGELOG.md

@@ -1,3 +1,12 @@
+## 5.2.4
+  - Fix: compatibility with all (>= 3.0) rufus-scheduler versions [#97](https://github.com/logstash-plugins/logstash-integration-jdbc/pull/97) 
+
+## 5.2.3
+  - Performance: avoid contention on scheduler execution [#103](https://github.com/logstash-plugins/logstash-integration-jdbc/pull/103)
+
+## 5.2.2
+  - Feat: name scheduler threads + redirect error logging [#102](https://github.com/logstash-plugins/logstash-integration-jdbc/pull/102)
+
 ## 5.2.1
   - Refactor: isolate paginated normal statement algorithm in a separate handler [#101](https://github.com/logstash-plugins/logstash-integration-jdbc/pull/101)
 

data/Gemfile

@@ -9,3 +9,5 @@ if Dir.exist?(logstash_path) && use_logstash_source
   gem 'logstash-core', :path => "#{logstash_path}/logstash-core"
   gem 'logstash-core-plugin-api', :path => "#{logstash_path}/logstash-core-plugin-api"
 end
+
+gem 'rufus-scheduler', ENV['RUFUS_SCHEDULER_VERSION'] if ENV['RUFUS_SCHEDULER_VERSION']

data/lib/logstash/filters/jdbc/loader_schedule.rb

@@ -4,60 +4,26 @@ require "rufus/scheduler"
 
 module LogStash module Filters module Jdbc
   class LoaderSchedule < Validatable
-    attr_reader :schedule_frequency, :loader_schedule
-
-    def to_log_string
-      message = ""
-      message.concat "these months in the year [#{@cronline.months.to_a.join(", ")}];" unless @cronline.months.nil?
-      message.concat "these days in the month [#{@cronline.days.to_a.join(", ")}];" unless @cronline.days.nil?
-      message.concat "these hours in the day [#{@cronline.hours.to_a.join(", ")}];" unless @cronline.hours.nil?
-      message.concat "these minutes in the hour [#{@cronline.minutes.to_a.join(", ")}];" unless @cronline.minutes.nil?
-      message.concat "these seconds in the minute [#{@cronline.seconds.to_a.join(", ")}]" unless @cronline.seconds.nil?
-      if !message.empty?
-        message.prepend "Scheduled for: "
-      end
-      message
-    end
+    attr_reader :loader_schedule
 
     private
 
-    def post_initialize
-      if valid?
-        # From the Rufus::Scheduler docs:
-        #   By default, rufus-scheduler sleeps 0.300 second between every step.
-        #   At each step it checks for jobs to trigger and so on.
-        # set the frequency to 2.5 seconds if we are not reloading in the seconds timeframe
-        # rufus scheduler thread should respond to stop quickly enough.
-        if only_seconds_set?
-          @schedule_frequency = 0.3
-        else
-          @schedule_frequency = 2.5
-        end
-      end
-    end
-
-
-    def only_seconds_set?
-      @cronline.seconds &&
-        @cronline.minutes.nil? &&
-        @cronline.hours.nil? &&
-        @cronline.days.nil? &&
-        @cronline.months.nil?
-    end
-
+    # @overload
     def parse_options
       @loader_schedule = @options
 
-      unless @loader_schedule.is_a?(String)
+      if @loader_schedule.is_a?(String)
+        begin
+          # Rufus::Scheduler 3.0 - 3.6 methods signature: parse_cron(o, opts)
+          # since Rufus::Scheduler 3.7 methods signature: parse_cron(o, opts={})
+          @cronline = Rufus::Scheduler.parse_cron(@loader_schedule, {})
+        rescue => e
+          @option_errors << "The loader_schedule option is invalid: #{e.message}"
+        end
+      else
         @option_errors << "The loader_schedule option must be a string"
       end
 
-      begin
-        @cronline = Rufus::Scheduler::CronLine.new(@loader_schedule)
-      rescue => e
-        @option_errors << "The loader_schedule option is invalid: #{e.message}"
-      end
-
       @valid = @option_errors.empty?
     end
   end

data/lib/logstash/filters/jdbc/repeating_load_runner.rb

@@ -3,8 +3,6 @@ require_relative "single_load_runner"
 
 module LogStash module Filters module Jdbc
   class RepeatingLoadRunner < SingleLoadRunner
-   # info - attr_reader :local, :loaders, :preloaders
-
     def repeated_load
       local.repopulate_all(loaders)
       @reload_counter.increment

data/lib/logstash/filters/jdbc/single_load_runner.rb

@@ -26,10 +26,6 @@ module LogStash module Filters module Jdbc
     def repeated_load
     end
 
-    def call
-      repeated_load
-    end
-
     def reload_count
       @reload_counter.value
     end

data/lib/logstash/filters/jdbc_static.rb

@@ -3,6 +3,7 @@ require "logstash-integration-jdbc_jars"
 require "logstash/filters/base"
 require "logstash/namespace"
 require "logstash/plugin_mixins/ecs_compatibility_support"
+require "logstash/plugin_mixins/jdbc/scheduler"
 require_relative "jdbc/loader"
 require_relative "jdbc/loader_schedule"
 require_relative "jdbc/repeating_load_runner"
@@ -191,17 +192,15 @@ module LogStash module Filters class JdbcStatic < LogStash::Filters::Base
     @processor = Jdbc::LookupProcessor.new(@local_lookups, global_lookup_options)
     runner_args.unshift(@processor.local)
     if @loader_schedule
-      args = []
       @loader_runner = Jdbc::RepeatingLoadRunner.new(*runner_args)
       @loader_runner.initial_load
-      cronline = Jdbc::LoaderSchedule.new(@loader_schedule)
-      cronline.to_log_string.tap do |msg|
-        logger.info("Scheduler operations: #{msg}") unless msg.empty?
+      @scheduler = LogStash::PluginMixins::Jdbc::Scheduler.
+          start_cron_scheduler(@loader_schedule, thread_name: "[#{id}]-jdbc_static__scheduler") { @loader_runner.repeated_load }
+      cron_job = @scheduler.cron_jobs.first
+      if cron_job
+        frequency = cron_job.respond_to?(:rough_frequency) ? cron_job.rough_frequency : cron_job.frequency
+        logger.info("Loaders will execute every #{frequency} seconds", loader_schedule: @loader_schedule)
       end
-      logger.info("Scheduler scan for work frequency is: #{cronline.schedule_frequency}")
-      rufus_args = {:max_work_threads => 1, :frequency => cronline.schedule_frequency}
-      @scheduler = Rufus::Scheduler.new(rufus_args)
-      @scheduler.cron(cronline.loader_schedule, @loader_runner)
     else
       @loader_runner = Jdbc::SingleLoadRunner.new(*runner_args)
       @loader_runner.initial_load

data/lib/logstash/inputs/jdbc.rb

@@ -3,6 +3,7 @@ require "logstash/inputs/base"
 require "logstash/namespace"
 require "logstash/plugin_mixins/jdbc/common"
 require "logstash/plugin_mixins/jdbc/jdbc"
+require "logstash/plugin_mixins/jdbc/scheduler"
 require "logstash/plugin_mixins/ecs_compatibility_support"
 require "logstash/plugin_mixins/ecs_compatibility_support/target_check"
 require "logstash/plugin_mixins/validator_support/field_reference_validation_adapter"
@@ -293,11 +294,9 @@ module LogStash module Inputs class Jdbc < LogStash::Inputs::Base
   def run(queue)
     load_driver
     if @schedule
-      @scheduler = Rufus::Scheduler.new(:max_work_threads => 1)
-      @scheduler.cron @schedule do
-        execute_query(queue)
-      end
-
+      # input thread (Java) name example "[my-oracle]<jdbc"
+      @scheduler = LogStash::PluginMixins::Jdbc::Scheduler.
+          start_cron_scheduler(@schedule, thread_name: "[#{id}]<jdbc__scheduler") { execute_query(queue) }
       @scheduler.join
     else
       execute_query(queue)

data/lib/logstash/plugin_mixins/jdbc/jdbc.rb

@@ -220,7 +220,8 @@ module LogStash  module PluginMixins module Jdbc
         end
         success = true
       rescue Sequel::DatabaseConnectionError, Sequel::DatabaseError, Java::JavaSql::SQLException => e
-        details = { :exception => e.message }
+        details = { exception: e.class, message: e.message }
+        details[:cause] = e.cause.inspect if e.cause
         details[:backtrace] = e.backtrace if @logger.debug?
         @logger.warn("Exception when executing JDBC query", details)
       else

data/lib/logstash/plugin_mixins/jdbc/scheduler.rb

@@ -0,0 +1,175 @@
+require 'rufus/scheduler'
+
+require 'logstash/util/loggable'
+
+module LogStash module PluginMixins module Jdbc
+  class Scheduler < Rufus::Scheduler
+
+    include LogStash::Util::Loggable
+
+    # Rufus::Scheduler >= 3.4 moved the Time impl into a gem EoTime = ::EtOrbi::EoTime`
+    # Rufus::Scheduler 3.1 - 3.3 using it's own Time impl `Rufus::Scheduler::ZoTime`
+    TimeImpl = defined?(Rufus::Scheduler::EoTime) ? Rufus::Scheduler::EoTime :
+                   (defined?(Rufus::Scheduler::ZoTime) ? Rufus::Scheduler::ZoTime : ::Time)
+
+    # @param cron [String] cron-line
+    # @param opts [Hash] scheduler options
+    # @return scheduler instance
+    def self.start_cron_scheduler(cron, opts = {}, &block)
+      unless block_given?
+        raise ArgumentError, 'missing (cron scheduler) block - worker task to execute'
+      end
+      scheduler = new_scheduler(opts)
+      scheduler.schedule_cron(cron, &block)
+      scheduler
+    end
+
+    # @param opts [Hash] scheduler options
+    # @return scheduler instance
+    def self.new_scheduler(opts)
+      unless opts.key?(:thread_name)
+        raise ArgumentError, 'thread_name: option is required to be able to distinguish multiple scheduler threads'
+      end
+      opts[:max_work_threads] ||= 1
+      # amount the scheduler thread sleeps between checking whether jobs
+      # should trigger, default is 0.3 which is a bit too often ...
+      # in theory the cron expression '* * * * * *' supports running jobs
+      # every second but this is very rare, we could potentially go higher
+      opts[:frequency] ||= 1.0
+
+      new(opts)
+    end
+
+    # @overload
+    def timeout_jobs
+      # Rufus relies on `Thread.list` which is a blocking operation and with many schedulers
+      # (and threads) within LS will have a negative impact on performance as scheduler
+      # threads will end up waiting to obtain the `Thread.list` lock.
+      #
+      # However, this isn't necessary we can easily detect whether there are any jobs
+      # that might need to timeout: only when `@opts[:timeout]` is set causes worker thread(s)
+      # to have a `Thread.current[:rufus_scheduler_timeout]` that is not nil
+      return unless @opts[:timeout]
+      super
+    end
+
+    # @overload
+    def work_threads(query = :all)
+      if query == :__all_no_cache__ # special case from JobDecorator#start_work_thread
+        @_work_threads = nil # when a new worker thread is being added reset
+        return super(:all)
+      end
+
+      # Gets executed every time a job is triggered, we're going to cache the
+      # worker threads for this scheduler (to avoid `Thread.list`) - they only
+      # change when a new thread is being started from #start_work_thread ...
+      work_threads = @_work_threads
+      if work_threads.nil?
+        work_threads = threads.select { |t| t[:rufus_scheduler_work_thread] }
+        @_work_threads = work_threads
+      end
+
+      case query
+      when :active then work_threads.select { |t| t[:rufus_scheduler_job] }
+      when :vacant then work_threads.reject { |t| t[:rufus_scheduler_job] }
+      else work_threads
+      end
+    end
+
+    # @overload
+    def on_error(job, err)
+      details = { exception: err.class, message: err.message, backtrace: err.backtrace }
+      details[:cause] = err.cause if err.cause
+
+      details[:now] = debug_format_time(TimeImpl.now)
+      details[:last_time] = (debug_format_time(job.last_time) rescue nil)
+      details[:next_time] = (debug_format_time(job.next_time) rescue nil)
+      details[:job] = job
+
+      details[:opts] = @opts
+      details[:started_at] = started_at
+      details[:thread] = thread.inspect
+      details[:jobs_size] = @jobs.size
+      details[:work_threads_size] = work_threads.size
+      details[:work_queue_size] = work_queue.size
+
+      logger.error("Scheduler intercepted an error:", details)
+
+    rescue => e
+      logger.error("Scheduler failed in #on_error #{e.inspect}")
+    end
+
+    def debug_format_time(time)
+      # EtOrbi::EoTime used by (newer) Rufus::Scheduler has to_debug_s https://git.io/JyiPj
+      time.respond_to?(:to_debug_s) ? time.to_debug_s : time.strftime("%Y-%m-%dT%H:%M:%S.%L")
+    end
+    private :debug_format_time
+
+    # @private helper used by JobDecorator
+    def work_thread_name_prefix
+      ( @opts[:thread_name] || "#{@thread_key}_scheduler" ) + '_worker-'
+    end
+
+    protected
+
+    # @overload
+    def start
+      ret = super() # @thread[:name] = @opts[:thread_name] || "#{@thread_key}_scheduler"
+
+      # at least set thread.name for easier thread dump analysis
+      if @thread.is_a?(Thread) && @thread.respond_to?(:name=)
+        @thread.name = @thread[:name] if @thread[:name]
+      end
+
+      ret
+    end
+
+    # @overload
+    def do_schedule(job_type, t, callable, opts, return_job_instance, block)
+      job_or_id = super
+
+      job_or_id.extend JobDecorator if return_job_instance
+
+      job_or_id
+    end
+
+    module JobDecorator
+
+      def start_work_thread
+        prev_thread_count = @scheduler.work_threads.size
+
+        ret = super() # does not return Thread instance in 3.0
+
+        work_threads = @scheduler.work_threads(:__all_no_cache__)
+        while prev_thread_count == work_threads.size # very unlikely
+          Thread.pass
+          work_threads = @scheduler.work_threads(:__all_no_cache__)
+        end
+
+        work_thread_name_prefix = @scheduler.work_thread_name_prefix
+
+        work_threads.sort! do |t1, t2|
+          if t1[:name].nil?
+            t2[:name].nil? ? 0 : +1 # nils at the end
+          elsif t2[:name].nil?
+            t1[:name].nil? ? 0 : -1
+          else
+            t1[:name] <=> t2[:name]
+          end
+        end
+
+        work_threads.each_with_index do |thread, i|
+          unless thread[:name]
+            thread[:name] = "#{work_thread_name_prefix}#{sprintf('%02i', i)}"
+            thread.name = thread[:name] if thread.respond_to?(:name=)
+            # e.g. "[oracle]<jdbc_scheduler_worker-00"
+          end
+        end
+
+        ret
+      end
+
+    end
+
+  end
+end end end

data/lib/logstash/plugin_mixins/jdbc/value_tracking.rb

@@ -6,9 +6,6 @@ module LogStash module PluginMixins module Jdbc
 
     def self.build_last_value_tracker(plugin)
       handler = plugin.record_last_run ? FileHandler.new(plugin.last_run_metadata_path) : NullFileHandler.new(plugin.last_run_metadata_path)
-      if plugin.record_last_run
-        handler = FileHandler.new(plugin.last_run_metadata_path)
-      end
       if plugin.clean_run
         handler.clean
       end

data/logstash-integration-jdbc.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name = 'logstash-integration-jdbc'
-  s.version         = '5.2.1'
+  s.version         = '5.2.4'
   s.licenses = ['Apache License (2.0)']
   s.summary         = "Integration with JDBC - input and filter plugins"
   s.description     = "This gem is a Logstash plugin required to be installed on top of the Logstash core pipeline using $LS_HOME/bin/logstash-plugin install gemname. This gem is not a stand-alone program"
@@ -34,8 +34,9 @@ Gem::Specification.new do |s|
 
   s.add_runtime_dependency 'tzinfo'
   s.add_runtime_dependency 'tzinfo-data'
-  # 3.5 limitation is required for jdbc-static loading schedule
-  s.add_runtime_dependency 'rufus-scheduler', '< 3.5'
+  # plugin maintains compatibility with < 3.5 (3.0.9)
+  # but works with newer rufus-scheduler >= 3.5 as well
+  s.add_runtime_dependency 'rufus-scheduler'
   s.add_runtime_dependency 'logstash-mixin-ecs_compatibility_support', '~>1.3'
   s.add_runtime_dependency "logstash-mixin-validator_support", '~> 1.0'
   s.add_runtime_dependency "logstash-mixin-event_support", '~> 1.0'

data/spec/filters/jdbc/repeating_load_runner_spec.rb

@@ -17,7 +17,7 @@ module LogStash module Filters module Jdbc
         expect(local_db).to receive(:populate_all).once.with(loaders)
         expect(local_db).to receive(:repopulate_all).once.with(loaders)
         runner.initial_load
-        subject.call
+        subject.repeated_load
       end
     end
   end

data/spec/inputs/integration/integ_spec.rb

@@ -70,7 +70,7 @@ describe LogStash::Inputs::Jdbc, :integration => true do
       plugin.register
       expect( plugin ).to receive(:log_java_exception)
       expect(plugin.logger).to receive(:warn).once.with("Exception when executing JDBC query",
-                                                        hash_including(:exception => instance_of(String)))
+                                                        hash_including(:message => instance_of(String)))
       q = Queue.new
       expect{ plugin.run(q) }.not_to raise_error
     end

data/spec/plugin_mixins/jdbc/scheduler_spec.rb

@@ -0,0 +1,78 @@
+# encoding: utf-8
+require "logstash/devutils/rspec/spec_helper"
+require "logstash/plugin_mixins/jdbc/scheduler"
+
+describe LogStash::PluginMixins::Jdbc::Scheduler do
+
+  let(:thread_name) { '[test]<jdbc_scheduler' }
+
+  let(:opts) do
+    { :max_work_threads => 2, :thread_name => thread_name }
+  end
+
+  subject(:scheduler) { LogStash::PluginMixins::Jdbc::Scheduler.new(opts) }
+
+  after { scheduler.stop(:wait) }
+
+  it "sets scheduler thread name" do
+    expect( scheduler.thread.name ).to include thread_name
+  end
+
+  context 'cron schedule' do
+
+    before do
+      scheduler.schedule_cron('* * * * * *') { sleep 1.25 } # every second
+    end
+
+    it "sets worker thread names" do
+      sleep 3.0
+      threads = scheduler.work_threads
+      threads.sort! { |t1, t2| (t1.name || '') <=> (t2.name || '') }
+
+      expect( threads.size ).to eql 2
+      expect( threads.first.name ).to eql "#{thread_name}_worker-00"
+      expect( threads.last.name ).to eql "#{thread_name}_worker-01"
+    end
+
+  end
+
+  context 'every 1s' do
+
+    before do
+      scheduler.schedule_in('1s') { raise 'TEST' } # every second
+    end
+
+    it "logs errors handled" do
+      expect( scheduler.logger ).to receive(:error).with /Scheduler intercepted an error/, hash_including(:message => 'TEST')
+      sleep 1.5
+    end
+
+  end
+
+  context 'work threads' do
+
+    let(:opts) { super().merge :max_work_threads => 3 }
+
+    let(:counter) { java.util.concurrent.atomic.AtomicLong.new(0) }
+
+    before do
+      scheduler.schedule_cron('* * * * * *') { counter.increment_and_get; sleep 3.25 } # every second
+    end
+
+    it "are working" do
+      sleep(0.05) while counter.get == 0
+      expect( scheduler.work_threads.size ).to eql 1
+      sleep(0.05) while counter.get == 1
+      expect( scheduler.work_threads.size ).to eql 2
+      sleep(0.05) while counter.get == 2
+      expect( scheduler.work_threads.size ).to eql 3
+
+      sleep 1.25
+      expect( scheduler.work_threads.size ).to eql 3
+      sleep 1.25
+      expect( scheduler.work_threads.size ).to eql 3
+    end
+
+  end
+
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: logstash-integration-jdbc
 version: !ruby/object:Gem::Version
-  version: 5.2.1
+  version: 5.2.4
 platform: ruby
 authors:
 - Elastic
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-12-22 00:00:00.000000000 Z
+date: 2022-03-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement
@@ -131,17 +131,17 @@ dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - "<"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '3.5'
+        version: '0'
   name: rufus-scheduler
   prerelease: false
   type: :runtime
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - "<"
+    - - ">="
       - !ruby/object:Gem::Version
-        version: '3.5'
+        version: '0'
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement
     requirements:
@@ -278,6 +278,7 @@ files:
 - lib/logstash/plugin_mixins/jdbc/checked_count_logger.rb
 - lib/logstash/plugin_mixins/jdbc/common.rb
 - lib/logstash/plugin_mixins/jdbc/jdbc.rb
+- lib/logstash/plugin_mixins/jdbc/scheduler.rb
 - lib/logstash/plugin_mixins/jdbc/statement_handler.rb
 - lib/logstash/plugin_mixins/jdbc/value_tracking.rb
 - lib/logstash/plugin_mixins/jdbc_streaming.rb
@@ -306,6 +307,7 @@ files:
 - spec/helpers/derbyrun.jar
 - spec/inputs/integration/integ_spec.rb
 - spec/inputs/jdbc_spec.rb
+- spec/plugin_mixins/jdbc/scheduler_spec.rb
 - spec/plugin_mixins/jdbc_streaming/parameter_handler_spec.rb
 - vendor/jar-dependencies/org/apache/derby/derby/10.14.1.0/derby-10.14.1.0.jar
 - vendor/jar-dependencies/org/apache/derby/derbyclient/10.14.1.0/derbyclient-10.14.1.0.jar
@@ -358,4 +360,5 @@ test_files:
 - spec/helpers/derbyrun.jar
 - spec/inputs/integration/integ_spec.rb
 - spec/inputs/jdbc_spec.rb
+- spec/plugin_mixins/jdbc/scheduler_spec.rb
 - spec/plugin_mixins/jdbc_streaming/parameter_handler_spec.rb