feat(tracing): OTLP trace metrics export [DO NOT MERGE]

[mabdinur]

Warning

DO NOT MERGE — pins libdatadog to a feature branch. Companion: DataDog/libdatadog#2067.

Overview

Wires the libdatadog OTLP trace-metrics export into dd-trace-py. When enabled, eligible spans are aggregated natively in Rust into a traces.span.sdk.metrics.duration histogram and exported to a /v1/metrics endpoint alongside traces. dd-trace-py's role is configuration only — stats aggregation, OTLP encoding, and HTTP export all run inside libdatadog.

Non-obvious details for reviewers

• Enabled via OTEL_TRACES_SPAN_METRICS_ENABLED. Auto-enables when both OTEL_TRACES_EXPORTER=otlp and DD_METRICS_OTEL_ENABLED=true are set.
• Flush cadence is fixed at 10s and is not driven by OTEL_METRIC_EXPORT_INTERVAL. Use _DD_TRACE_METRICS_OTEL_FLUSH_INTERVAL (ms) to shorten it in tests.
• OTEL_EXPORTER_OTLP_METRICS_ENDPOINT is passed to libdatadog verbatim — it must include /v1/metrics; it is not auto-suffixed.
• _dd.stats_computed: "true" is set on every outbound OTLP trace ResourceSpans to tell Datadog Agent OTLP receivers to skip their concentrator and prevent double-counted APM metrics. Backwards compatible with released Agents.
• src/native/Cargo.lock is pinned to munir/otlp-trace-metrics branch head (9cd1ad5).

Testing: tests/opentelemetry/test_otlp_trace.py::test_otlp_trace_metrics_exported_via_http. Cross-tracer coverage in DataDog/system-tests parametric suite.

[cit-pr-commenter-54b7da]

Codeowners resolved as

src/native/Cargo.lock                                                   @DataDog/apm-core-python

[datadog-datadog-prod-us1]

 

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 15 Pipeline jobs failed

DataDog/apm-reliability/dd-trace-py | contrib/opentelemetry 1/4     

🧪 1 Test failed

test_otlp_trace_metrics_exported_via_http[py3.12] from test_otlp_trace.py   (Fix with Cursor)

Expected status 0, got 1.
=== Captured STDOUT ===
=== End of captured STDOUT ===
=== Captured STDERR ===
Traceback (most recent call last):
  File "tests/opentelemetry/test_otlp_trace.py", line 139, in <module>
    assert metrics_payload is not None, "No OTLP metrics payload received by mock server"
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: No OTLP metrics payload received by mock server
=== End of captured STDERR ===

DataDog/apm-reliability/dd-trace-py | contrib/opentelemetry 2/4     

🧪 1 Test failed

test_otlp_trace_metrics_exported_via_http[py3.12] from test_otlp_trace.py   (Fix with Cursor)

Expected status 0, got 1.
=== Captured STDOUT ===
=== End of captured STDOUT ===
=== Captured STDERR ===
Traceback (most recent call last):
  File "tests/opentelemetry/test_otlp_trace.py", line 139, in <module>
    assert metrics_payload is not None, "No OTLP metrics payload received by mock server"
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: No OTLP metrics payload received by mock server
=== End of captured STDERR ===

DataDog/apm-reliability/dd-trace-py | contrib/opentelemetry 3/4     

🧪 1 Test failed

test_otlp_trace_metrics_exported_via_http[py3.12] from test_otlp_trace.py   (Fix with Cursor)

Expected status 0, got 1.
=== Captured STDOUT ===
=== End of captured STDOUT ===
=== Captured STDERR ===
Traceback (most recent call last):
  File "tests/opentelemetry/test_otlp_trace.py", line 139, in <module>
    assert metrics_payload is not None, "No OTLP metrics payload received by mock server"
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: No OTLP metrics payload received by mock server
=== End of captured STDERR ===

View all 15 failed jobs.

ℹ️ Info

No other issues found (see more)

❄️ No new flaky tests detected

🔄 Datadog retried 6 tests - 0 passed on retry

🚧 19 tests that failed were ignored due to quarantine

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: c374358 | Docs | Datadog PR Page | Give us feedback!}

[pr-commenter]

Benchmarks

Benchmark execution time: 2026-06-16 01:15:01

Comparing candidate commit c374358 in PR branch munir/otlp-trace-metrics-export with baseline commit fd67a37 in branch main.

Found 0 performance improvements and 5 performance regressions! Performance is the same for 612 metrics, 10 unstable metrics.

scenario:iastaspects-stringio_noaspect

• 🟥 execution_time [+42.099µs; +48.949µs] or [+11.988%; +13.939%]

scenario:iastaspectsospath-ospathbasename_aspect

• 🟥 execution_time [+93.789µs; +101.051µs] or [+21.800%; +23.487%]

scenario:span-start

• 🟥 execution_time [+1.150ms; +1.290ms] or [+7.359%; +8.254%]

scenario:startup-ddtrace_run_send_span

• 🟥 execution_time [+1.602s; +1.605s] or [+86.173%; +86.351%]

scenario:telemetryaddmetric-1-count-metric-1-times

• 🟥 execution_time [+294.949ns; +333.579ns] or [+14.230%; +16.094%]