[receiver/datadog] Support 128 bits TraceID coming from Datadog

With this commit, we add support for 128 bits TraceIDs coming from Datadog
instrumented services. This can happen when an OTel instrumented service calls
a downstream Datadog instrumented one. Datadog instrumentation libraries store
the 128 bits TraceID into two different fields:
* TraceID: lower 64 bits of the 128 bits TraceID
* _dd.p.tid: upper 64 bits of the 128 bits TraceID

This commit adds logic that reconstructs the 128 bits TraceID. Before this
commit, only the lower 64 bits were used as TraceID.

Fixes open-telemetry#36926

Author: xiu

.chloggen/fix_datadogreceiver_128bits_traceid.yaml

@@ -0,0 +1,27 @@
+# Use this changelog template to create an entry for release notes.
+
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: bug_fix
+
+# The name of the component, or a single word describing the area of concern, (e.g. filelogreceiver)
+component: datadogreceiver
+
+# A brief description of the change.  Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Support 128 bits TraceID coming from Datadog
+
+# Mandatory: One or more tracking issues related to the change. You can use the PR number here if no issue exists.
+issues: [36926]
+
+# (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:
+
+# 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]

receiver/datadogreceiver/internal/translator/traces_translator.go

@@ -20,7 +20,7 @@ import (
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	"go.opentelemetry.io/collector/pdata/ptrace"
 	semconv "go.opentelemetry.io/collector/semconv/v1.16.0"
-	"go.opentelemetry.io/otel/trace"
+	oteltrace "go.opentelemetry.io/otel/trace"
 	"google.golang.org/protobuf/proto"
 
 	"github.com/open-telemetry/opentelemetry-collector-contrib/receiver/datadogreceiver/internal/translator/header"
@@ -90,6 +90,13 @@ func ToTraces(payload *pb.TracerPayload, req *http.Request) ptrace.Traces {
 	// now instead of being a span level attribute.
 	groupByService := make(map[string]ptrace.SpanSlice)
 
+	// Datadog traces split a 128 bits trace id in two parts: TraceID and Tags._dd_p_tid. This happens if the
+	// instrumented service received a TraceContext from an OTel instrumented service. When it happens, we need
+	// to concatenate the two into newSpan.TraceID.
+	// The following map keeps track of the TraceIDs we process as only the first span has the upper 64 bits from the
+	// 128 bits trace ID.
+	map64to128 := make(map[uint64]pcommon.TraceID)
+
 	for _, trace := range traces {
 		for _, span := range trace {
 			slice, exist := groupByService[span.Service]
@@ -101,7 +108,24 @@ func ToTraces(payload *pb.TracerPayload, req *http.Request) ptrace.Traces {
 
 			_ = tagsToSpanLinks(span.GetMeta(), newSpan.Links())
 
-			newSpan.SetTraceID(uInt64ToTraceID(0, span.TraceID))
+			// Reconstructing the 128 bits TraceID, if available or cached.
+			// Note: This may not be resilient to related spans being flushed separately in datadog's tracing libraries.
+			//       It might also not work if multiple datadog instrumented services are chained.
+			if val, ok := span.Meta["_dd.p.tid"]; ok {
+				hexTraceID := val + strconv.FormatUint(span.TraceID, 16)
+				traceID, err := oteltrace.TraceIDFromHex(hexTraceID)
+				if err != nil {
+					panic(err)
+				}
+				newSpan.SetTraceID(pcommon.TraceID(traceID))
+
+				// Child spans don't have _dd.p.tid, we cache it.
+				map64to128[span.TraceID] = pcommon.TraceID(traceID)
+			} else if val, ok := map64to128[span.TraceID]; ok {
+				newSpan.SetTraceID(val)
+			} else {
+				newSpan.SetTraceID(uInt64ToTraceID(0, span.TraceID))
+			}
 			newSpan.SetSpanID(uInt64ToSpanID(span.SpanID))
 			newSpan.SetStartTimestamp(pcommon.Timestamp(span.Start))
 			newSpan.SetEndTimestamp(pcommon.Timestamp(span.Start + span.Duration))
@@ -195,14 +219,14 @@ func tagsToSpanLinks(tags map[string]string, dest ptrace.SpanLinkSlice) error {
 		link := dest.AppendEmpty()
 
 		// Convert trace id.
-		rawTrace, errTrace := trace.TraceIDFromHex(span.TraceID)
+		rawTrace, errTrace := oteltrace.TraceIDFromHex(span.TraceID)
 		if errTrace != nil {
 			return errTrace
 		}
 		link.SetTraceID(pcommon.TraceID(rawTrace))
 
 		// Convert span id.
-		rawSpan, errSpan := trace.SpanIDFromHex(span.SpanID)
+		rawSpan, errSpan := oteltrace.SpanIDFromHex(span.SpanID)
 		if errSpan != nil {
 			return errSpan
 		}

receiver/datadogreceiver/internal/translator/traces_translator_test.go

@@ -221,3 +221,50 @@ func TestUpsertHeadersAttributes(t *testing.T) {
 	assert.True(t, ok)
 	assert.Equal(t, "dotnet", val.Str())
 }
+
+func TestToTraces64to128bits(t *testing.T) {
+	// Test that we properly reconstruct a 128 bits TraceID from Datadog spans.
+	// This is necessary when a datadog instrumented service has been called by an OTel instrumented one as Datadog will
+	// split the TraceID into two different values:
+	// * TraceID holds the lower 64 bits
+	// * `_dd.p.tid` holds the upper 64 bits
+	expectedTraceID := "f233b7e1421e8bde1d99f09757cf199d"
+	spanIDParentOtel, _ := strconv.ParseUint("6b953724b399048a", 16, 32)
+	spanIDParentDD, _ := strconv.ParseUint("039f8ec65ed09993", 16, 32)
+	spanIDChildDD, _ := strconv.ParseUint("5ab19b8ebe922796", 16, 32)
+
+	spans := []pb.Span{
+		{
+			TraceID:  2133000431340558749,
+			SpanID:   spanIDParentDD,
+			ParentID: spanIDParentOtel,
+			Meta: map[string]string{
+				"_dd.p.tid": "f233b7e1421e8bde",
+			},
+		},
+		{
+			TraceID:  2133000431340558749,
+			SpanID:   spanIDChildDD,
+			ParentID: spanIDParentDD,
+		},
+	}
+	payload := &pb.TracerPayload{
+		Chunks: traceChunksFromSpans(spans),
+	}
+
+	req := &http.Request{
+		Header: http.Header{},
+	}
+	req.Header.Set(header.Lang, "go")
+
+	traces := ToTraces(payload, req)
+	assert.Equal(t, 2, traces.SpanCount(), "Expected 2 spans")
+
+	for _, rs := range traces.ResourceSpans().All() {
+		for _, ss := range rs.ScopeSpans().All() {
+			for _, span := range ss.Spans().All() {
+				assert.Equal(t, expectedTraceID, span.TraceID().String(), "Expected TraceID to be "+expectedTraceID)
+			}
+		}
+	}
+}