logstash-filter-aggregate 2.8.0 → 2.9.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8cbb27a58c0339ae9f7908d18250a5f91c456f6f
-  data.tar.gz: e86b38aa410918fb62cf7050be5fd8b9f44d6ed9
+  metadata.gz: 468412047e7db18515ba09a4a112753019e7bcb1
+  data.tar.gz: db82a1a0b681c38c600d72a8bc0a9d8215013437
 SHA512:
-  metadata.gz: d14e1e14cdf0342db92e06ceac124f4c0d384183e4074f7da55b87e27d117d4feddc63844b3a7b06b196d704154a44005bae18e8b3f135d8008f877ef6bafe76
-  data.tar.gz: 090d1e070fab264a8f4b1c7150e6d4161e357ab146cc0303480cc6ee0aebdcba80659f9a7dba02ffadf0939f47140a7038f83aa571b9053e2cd9d1fb58a2b91f
+  metadata.gz: 4b5ac8ba6307a2e41d582655b34ecd4d2c2da808ac0f49dc332107676631eb32155e19dd3c870783e06babd39f55a7f2e5346d3c1d565d3a4de5342a0969452e
+  data.tar.gz: fd827e4944b5f30dd3b53561a3914668911861584f125302136179d4d27d4cfb66f957438994b6f42c188eb778264da5f5c46452ff216ddba6d66d9566536512

data/CHANGELOG.md

@@ -1,9 +1,16 @@
+## 2.9.0
+  - new feature: add ability to dynamically define a custom `timeout` or `inactivity_timeout` in `code` block (fix issues [#91](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/91) and [#92](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/92))
+  - new feature: add meta informations available in `code` block through `map_meta` variable
+  - new feature: add Logstash metrics, specific to aggregate plugin: aggregate_maps, pushed_events, task_timeouts, code_errors, timeout_code_errors
+  - new feature: validate at startup that `map_action` option equals to 'create', 'update' or 'create_or_update'
+
 ## 2.8.0
-  - new feature: add 'timeout_timestamp_field' option.  
-    When set, this option lets to compute timeout based on event timestamp field (and not system time). It's particularly useful when processing old logs.
+  - new feature: add 'timeout_timestamp_field' option (fix issue [#81](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/81))  
+    When set, this option lets to compute timeout based on event timestamp field (and not system time).  
+    It's particularly useful when processing old logs.
 
 ## 2.7.2
-  - bugfix: fix synchronisation issue at Logstash shutdown (#75)
+  - bugfix: fix synchronisation issue at Logstash shutdown (issue [#75](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/75))
 
 ## 2.7.1
   - docs: update gemspec summary
@@ -32,7 +39,7 @@
   Events for a given `task_id` will be aggregated for as long as they keep arriving within the defined `inactivity_timeout` option - the inactivity timeout is reset each time a new event happens. On the contrary, `timeout` is never reset and happens after `timeout` seconds since aggregation map creation.
 
 ## 2.5.2
-- bugfix: fix 'aggregate_maps_path' load (issue #62). Re-start of Logstash died when no data were provided in 'aggregate_maps_path' file for some aggregate task_id patterns
+- bugfix: fix 'aggregate_maps_path' load (issue [#62](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/62)). Re-start of Logstash died when no data were provided in 'aggregate_maps_path' file for some aggregate task_id patterns
 - enhancement: at Logstash startup, check that 'task_id' option contains a field reference expression (else raise error)
 - docs: enhance examples
 - docs: precise that tasks are tied to their task_id pattern, even if they have same task_id value
@@ -50,7 +57,7 @@
  - breaking: need Logstash 2.4 or later  
 
 ## 2.4.0
- - new feature: You can now define timeout options per task_id pattern (#42)  
+ - new feature: You can now define timeout options per task_id pattern (fix issue [#42](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/42))  
  timeout options are : `timeout, timeout_code, push_map_as_event_on_timeout, push_previous_map_as_event, timeout_task_id_field, timeout_tags`
  - validation: a configuration error is thrown at startup if you define any timeout option on several aggregate filters for the same task_id pattern
  - breaking: if you use `aggregate_maps_path` option, storage format has changed. So you have to delete `aggregate_maps_path` file before starting Logstash
@@ -84,14 +91,14 @@
  - internal,deps: New dependency requirements for logstash-core for the 5.0 release
 
 ## 2.0.3
- - bugfix: fix issue #10 : numeric task_id is now well processed
+ - bugfix: fix issue [#10](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/10) : numeric task_id is now well processed
 
 ## 2.0.2
- - bugfix: fix issue #5 : when code call raises an exception, the error is logged and the event is tagged '_aggregateexception'. It avoids logstash crash.
+ - bugfix: fix issue [#5](https://github.com/logstash-plugins/logstash-filter-aggregate/issues/5) : when code call raises an exception, the error is logged and the event is tagged '_aggregateexception'. It avoids logstash crash.
 
 ## 2.0.0
- - internal: Plugins were updated to follow the new shutdown semantic, this mainly allows Logstash to instruct input plugins to terminate gracefully, 
-   instead of using Thread.raise on the plugins' threads. Ref: https://github.com/elastic/logstash/pull/3895
+ - internal: Plugins were updated to follow the new shutdown semantic, this mainly allows Logstash to instruct input plugins to terminate gracefully, instead of using Thread.raise on the plugins' threads.  
+   Ref: https://github.com/elastic/logstash/pull/3895
  - internal,deps: Dependency on logstash-core update to 2.0
 
 ## 0.1.3

data/docs/index.asciidoc

@@ -397,11 +397,20 @@ Example:
   * Value type is <<string,string>>
   * There is no default value for this setting.
 
-The code to execute to update map, using current event.
+The code to execute to update aggregated map, using current event.
 
-Or on the contrary, the code to execute to update event, using current map.
+Or on the contrary, the code to execute to update event, using aggregated map.
 
-You will have a 'map' variable and an 'event' variable available (that is the event itself).
+Available variables are :
+
+`event`: current Logstash event
+
+`map`: aggregated map associated to `task_id`, containing key/value pairs. Data structure is a ruby http://ruby-doc.org/core-1.9.1/Hash.html[Hash]
+
+`map_meta`: meta informations associated to aggregate map. It allows to set a custom `timeout` or `inactivity_timeout`. 
+It allows also to get `creation_timestamp`, `lastevent_timestamp` and `task_id`.
+
+When option push_map_as_event_on_timeout=true, if you set `map_meta.timeout=0` in `code` block, then aggregated map is immediately pushed as a new event.
 
 Example:
 [source,ruby]

data/lib/logstash/filters/aggregate.rb

@@ -20,7 +20,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
 
   config :code, :validate => :string, :required => true
 
-  config :map_action, :validate => :string, :default => "create_or_update"
+  config :map_action, :validate => ["create", "update", "create_or_update"], :default => "create_or_update"
 
   config :end_of_task, :validate => :boolean, :default => false
 
@@ -51,6 +51,8 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
   # pointer to current pipeline context
   attr_accessor :current_pipeline
 
+  # boolean indicating if expired maps should be checked on every flush call (typically because custom timeout has beeen set on a map)
+  attr_accessor :check_expired_maps_on_every_flush
 
   # ################ #
   # STATIC VARIABLES #
@@ -81,7 +83,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     end
 
     # process lambda expression to call in each filter call
-    eval("@codeblock = lambda { |event, map| #{@code} }", binding, "(aggregate filter code)")
+    eval("@codeblock = lambda { |event, map, map_meta| #{@code} }", binding, "(aggregate filter code)")
 
     # process lambda expression to call in the timeout case or previous event case
     if @timeout_code
@@ -140,6 +142,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
 
       # init aggregate_maps
       @current_pipeline.aggregate_maps[@task_id] ||= {}
+      update_aggregate_maps_metric()
 
     end
   end
@@ -202,8 +205,9 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
 
         # create aggregate map
         creation_timestamp = reference_timestamp(event)
-        aggregate_maps_element = LogStash::Filters::Aggregate::Element.new(creation_timestamp)
+        aggregate_maps_element = LogStash::Filters::Aggregate::Element.new(creation_timestamp, task_id)
         @current_pipeline.aggregate_maps[@task_id][task_id] = aggregate_maps_element
+        update_aggregate_maps_metric()
       else
         return if @map_action == "create"
       end
@@ -214,7 +218,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
       # execute the code to read/update map and event
       map = aggregate_maps_element.map
       begin
-        @codeblock.call(event, map)
+        @codeblock.call(event, map, aggregate_maps_element)
         @logger.debug("Aggregate successful filter code execution", :code => @code)
         noError = true
       rescue => exception
@@ -224,10 +228,17 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
                       :map => map,
                       :event_data => event.to_hash_with_metadata)
         event.tag("_aggregateexception")
+        metric.increment(:code_errors)
       end
 
       # delete the map if task is ended
       @current_pipeline.aggregate_maps[@task_id].delete(task_id) if @end_of_task
+      update_aggregate_maps_metric()
+
+      # process custom timeout set by code block
+      if (aggregate_maps_element.timeout || aggregate_maps_element.inactivity_timeout)
+        event_to_yield = process_map_timeout(aggregate_maps_element)
+      end
 
     end
 
@@ -238,6 +249,25 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     yield event_to_yield if event_to_yield
   end
 
+  # Process a custom timeout defined in aggregate map element
+  # Returns an event to yield if timeout=0 and push_map_as_event_on_timeout=true
+  def process_map_timeout(element)
+    event_to_yield = nil
+    init_pipeline_timeout_management()
+    if (element.timeout == 0 || element.inactivity_timeout == 0)
+      @current_pipeline.aggregate_maps[@task_id].delete(element.task_id)
+      if @current_pipeline.flush_instance_map[@task_id].push_map_as_event_on_timeout
+        event_to_yield = create_timeout_event(element.map, element.task_id)
+      end
+      @logger.debug("Aggregate remove expired map with task_id=#{element.task_id} and custom timeout=0")
+      metric.increment(:task_timeouts)
+      update_aggregate_maps_metric()
+    else
+      @current_pipeline.flush_instance_map[@task_id].check_expired_maps_on_every_flush ||= true
+    end
+    return event_to_yield
+  end
+
   # Create a new event from the aggregation_map and the corresponding task_id
   # This will create the event and
   #  if @timeout_task_id_field is set, it will set the task_id on the timeout event
@@ -255,7 +285,8 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
 
     LogStash::Util::Decorators.add_tags(@timeout_tags, event_to_yield, "filters/#{self.class.name}")
 
-    # Call code block if available
+
+    # Call timeout code block if available
     if @timeout_code
       begin
         @timeout_codeblock.call(event_to_yield)
@@ -265,9 +296,12 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
                       :timeout_code => @timeout_code,
                       :timeout_event_data => event_to_yield.to_hash_with_metadata)
         event_to_yield.tag("_aggregateexception")
+        metric.increment(:timeout_code_errors)
       end
     end
 
+    metric.increment(:pushed_events)
+
     return event_to_yield
   end
 
@@ -276,6 +310,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     previous_entry = @current_pipeline.aggregate_maps[@task_id].shift()
     previous_task_id = previous_entry[0]
     previous_map = previous_entry[1].map
+    update_aggregate_maps_metric()
     return create_timeout_event(previous_map, previous_task_id)
   end
 
@@ -287,13 +322,13 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
   # This method is invoked by LogStash every 5 seconds.
   def flush(options = {})
 
-    @logger.debug("Aggregate flush call with #{options}")
+    @logger.trace("Aggregate flush call with #{options}")
 
     # init flush/timeout properties for current pipeline
     init_pipeline_timeout_management()
     
     # launch timeout management only every interval of (@inactivity_timeout / 2) seconds or at Logstash shutdown
-    if @current_pipeline.flush_instance_map[@task_id] == self && @current_pipeline.aggregate_maps[@task_id] && (!@current_pipeline.last_flush_timestamp_map.has_key?(@task_id) || Time.now > @current_pipeline.last_flush_timestamp_map[@task_id] + @inactivity_timeout / 2 || options[:final])
+    if @current_pipeline.flush_instance_map[@task_id] == self && @current_pipeline.aggregate_maps[@task_id] && (!@current_pipeline.last_flush_timestamp_map.has_key?(@task_id) || Time.now > @current_pipeline.last_flush_timestamp_map[@task_id] + @inactivity_timeout / 2 || options[:final] || @check_expired_maps_on_every_flush)
       events_to_flush = remove_expired_maps()
 
       # at Logstash shutdown, if push_previous_map_as_event is enabled, it's important to force flush (particularly for jdbc input plugin)
@@ -302,6 +337,8 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
           events_to_flush << extract_previous_map_as_event()
         end
       end
+      
+      update_aggregate_maps_metric()
 
       # tag flushed events, indicating "final flush" special event
       if options[:final]
@@ -335,6 +372,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     if @current_pipeline.flush_instance_map[@task_id] == self
       if @timeout.nil?
         @timeout = @current_pipeline.default_timeout
+        @logger.debug("Aggregate timeout for '#{@task_id}' pattern: #{@timeout} seconds")
       end
       if @inactivity_timeout.nil?
         @inactivity_timeout = @timeout
@@ -347,23 +385,32 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
   # If @push_previous_map_as_event option is set, or @push_map_as_event_on_timeout is set, expired maps are returned as new events to be flushed to Logstash pipeline.
   def remove_expired_maps()
     events_to_flush = []
-    min_timestamp = Time.now - @timeout
-    min_inactivity_timestamp = Time.now - @inactivity_timeout
+    default_min_timestamp = Time.now - @timeout
+    default_min_inactivity_timestamp = Time.now - @inactivity_timeout
 
     @current_pipeline.mutex.synchronize do
 
       @logger.debug("Aggregate remove_expired_maps call with '#{@task_id}' pattern and #{@current_pipeline.aggregate_maps[@task_id].length} maps")
 
       @current_pipeline.aggregate_maps[@task_id].delete_if do |key, element|
+        min_timestamp = element.timeout ? Time.now - element.timeout : default_min_timestamp
+        min_inactivity_timestamp = element.inactivity_timeout ? Time.now - element.inactivity_timeout : default_min_inactivity_timestamp
         if element.creation_timestamp + element.difference_from_creation_to_now < min_timestamp || element.lastevent_timestamp + element.difference_from_creation_to_now < min_inactivity_timestamp
           if @push_previous_map_as_event || @push_map_as_event_on_timeout
             events_to_flush << create_timeout_event(element.map, key)
           end
+          @logger.debug("Aggregate remove expired map with task_id=#{key}")
+          metric.increment(:task_timeouts)
           next true
         end
         next false
       end
     end
+    
+    # disable check_expired_maps_on_every_flush if there is not anymore maps
+    if @current_pipeline.aggregate_maps[@task_id].length == 0 && @check_expired_maps_on_every_flush
+      @check_expired_maps_on_every_flush = nil
+    end
 
     return events_to_flush
   end
@@ -382,14 +429,16 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
 
     event_to_flush = nil
     event_timestamp = reference_timestamp(event)
-    min_timestamp = event_timestamp - @timeout
-    min_inactivity_timestamp = event_timestamp - @inactivity_timeout
+    min_timestamp = element.timeout ? event_timestamp - element.timeout : event_timestamp - @timeout
+    min_inactivity_timestamp = element.inactivity_timeout ? event_timestamp - element.inactivity_timeout : event_timestamp - @inactivity_timeout
 
     if element.creation_timestamp < min_timestamp || element.lastevent_timestamp < min_inactivity_timestamp
       if @push_previous_map_as_event || @push_map_as_event_on_timeout
         event_to_flush = create_timeout_event(element.map, task_id)
       end
       @current_pipeline.aggregate_maps[@task_id].delete(task_id)
+      @logger.debug("Aggregate remove expired map with task_id=#{task_id}")
+      metric.increment(:task_timeouts)
     end
 
     return event_to_flush
@@ -428,7 +477,7 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     if @execution_context
       return @execution_context.pipeline_id
     else
-      return pipeline_id = "main"
+      return "main"
     end
   end
 
@@ -438,17 +487,28 @@ class LogStash::Filters::Aggregate < LogStash::Filters::Base
     return (@timeout_timestamp_field) ? event.get(@timeout_timestamp_field).time : Time.now
   end
 
+  # update "aggregate_maps" metric, with aggregate maps count associated to configured taskid pattern 
+  def update_aggregate_maps_metric()
+    aggregate_maps = @current_pipeline.aggregate_maps[@task_id]
+    if aggregate_maps
+      metric.gauge(:aggregate_maps, aggregate_maps.length)
+    end
+  end
+
 end # class LogStash::Filters::Aggregate
 
 # Element of "aggregate_maps"
 class LogStash::Filters::Aggregate::Element
 
-  attr_accessor :creation_timestamp, :lastevent_timestamp, :difference_from_creation_to_now, :map
+  attr_accessor :creation_timestamp, :lastevent_timestamp, :difference_from_creation_to_now, :timeout, :inactivity_timeout, :task_id, :map
 
-  def initialize(creation_timestamp)
+  def initialize(creation_timestamp, task_id)
     @creation_timestamp = creation_timestamp
     @lastevent_timestamp = creation_timestamp
     @difference_from_creation_to_now = (Time.now - creation_timestamp).to_i
+    @timeout = nil
+    @inactivity_timeout = nil
+    @task_id = task_id
     @map = {}
   end
 end

data/logstash-filter-aggregate.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name = 'logstash-filter-aggregate'
-  s.version         = '2.8.0'
+  s.version         = '2.9.0'
   s.licenses = ['Apache License (2.0)']
   s.summary = "Aggregates information from several events originating with a single task"
   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'

data/spec/filters/aggregate_spec.rb

@@ -389,4 +389,35 @@ describe LogStash::Filters::Aggregate do
     end
   end
 
+  context "custom timeout on map_meta, " do
+    describe "when map_meta.timeout=0, " do
+      it "should push a new aggregated event immediately" do
+        agg_filter = setup_filter({ "task_id" => "%{ppm_id}", "code" => "map['sql_duration'] = 2; map_meta.timeout = 0", "push_map_as_event_on_timeout" => true, "timeout" => 120 })
+        agg_filter.filter(event({"ppm_id" => "1"})) do |yield_event| 
+          expect(yield_event).not_to be_nil
+          expect(yield_event.get("sql_duration")).to eq(2)
+        end
+        expect(aggregate_maps["%{ppm_id}"]).to be_empty
+      end
+    end
+    describe "when map_meta.timeout=0 and push_map_as_event_on_timeout=false, " do
+      it "should just remove expired map and not push an aggregated event" do
+        agg_filter = setup_filter({ "task_id" => "%{ppm_id}", "code" => "map_meta.timeout = 0", "push_map_as_event_on_timeout" => false, "timeout" => 120 })
+        agg_filter.filter(event({"ppm_id" => "1"}))  { |yield_event| fail "it shouldn't have yield event" } 
+        expect(aggregate_maps["%{ppm_id}"]).to be_empty
+      end
+    end
+    describe "when map_meta.inactivity_timeout=1, " do
+      it "should push a new aggregated event at next flush call" do
+        agg_filter = setup_filter({ "task_id" => "%{ppm_id}", "code" => "map['sql_duration'] = 2; map_meta.inactivity_timeout = 1", "push_map_as_event_on_timeout" => true, "timeout" => 120 })
+        agg_filter.filter(event({"ppm_id" => "1"})) { |yield_event| fail "it shouldn't have yield event" }
+        expect(aggregate_maps["%{ppm_id}"].size).to eq(1)
+        sleep(2)
+        events_to_flush = agg_filter.flush()
+        expect(events_to_flush.size).to eq(1)
+        expect(aggregate_maps["%{ppm_id}"]).to be_empty
+      end
+    end
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: logstash-filter-aggregate
 version: !ruby/object:Gem::Version
-  version: 2.8.0
+  version: 2.9.0
 platform: ruby
 authors:
 - Elastic
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2018-03-18 00:00:00.000000000 Z
+date: 2018-11-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   requirement: !ruby/object:Gem::Requirement