[exporter/elasticsearch] remove dedup config (#33776)

**Description:**

Remove the `dedup` configuration setting, and always de-duplicate.
Elasticsearch does not permit duplicate keys in JSON objects, and this
configuration is adding more complexity to the code than it's worth.

I've simplified the `internal/objmodel` API slightly, unexporting the
`Sort` methods, which are internally called by the now unconditional
call to `Dedup`.

**Link to tracking Issue:**

Closes
#33773

**Testing:**

Ran the unit tests, which cover deduplication. None of the tests in
package elasticsearchexporter covered `dedup: false`.

**Documentation:** <Describe the documentation added.>

---------

Co-authored-by: Carson Ip <carsonip@users.noreply.github.com>
Co-authored-by: Andrzej Stencel <andrzej.stencel@elastic.co>

.chloggen/elasticsearch-always-dedup.yaml

@@ -0,0 +1,29 @@
+# Use this changelog template to create an entry for release notes.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: breaking
+
+# The name of the component, or a single word describing the area of concern, (e.g. filelogreceiver)
+component: elasticsearchexporter
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Make "dedup" option no-op, always de-duplicate.
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [33773]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+subtext:
+  Elasticsearch does not permit duplicate keys in JSON objects,
+  so there is no value in being able to configure deduplication.
+
+# If your change doesn't affect end users or the exported elements of any package,
+# you should instead start your pull request title with [chore] or use the "Skip Changelog" label.
+# Optional: The change log or logs in which this entry should be included.
+# e.g. '[user]' or '[user, api]'
+# Include 'user' if the change is relevant to end users.
+# Include 'api' if there is a change to a library API.
+# Default: '[user]'
+change_logs: [user]

cmd/otelcontribcol/go.mod

@@ -727,7 +727,7 @@ require (
 	github.com/syndtr/gocapability v0.0.0-20200815063812-42c35b437635 // indirect
 	github.com/tencentcloud/tencentcloud-sdk-go/tencentcloud/common v1.0.954 // indirect
 	github.com/tg123/go-htpasswd v1.2.2 // indirect
-	github.com/tidwall/gjson v1.14.2 // indirect
+	github.com/tidwall/gjson v1.17.1 // indirect
 	github.com/tidwall/match v1.1.1 // indirect
 	github.com/tidwall/pretty v1.2.0 // indirect
 	github.com/tidwall/tinylru v1.1.0 // indirect

exporter/elasticsearchexporter/README.md

@@ -129,10 +129,9 @@ behaviours, which may be configured through the following settings:
     - `raw`: Omit the `Attributes.` string prefixed to field names for log and 
              span attributes as well as omit the `Events.` string prefixed to
              field names for span events. 
-  - `dedup` (default=true; DEPRECATED, in future deduplication will always be enabled):
-    Try to find and remove duplicate fields/attributes from events before publishing
-    to Elasticsearch. Some structured logging libraries can produce duplicate fields
-    (for example zap). Elasticsearch will reject documents that have duplicate fields.
+  - `dedup` (DEPRECATED). This configuration is deprecated and non-operational,
+    and will be removed in the future. Object keys are always deduplicated to
+    avoid Elasticsearch rejecting documents.
   - `dedot` (default=true; DEPRECATED, in future dedotting will always be enabled
     for ECS mode, and never for other modes): When enabled attributes with `.`
     will be split into proper json objects.

exporter/elasticsearchexporter/config.go

@@ -160,12 +160,12 @@ type MappingsSettings struct {
 	// Mode configures the field mappings.
 	Mode string `mapstructure:"mode"`
 
-	// Try to find and remove duplicate fields
+	// Dedup is non-operational, and will be removed in the future.
 	//
-	// Deprecated: [v0.104.0] deduplication will always be applied in future,
-	// with no option to disable. Disabling deduplication is not meaningful,
-	// as Elasticsearch will reject documents with duplicate JSON object keys.
-	Dedup bool `mapstructure:"dedup"`
+	// Deprecated: [v0.104.0] deduplication is always enabled, and cannot be
+	// disabled. Disabling deduplication is not meaningful, as Elasticsearch
+	// will always reject documents with duplicate JSON object keys.
+	Dedup *bool `mapstructure:"dedup,omitempty"`
 
 	// Deprecated: [v0.104.0] dedotting will always be applied for ECS mode
 	// in future, and never for other modes. Elasticsearch's "dot_expander"
@@ -322,8 +322,8 @@ func (cfg *Config) MappingMode() MappingMode {
 }
 
 func logConfigDeprecationWarnings(cfg *Config, logger *zap.Logger) {
-	if !cfg.Mapping.Dedup {
-		logger.Warn("dedup has been deprecated, and will always be enabled in future")
+	if cfg.Mapping.Dedup != nil {
+		logger.Warn("dedup is deprecated, and is always enabled")
 	}
 	if cfg.Mapping.Dedot && cfg.MappingMode() != MappingECS || !cfg.Mapping.Dedot && cfg.MappingMode() == MappingECS {
 		logger.Warn("dedot has been deprecated: in the future, dedotting will always be performed in ECS mode only")

exporter/elasticsearchexporter/config_test.go

@@ -100,7 +100,6 @@ func TestConfig(t *testing.T) {
 				},
 				Mapping: MappingsSettings{
 					Mode:  "none",
-					Dedup: true,
 					Dedot: true,
 				},
 				LogstashFormat: LogstashFormatSettings{
@@ -162,7 +161,6 @@ func TestConfig(t *testing.T) {
 				},
 				Mapping: MappingsSettings{
 					Mode:  "none",
-					Dedup: true,
 					Dedot: true,
 				},
 				LogstashFormat: LogstashFormatSettings{
@@ -224,7 +222,6 @@ func TestConfig(t *testing.T) {
 				},
 				Mapping: MappingsSettings{
 					Mode:  "none",
-					Dedup: true,
 					Dedot: true,
 				},
 				LogstashFormat: LogstashFormatSettings{

exporter/elasticsearchexporter/exporter.go

@@ -44,7 +44,6 @@ func newExporter(
 	}
 
 	model := &encodeModel{
-		dedup: cfg.Mapping.Dedup,
 		dedot: cfg.Mapping.Dedot,
 		mode:  cfg.MappingMode(),
 	}

exporter/elasticsearchexporter/factory.go

@@ -76,7 +76,6 @@ func createDefaultConfig() component.Config {
 		},
 		Mapping: MappingsSettings{
 			Mode:  "none",
-			Dedup: true,
 			Dedot: true,
 		},
 		LogstashFormat: LogstashFormatSettings{

exporter/elasticsearchexporter/factory_test.go

@@ -100,8 +100,9 @@ func TestFactory_CreateLogsAndTracesExporterWithDeprecatedIndexOption(t *testing
 func TestFactory_DedupDeprecated(t *testing.T) {
 	factory := NewFactory()
 	cfg := withDefaultConfig(func(cfg *Config) {
+		dedup := false
 		cfg.Endpoint = "http://testing.invalid:9200"
-		cfg.Mapping.Dedup = false
+		cfg.Mapping.Dedup = &dedup
 		cfg.Mapping.Dedot = false // avoid dedot warnings
 	})
 
@@ -123,9 +124,9 @@ func TestFactory_DedupDeprecated(t *testing.T) {
 
 	records := logObserver.AllUntimed()
 	assert.Len(t, records, 3)
-	assert.Equal(t, "dedup has been deprecated, and will always be enabled in future", records[0].Message)
-	assert.Equal(t, "dedup has been deprecated, and will always be enabled in future", records[1].Message)
-	assert.Equal(t, "dedup has been deprecated, and will always be enabled in future", records[2].Message)
+	assert.Equal(t, "dedup is deprecated, and is always enabled", records[0].Message)
+	assert.Equal(t, "dedup is deprecated, and is always enabled", records[1].Message)
+	assert.Equal(t, "dedup is deprecated, and is always enabled", records[2].Message)
 }
 
 func TestFactory_DedotDeprecated(t *testing.T) {

exporter/elasticsearchexporter/go.mod

@@ -11,6 +11,7 @@ require (
 	github.com/open-telemetry/opentelemetry-collector-contrib/internal/common v0.104.0
 	github.com/open-telemetry/opentelemetry-collector-contrib/internal/coreinternal v0.104.0
 	github.com/stretchr/testify v1.9.0
+	github.com/tidwall/gjson v1.17.1
 	go.opentelemetry.io/collector/component v0.104.1-0.20240712081520-6227646b0146
 	go.opentelemetry.io/collector/config/configauth v0.104.1-0.20240712081520-6227646b0146
 	go.opentelemetry.io/collector/config/configcompression v1.11.1-0.20240712081520-6227646b0146
@@ -64,6 +65,8 @@ require (
 	github.com/prometheus/common v0.55.0 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect
 	github.com/rs/cors v1.11.0 // indirect
+	github.com/tidwall/match v1.1.1 // indirect
+	github.com/tidwall/pretty v1.2.0 // indirect
 	go.elastic.co/apm/module/apmzap/v2 v2.6.0 // indirect
 	go.elastic.co/apm/v2 v2.6.0 // indirect
 	go.elastic.co/fastjson v1.3.0 // indirect

exporter/elasticsearchexporter/internal/objmodel/objmodel.go

@@ -180,15 +180,14 @@ func (doc *Document) AddEvents(key string, events ptrace.SpanEventSlice) {
 	}
 }
 
-// Sort sorts all fields in the document by key name.
-func (doc *Document) Sort() {
+func (doc *Document) sort() {
 	sort.SliceStable(doc.fields, func(i, j int) bool {
 		return doc.fields[i].key < doc.fields[j].key
 	})
 
 	for i := range doc.fields {
 		fld := &doc.fields[i]
-		fld.value.Sort()
+		fld.value.sort()
 	}
 }
 
@@ -199,7 +198,7 @@ func (doc *Document) Sort() {
 func (doc *Document) Dedup() {
 	// 1. Always ensure the fields are sorted, Dedup support requires
 	// Fields to be sorted.
-	doc.Sort()
+	doc.sort()
 
 	// 2. rename fields if a primitive value is overwritten by an object.
 	//    For example the pair (path.x=1, path.x.a="test") becomes:
@@ -223,7 +222,7 @@ func (doc *Document) Dedup() {
 		}
 	}
 	if renamed {
-		doc.Sort()
+		doc.sort()
 	}
 
 	// 3. mark duplicates as 'ignore'
@@ -423,14 +422,13 @@ func ValueFromAttribute(attr pcommon.Value) Value {
 	}
 }
 
-// Sort recursively sorts all keys in docuemts held by the value.
-func (v *Value) Sort() {
+func (v *Value) sort() {
 	switch v.kind {
 	case KindObject:
-		v.doc.Sort()
+		v.doc.sort()
 	case KindArr:
 		for i := range v.arr {
-			v.arr[i].Sort()
+			v.arr[i].sort()
 		}
 	}
 }

exporter/elasticsearchexporter/internal/objmodel/objmodel_test.go

@@ -80,46 +80,11 @@ func TestObjectModel_CreateMap(t *testing.T) {
 	for name, test := range tests {
 		t.Run(name, func(t *testing.T) {
 			doc := test.build()
-			doc.Sort()
 			assert.Equal(t, test.want, doc)
 		})
 	}
 }
 
-func TestDocument_Sort(t *testing.T) {
-	tests := map[string]struct {
-		build func() Document
-		want  Document
-	}{
-		"keys are sorted": {
-			build: func() (doc Document) {
-				doc.AddInt("z", 26)
-				doc.AddInt("a", 1)
-				return doc
-			},
-			want: Document{[]field{{"a", IntValue(1)}, {"z", IntValue(26)}}},
-		},
-		"sorting is stable": {
-			build: func() (doc Document) {
-				doc.AddInt("a", 1)
-				doc.AddInt("c", 3)
-				doc.AddInt("a", 2)
-				return doc
-			},
-			want: Document{[]field{{"a", IntValue(1)}, {"a", IntValue(2)}, {"c", IntValue(3)}}},
-		},
-	}
-
-	for name, test := range tests {
-		t.Run(name, func(t *testing.T) {
-			doc := test.build()
-			doc.Sort()
-			assert.Equal(t, test.want, doc)
-		})
-	}
-
-}
-
 func TestObjectModel_Dedup(t *testing.T) {
 	tests := map[string]struct {
 		build func() Document
@@ -200,7 +165,6 @@ func TestObjectModel_Dedup(t *testing.T) {
 	for name, test := range tests {
 		t.Run(name, func(t *testing.T) {
 			doc := test.build()
-			doc.Sort()
 			doc.Dedup()
 			assert.Equal(t, test.want, doc)
 		})

exporter/elasticsearchexporter/model.go

@@ -76,7 +76,6 @@ type mappingModel interface {
 //
 // See: https://github.com/open-telemetry/oteps/blob/master/text/logs/0097-log-data-model.md
 type encodeModel struct {
-	dedup bool
 	dedot bool
 	mode  MappingMode
 }
@@ -95,13 +94,9 @@ func (m *encodeModel) encodeLog(resource pcommon.Resource, record plog.LogRecord
 	default:
 		document = m.encodeLogDefaultMode(resource, record, scope)
 	}
+	document.Dedup()
 
 	var buf bytes.Buffer
-	if m.dedup {
-		document.Dedup()
-	} else if m.dedot {
-		document.Sort()
-	}
 	err := document.Serialize(&buf, m.dedot)
 	return buf.Bytes(), err
 }
@@ -171,11 +166,7 @@ func (m *encodeModel) encodeLogECSMode(resource pcommon.Resource, record plog.Lo
 }
 
 func (m *encodeModel) encodeDocument(document objmodel.Document) ([]byte, error) {
-	if m.dedup {
-		document.Dedup()
-	} else if m.dedot {
-		document.Sort()
-	}
+	document.Dedup()
 
 	var buf bytes.Buffer
 	err := document.Serialize(&buf, m.dedot)
@@ -225,12 +216,7 @@ func (m *encodeModel) encodeSpan(resource pcommon.Resource, span ptrace.Span, sc
 	m.encodeEvents(&document, span.Events())
 	document.AddInt("Duration", durationAsMicroseconds(span.StartTimestamp().AsTime(), span.EndTimestamp().AsTime())) // unit is microseconds
 	document.AddAttributes("Scope", scopeToAttributes(scope))
-
-	if m.dedup {
-		document.Dedup()
-	} else if m.dedot {
-		document.Sort()
-	}
+	document.Dedup()
 
 	var buf bytes.Buffer
 	err := document.Serialize(&buf, m.dedot)