feat(traces): migrate trace propagation to dd-trace-rs by duncanista · Pull Request #1089 · DataDog/datadog-lambda-extension

duncanista · 2026-03-06T22:41:17Z

Overview

Migrate trace context propagation from a local implementation to the datadog-opentelemetry crate in dd-trace-rs. This removes ~750 lines of propagation logic (Datadog headers, W3C TraceContext) that we were maintaining locally in favor of the shared upstream
implementation.

Key changes:

Add datadog-opentelemetry as a dependency
Delete local propagation modules: traces/context.rs, traces/propagation/error.rs, traces/propagation/text_map_propagator.rs
Keep a thin DatadogCompositePropagator wrapper that adds ot-baggage-* header extraction, which is not yet supported upstream
Update all consumers to import SpanContext, Sampling, TracePropagationStyle, and header constants directly from datadog-opentelemetry
Adapt to upstream types: u128 trace IDs (cast to u64 for Datadog protocol), Sampling struct, SamplingPriority newtype

Testing

All 476 existing tests pass
Propagation tests updated to use Datadog/TraceContext styles (upstream does not expose B3/B3Multi variants)
Config parsing tests updated to reflect valid upstream style names

Replace local trace extraction/propagation implementation with datadog-opentelemetry crate from dd-trace-rs. This removes maintenance burden for Datadog/TraceContext propagator logic. - Add datadog-opentelemetry dependency - Remove local propagation modules (context, error, text_map_propagator) - Use DatadogCompositePropagator from dd-trace-rs with thin wrapper for ot-baggage-* header extraction (not yet supported upstream) - Update all consumers to import directly from datadog-opentelemetry - Adapt to u128 trace_id, Sampling struct, SamplingPriority types

duncanista · 2026-03-06T22:41:42Z

Merge needs to be held until dd-trace-rs changes are on main or released

…ace IDs The truncation is intentional — Datadog protocol transmits the lower 64 bits via x-datadog-trace-id, with upper bits in _dd.p.tid tag.

Copilot

Pull request overview

This PR migrates trace context propagation from Bottlecap’s local implementation to datadog-opentelemetry (from dd-trace-rs), removing local Datadog/W3C propagation code while keeping a small wrapper to support ot-baggage-* extraction.

Changes:

Add datadog-opentelemetry dependency and adapt config to satisfy its propagation configuration trait.
Remove local propagation/context modules and update call sites to use upstream SpanContext, Sampling, TracePropagationStyle, and header constants.
Keep a thin DatadogCompositePropagator wrapper to add ot-baggage-* extraction not supported upstream yet.

Reviewed changes

Copilot reviewed 19 out of 20 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
bottlecap/src/traces/propagation/text_map_propagator.rs	Deleted local header/W3C TraceContext propagators and tests.
bottlecap/src/traces/propagation/error.rs	Deleted local propagation error type.
bottlecap/src/traces/context.rs	Deleted local `SpanContext`/`Sampling` types in favor of upstream.
bottlecap/src/traces/mod.rs	Removes `context` module export now that it’s deleted.
bottlecap/src/traces/propagation/mod.rs	Replaces local multi-propagator logic with wrapper around upstream composite + baggage extraction.
bottlecap/src/traces/propagation/carrier.rs	Switches carrier traits to upstream and introduces `JsonCarrier` newtypes for `serde_json::Value`.
bottlecap/src/otlp/transform.rs	Imports Datadog trace-id-high constant from upstream.
bottlecap/src/lifecycle/listener.rs	Updates injected headers to use upstream context/constants and u128 trace IDs.
bottlecap/src/lifecycle/invocation/processor.rs	Updates extraction to use upstream propagator and JSON carriers; adjusts for u128 trace IDs.
bottlecap/src/lifecycle/invocation/processor_service.rs	Updates `SpanContext` import to upstream.
bottlecap/src/lifecycle/invocation/span_inferrer.rs	Replaces local propagator trait usage with `DatadogCompositePropagator`.
bottlecap/src/lifecycle/invocation/context.rs	Switches to upstream `SpanContext`.
bottlecap/src/lifecycle/invocation/triggers/step_function_event.rs	Updates to upstream `SpanContext`/`Sampling` and adds local `_dd.p.*` parsing helper.
bottlecap/src/lifecycle/invocation/triggers/sqs_event.rs	Updates to upstream `SpanContext`/`Sampling` and u128 trace IDs.
bottlecap/src/config/trace_propagation_style.rs	Uses upstream `TracePropagationStyle` and keeps only deserialization logic.
bottlecap/src/config/mod.rs	Implements upstream `PropagationConfig` for `Config`; updates tests to match supported styles.
bottlecap/src/config/env.rs	Adjusts env parsing tests/defaults to supported upstream style names.
bottlecap/src/config/yaml.rs	Adjusts YAML sample/tests to supported upstream style names.
bottlecap/Cargo.toml	Adds git dependency on `datadog-opentelemetry`.
bottlecap/Cargo.lock	Updates lockfile with new dependency tree introduced by `datadog-opentelemetry`.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

Copilot · 2026-03-06T22:46:52Z

bottlecap/src/traces/propagation/mod.rs

+        Some(context)
    }

    fn attach_baggage(context: &mut SpanContext, carrier: &dyn Extractor) {


attach_baggage checks key.strip_prefix("ot-baggage-") on the raw key returned by carrier.keys(), which is case-sensitive. For carriers that don't normalize header keys (e.g., JSON event headers), baggage headers like OT-Baggage-foo won't be extracted. Consider doing an ASCII-lowercase prefix check (or normalizing keys) so extraction is case-insensitive like HTTP headers.

Copilot · 2026-03-06T22:46:52Z

bottlecap/src/lifecycle/invocation/triggers/step_function_event.rs

+fn extract_propagation_tags(tags_str: &str) -> HashMap<String, String> {
+    tags_str
+        .split(',')
+        .filter_map(|pair| {
+            let (k, v) = pair.split_once('=')?;
+            if k.starts_with("_dd.p.") {
+                Some((k.to_string(), v.to_string()))
+            } else {
+                None
+            }
+        })
+        .collect()
+}


extract_propagation_tags duplicates the same _dd.p.* parsing logic added in lifecycle/invocation/processor.rs. Consider centralizing this into a shared helper (e.g., in traces::propagation) to avoid the two implementations drifting (and potentially add trimming of whitespace around pairs/keys in one place).

Copilot · 2026-03-06T22:46:52Z

bottlecap/src/lifecycle/invocation/processor.rs

        if let Some(request_id) = span.meta.get("request_id") {
            self.context_buffer.add_tracer_span(request_id, span);
        }
    }
 }

+fn extract_propagation_tags(carrier: &HashMap<String, String>) -> HashMap<String, String> {
+    let carrier_tags = carrier.get(DATADOG_TAGS_KEY).map_or("", String::as_str);
+    carrier_tags
+        .split(',')
+        .filter_map(|pair| {
+            let (k, v) = pair.split_once('=')?;
+            if k.starts_with("_dd.p.") {
+                Some((k.to_string(), v.to_string()))


This _dd.p.* parsing helper is now duplicated (also in step_function_event.rs). Consider moving this logic into a single shared helper to keep propagation tag parsing consistent across trigger sources (and to make it easier to apply improvements like trimming/validation in one place).

Copilot · 2026-03-06T22:46:52Z

bottlecap/src/lifecycle/listener.rs

    fn inject_span_context_to_headers(headers: &mut HeaderMap, span_context: &SpanContext) {
        headers.insert(
            DATADOG_TRACE_ID_KEY,
-            span_context
-                .trace_id
-                .to_string()
-                .parse()
-                .expect("Failed to parse trace id"),
+            {
+                #[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits
+                let trace_id = span_context.trace_id as u64;
+                trace_id
+            }
+            .to_string()
+            .parse()
+            .expect("Failed to parse trace id"),
        );



SpanContext.trace_id is now a u128, but this uses as u64 which silently truncates. If the intended behavior is to send the lower 64 bits for Datadog headers, make that explicit (mask/helper) to document intent. Also, this helper currently only injects a subset of the Datadog propagation state into headers; consider serializing all relevant _dd.p.* propagation tags from span_context.tags into x-datadog-tags so important fields (e.g., 128-bit high bits and other propagated tags) aren't lost.

Copilot · 2026-03-06T22:46:53Z

bottlecap/src/lifecycle/invocation/processor.rs

+            #[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits
+            {
+                context.invocation_span.trace_id = sc.trace_id as u64;
+            }


sc.trace_id as u64 silently truncates the u128 trace id when storing it into invocation_span.trace_id. If the system relies on lower-64-bits semantics, prefer an explicit conversion (mask/helper) and ensure higher-order bits are preserved via _dd.p.tid in tags when needed.

Suggested change

#[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits

{

context.invocation_span.trace_id = sc.trace_id as u64;

}

// Datadog protocol uses the lower 64 bits for the trace_id field; preserve full 128 bits via _dd.p.tid

let full_trace_id = sc.trace_id;

let lower_64_trace_id = (full_trace_id & 0xffff_ffff_ffff_ffff) as u64;

context.invocation_span.trace_id = lower_64_trace_id;

// Store full 128-bit trace id as 32-char hex in Datadog internal tag

context

.invocation_span

.meta

.entry("_dd.p.tid".to_string())

.or_insert_with(|| format!("{:032x}", full_trace_id));

Copilot · 2026-03-06T22:46:53Z

bottlecap/src/lifecycle/invocation/processor.rs

+            #[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits
+            {
+                trace_id = sc.trace_id as u64;
+            }
            parent_id = sc.span_id;
            tags.extend(sc.tags.clone());


Same truncation issue as earlier: sc.trace_id as u64 will drop higher 64 bits. Consider an explicit lower-64-bits conversion and ensure _dd.p.tid is present in the propagated tags when the upper bits are non-zero.

Suggested change

#[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits

{

trace_id = sc.trace_id as u64;

}

parent_id = sc.span_id;

tags.extend(sc.tags.clone());

// Datadog protocol uses the lower 64 bits for the trace id field.

let full_trace_id = sc.trace_id;

let lower_64 = (full_trace_id & ((1u128 << 64) - 1)) as u64;

let upper_64 = full_trace_id >> 64;

trace_id = lower_64;

parent_id = sc.span_id;

tags.extend(sc.tags.clone());

// When the upper 64 bits are non-zero, ensure _dd.p.tid is present

// so that the full 128-bit trace id can be reconstructed downstream.

if upper_64 != 0 && !tags.contains_key("_dd.p.tid") {

let upper_hex = format!("{:016x}", upper_64);

tags.insert("_dd.p.tid".to_string(), upper_hex);

}

duncanista requested a review from a team as a code owner March 6, 2026 22:41

duncanista requested review from Copilot and jchrostek-dd March 6, 2026 22:41

Copilot started reviewing on behalf of duncanista March 6, 2026 22:41 View session

duncanista added 2 commits March 6, 2026 17:45

fix(traces): suppress clippy cast_possible_truncation for u128→u64 tr…

baf4f6f

…ace IDs The truncation is intentional — Datadog protocol transmits the lower 64 bits via x-datadog-trace-id, with upper bits in _dd.p.tid tag.

fmt

e03b162

Copilot AI reviewed Mar 6, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(traces): migrate trace propagation to dd-trace-rs#1089

feat(traces): migrate trace propagation to dd-trace-rs#1089
duncanista wants to merge 3 commits intomainfrom
jordan.gonzalez/propagation/migrate-to-dd-trace-rs

duncanista commented Mar 6, 2026

Uh oh!

duncanista commented Mar 6, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-            #[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits
-            {
-                context.invocation_span.trace_id = sc.trace_id as u64;
-            }
+            // Datadog protocol uses the lower 64 bits for the trace_id field; preserve full 128 bits via _dd.p.tid
+            let full_trace_id = sc.trace_id;
+            let lower_64_trace_id = (full_trace_id & 0xffff_ffff_ffff_ffff) as u64;
+            context.invocation_span.trace_id = lower_64_trace_id;
+            // Store full 128-bit trace id as 32-char hex in Datadog internal tag
+            context
+                .invocation_span
+                .meta
+                .entry("_dd.p.tid".to_string())
+                .or_insert_with(|| format!("{:032x}", full_trace_id));

-            #[allow(clippy::cast_possible_truncation)] // Datadog protocol uses lower 64 bits
-            {
-                trace_id = sc.trace_id as u64;
-            }
-            parent_id = sc.span_id;
-            tags.extend(sc.tags.clone());
+            // Datadog protocol uses the lower 64 bits for the trace id field.
+            let full_trace_id = sc.trace_id;
+            let lower_64 = (full_trace_id & ((1u128 << 64) - 1)) as u64;
+            let upper_64 = full_trace_id >> 64;
+            trace_id = lower_64;
+            parent_id = sc.span_id;
+            tags.extend(sc.tags.clone());
+            // When the upper 64 bits are non-zero, ensure _dd.p.tid is present
+            // so that the full 128-bit trace id can be reconstructed downstream.
+            if upper_64 != 0 && !tags.contains_key("_dd.p.tid") {
+                let upper_hex = format!("{:016x}", upper_64);
+                tags.insert("_dd.p.tid".to_string(), upper_hex);
+            }

Conversation

duncanista commented Mar 6, 2026

Overview

Testing

Uh oh!

duncanista commented Mar 6, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants