Beats event processing and default fields (#10801)

This changes moves the generation of the event processing into it's
distinct package, such that the actual publisher pipeline will not
define any processors anymore. A new instance of a publisher pipeline
must not add fields on it's own.

This change converts the event processing pipline into the 'Supporter'
pattern, which is already used for Index Management.
As different beats ask for slightly different behavior in the event
processing (e.g. normalize, default builtins and so on), the
`processing.Support` can be used for customizations.

filebeat/channel/factory.go

@@ -125,13 +125,15 @@ func (f *OutletFactory) Create(p beat.Pipeline, cfg *common.Config, dynFields *c
 	}
 
 	client, err := p.ConnectWith(beat.ClientConfig{
-		PublishMode:   beat.GuaranteedSend,
-		EventMetadata: config.EventMetadata,
-		DynamicFields: dynFields,
-		Meta:          meta,
-		Fields:        fields,
-		Processor:     processors,
-		Events:        f.eventer,
+		PublishMode: beat.GuaranteedSend,
+		Processing: beat.ProcessingConfig{
+			EventMetadata: config.EventMetadata,
+			DynamicFields: dynFields,
+			Meta:          meta,
+			Fields:        fields,
+			Processor:     processors,
+		},
+		Events: f.eventer,
 	})
 	if err != nil {
 		return nil, err

heartbeat/monitors/task.go

@@ -139,9 +139,11 @@ func (t *configuredJob) Start() {
 	}
 
 	t.client, err = t.monitor.pipelineConnector.ConnectWith(beat.ClientConfig{
-		EventMetadata: t.config.EventMetadata,
-		Processor:     t.processors,
-		Fields:        fields,
+		Processing: beat.ProcessingConfig{
+			EventMetadata: t.config.EventMetadata,
+			Processor:     t.processors,
+			Fields:        fields,
+		},
 	})
 	if err != nil {
 		logp.Err("could not start monitor: %v", err)

journalbeat/input/input.go

@@ -125,10 +125,12 @@ func New(
 func (i *Input) Run() {
 	var err error
 	i.client, err = i.pipeline.ConnectWith(beat.ClientConfig{
-		PublishMode:   beat.GuaranteedSend,
-		EventMetadata: i.eventMeta,
-		Meta:          nil,
-		Processor:     i.processors,
+		PublishMode: beat.GuaranteedSend,
+		Processing: beat.ProcessingConfig{
+			EventMetadata: i.eventMeta,
+			Meta:          nil,
+			Processor:     i.processors,
+		},
 		ACKCount: func(n int) {
 			i.logger.Infof("journalbeat successfully published %d events", n)
 		},

libbeat/beat/pipeline.go

@@ -46,22 +46,7 @@ type Client interface {
 type ClientConfig struct {
 	PublishMode PublishMode
 
-	// EventMetadata configures additional fields/tags to be added to published events.
-	EventMetadata common.EventMetadata
-
-	// Meta provides additional meta data to be added to the Meta field in the beat.Event
-	// structure.
-	Meta common.MapStr
-
-	// Fields provides additional 'global' fields to be added to every event
-	Fields common.MapStr
-
-	// DynamicFields provides additional fields to be added to every event, supporting live updates
-	DynamicFields *common.MapStrPointer
-
-	// Processors passes additional processor to the client, to be executed before
-	// the pipeline processors.
-	Processor ProcessorList
+	Processing ProcessingConfig
 
 	// WaitClose sets the maximum duration to wait on ACK, if client still has events
 	// active non-acknowledged events in the publisher pipeline.
@@ -72,14 +57,6 @@ type ClientConfig struct {
 	// Events configures callbacks for common client callbacks
 	Events ClientEventer
 
-	// By default events are normalized within processor pipeline,
-	// if the normalization step should be skipped set this to true.
-	SkipNormalization bool
-
-	// By default events are decorated with agent metadata.
-	// To skip adding that metadata set this to true.
-	SkipAgentMetadata bool
-
 	// ACK handler strategies.
 	// Note: ack handlers are run in another go-routine owned by the publisher pipeline.
 	//       They should not block for to long, to not block the internal buffers for
@@ -101,6 +78,31 @@ type ClientConfig struct {
 	ACKLastEvent func(interface{})
 }
 
+// ProcessingConfig provides additional event processing settings a client can
+// pass to the publisher pipeline on Connect.
+type ProcessingConfig struct {
+	// EventMetadata configures additional fields/tags to be added to published events.
+	EventMetadata common.EventMetadata
+
+	// Meta provides additional meta data to be added to the Meta field in the beat.Event
+	// structure.
+	Meta common.MapStr
+
+	// Fields provides additional 'global' fields to be added to every event
+	Fields common.MapStr
+
+	// DynamicFields provides additional fields to be added to every event, supporting live updates
+	DynamicFields *common.MapStrPointer
+
+	// Processors passes additional processor to the client, to be executed before
+	// the pipeline processors.
+	Processor ProcessorList
+
+	// Private contains additional information to be passed to the processing
+	// pipeline builder.
+	Private interface{}
+}
+
 // ClientEventer provides access to internal client events.
 type ClientEventer interface {
 	Closing() // Closing indicates the client is being shutdown next

libbeat/cmd/instance/beat.go

@@ -62,6 +62,7 @@ import (
 	"github.com/elastic/beats/libbeat/paths"
 	"github.com/elastic/beats/libbeat/plugin"
 	"github.com/elastic/beats/libbeat/publisher/pipeline"
+	"github.com/elastic/beats/libbeat/publisher/processing"
 	svc "github.com/elastic/beats/libbeat/service"
 	"github.com/elastic/beats/libbeat/version"
 	sysinfo "github.com/elastic/go-sysinfo"
@@ -78,6 +79,8 @@ type Beat struct {
 
 	keystore keystore.Keystore
 	index    idxmgmt.Supporter
+
+	processing processing.Supporter
 }
 
 type beatConfig struct {
@@ -310,6 +313,7 @@ func (b *Beat) createBeater(bt beat.Creator) (beat.Beater, error) {
 			Logger:    logp.L().Named("publisher"),
 		},
 		b.Config.Pipeline,
+		b.processing,
 		b.makeOutputFactory(b.Config.Output),
 	)
 
@@ -593,6 +597,16 @@ func (b *Beat) configure(settings Settings) error {
 		imFactory = idxmgmt.MakeDefaultSupport(settings.ILM)
 	}
 	b.index, err = imFactory(nil, b.Beat.Info, b.RawConfig)
+	if err != nil {
+		return err
+	}
+
+	processingFactory := settings.Processing
+	if processingFactory == nil {
+		processingFactory = processing.MakeDefaultBeatSupport(true)
+	}
+	b.processing, err = processingFactory(b.Info, logp.L().Named("processors"), b.RawConfig)
+
 	return err
 }
 

libbeat/cmd/instance/settings.go

@@ -24,6 +24,7 @@ import (
 	"github.com/elastic/beats/libbeat/idxmgmt"
 	"github.com/elastic/beats/libbeat/idxmgmt/ilm"
 	"github.com/elastic/beats/libbeat/monitoring/report"
+	"github.com/elastic/beats/libbeat/publisher/processing"
 )
 
 // Settings contains basic settings for any beat to pass into GenRootCmd
@@ -40,4 +41,6 @@ type Settings struct {
 	// load custom index manager. The config object will be the Beats root configuration.
 	IndexManagement idxmgmt.SupportFactory
 	ILM             ilm.SupportFactory
+
+	Processing processing.SupportFactory
 }

libbeat/monitoring/report/elasticsearch/elasticsearch.go

@@ -37,6 +37,7 @@ import (
 	"github.com/elastic/beats/libbeat/outputs/outil"
 	"github.com/elastic/beats/libbeat/outputs/transport"
 	"github.com/elastic/beats/libbeat/publisher/pipeline"
+	"github.com/elastic/beats/libbeat/publisher/processing"
 	"github.com/elastic/beats/libbeat/publisher/queue"
 	"github.com/elastic/beats/libbeat/publisher/queue/memqueue"
 )
@@ -169,11 +170,16 @@ func makeReporter(beat beat.Info, settings report.Settings, cfg *common.Config)
 	outClient := outputs.NewFailoverClient(clients)
 	outClient = outputs.WithBackoff(outClient, config.Backoff.Init, config.Backoff.Max)
 
+	processing, err := processing.MakeDefaultSupport(true)(beat, log, common.NewConfig())
+	if err != nil {
+		return nil, err
+	}
+
 	pipeline, err := pipeline.New(
 		beat,
 		pipeline.Monitors{
 			Metrics: monitoring,
-			Logger:  logp.NewLogger(selector),
+			Logger:  log,
 		},
 		queueFactory,
 		outputs.Group{
@@ -184,6 +190,7 @@ func makeReporter(beat beat.Info, settings report.Settings, cfg *common.Config)
 		pipeline.Settings{
 			WaitClose:     0,
 			WaitCloseMode: pipeline.NoWaitOnClose,
+			Processors:    processing,
 		})
 	if err != nil {
 		return nil, err

libbeat/publisher/pipeline/module.go

@@ -26,7 +26,7 @@ import (
 	"github.com/elastic/beats/libbeat/logp"
 	"github.com/elastic/beats/libbeat/monitoring"
 	"github.com/elastic/beats/libbeat/outputs"
-	"github.com/elastic/beats/libbeat/processors"
+	"github.com/elastic/beats/libbeat/publisher/processing"
 	"github.com/elastic/beats/libbeat/publisher/queue"
 )
 
@@ -60,6 +60,7 @@ func Load(
 	beatInfo beat.Info,
 	monitors Monitors,
 	config Config,
+	processors processing.Supporter,
 	makeOutput func(outputs.Observer) (string, outputs.Group, error),
 ) (*Pipeline, error) {
 	log := monitors.Logger
@@ -71,28 +72,11 @@ func Load(
 		log.Info("Dry run mode. All output types except the file based one are disabled.")
 	}
 
-	processors, err := processors.New(config.Processors)
-	if err != nil {
-		return nil, fmt.Errorf("error initializing processors: %v", err)
-	}
-
 	name := beatInfo.Name
 	settings := Settings{
 		WaitClose:     0,
 		WaitCloseMode: NoWaitOnClose,
-		Disabled:      publishDisabled,
 		Processors:    processors,
-		Annotations: Annotations{
-			Event: config.EventMetadata,
-			Builtin: common.MapStr{
-				"host": common.MapStr{
-					"name": name,
-				},
-				"ecs": common.MapStr{
-					"version": "1.0.0-beta2",
-				},
-			},
-		},
 	}
 
 	queueBuilder, err := createQueueBuilder(config.Queue, monitors)