[OTLP EXPORTER] fix OTLP recordable attribute spec compliance by dbarker · Pull Request #4166 · open-telemetry/opentelemetry-cpp

dbarker · 2026-06-16T20:16:52Z

Fixes # (issue)

Fixes a few non-compliant areas of the otlp attribute spec 1) keys are not enforced to be unique, 2) byte values are not supported where required (due to an allow_bytes argument in the utils) .

See:

https://opentelemetry.io/docs/specs/otel/common/#attribute

Implementation MUST by default enforce that the exported attribute collections contain only unique keys.

https://opentelemetry.io/docs/specs/otel/common/#attribute-collections

An Attribute is a key-value pair, which MUST have the following properties:

The attribute key MUST be a non-null and non-empty string.
Case sensitivity of keys is preserved. Keys that differ in casing are treated as distinct keys.
The attribute value MUST be one of types defined in AnyValue.

Resources, Instrumentation Scopes, Metric points, Spans, Span Events, Span Links and Log Records, contain a collection of attributes.

Changes

Check if an attribute key exists and update its value on calls to SetAttribute for logs and spans
Support byte values for all OTLP AnyValue attributes (spans, logs, metrics, resource, ...)
Use a variant visitor to set the attribute values into the protobuf AnyValue message (replacing an if statement using nostd::get)
adds test for otlp_populate_attribute_utils

For significant contributions please make sure you have completed the following items:

CHANGELOG.md updated for non-trivial changes
Unit tests have been added
Changes in public API reviewed

…otlp attribute AnyValue types. Use a variant vistor instead of sequential if/else conditions to set otlp attributes. Populate otlp attribute and AnyValue tests.

codecov · 2026-06-16T21:06:24Z

Codecov Report

❌ Patch coverage is 92.19858% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.84%. Comparing base (de26178) to head (88e47f6).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...xporters/otlp/src/otlp_populate_attribute_utils.cc	90.75%	10 Missing ⚠️
sdk/src/logs/multi_recordable.cc	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4166      +/-   ##
==========================================
+ Coverage   82.83%   82.84%   +0.01%     
==========================================
  Files         406      406              
  Lines       16913    16951      +38     
==========================================
+ Hits        14009    14041      +32     
- Misses       2904     2910       +6

Files with missing lines	Coverage Δ
...opentelemetry/exporters/otlp/otlp_log_recordable.h	`100.00% <ø> (ø)`
...ude/opentelemetry/exporters/otlp/otlp_recordable.h	`100.00% <ø> (ø)`
exporters/otlp/src/otlp_log_recordable.cc	`40.13% <100.00%> (+4.08%)`	⬆️
exporters/otlp/src/otlp_metric_utils.cc	`92.03% <100.00%> (ø)`
exporters/otlp/src/otlp_recordable.cc	`91.06% <100.00%> (+0.80%)`	⬆️
sdk/src/trace/span.cc	`89.10% <100.00%> (+0.21%)`	⬆️
sdk/src/logs/multi_recordable.cc	`89.34% <50.00%> (-1.20%)`	⬇️
...xporters/otlp/src/otlp_populate_attribute_utils.cc	`91.72% <90.75%> (-1.23%)`	⬇️

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

dbarker · 2026-06-16T23:56:42Z

                                     const opentelemetry::common::AttributeValue &value) noexcept
 {
-  OtlpPopulateAttributeUtils::PopulateAttribute(proto_record_.add_attributes(), key, value, true);
+  for (auto &attr : *proto_record_.mutable_attributes())


Benchmark comparison:

This check to avoid duplicating attribute keys adds overhead:

~2.3% (~14ns) overhead to nominal spans with 6 attributes (OtlpSpanFixture/Nominal)

~136% (10us) to spans with max (128) attributes (OtlpSpanFixture/MaxAttributes)

This can be mitigated a few ways 1) add a SetAttributes method that takes key value iterable and does not check existing keys on creation of the span, 2) do not check attribute keys during span recording and perform deduplication when creating the protobuf request message on export.

branch: dbarker:add_otlp_recordable_benchmark (#4165)

----------------------------------------------------------------------------------- Benchmark Time CPU Iterations ----------------------------------------------------------------------------------- OtlpSpanFixture/Minimal_mean 342 ns 345 ns 5 OtlpSpanFixture/Minimal_median 343 ns 345 ns 5 OtlpSpanFixture/Minimal_stddev 2.23 ns 1.88 ns 5 OtlpSpanFixture/Minimal_cv 0.65 % 0.55 % 5 OtlpSpanFixture/Nominal_mean 596 ns 598 ns 5 OtlpSpanFixture/Nominal_median 594 ns 596 ns 5 OtlpSpanFixture/Nominal_stddev 10.5 ns 10.6 ns 5 OtlpSpanFixture/Nominal_cv 1.76 % 1.77 % 5 OtlpSpanFixture/MaxAttributes_mean 7517 ns 7521 ns 5 items_per_second=17.0197M/s OtlpSpanFixture/MaxAttributes_median 7489 ns 7492 ns 5 items_per_second=17.0841M/s OtlpSpanFixture/MaxAttributes_stddev 87.8 ns 87.8 ns 5 items_per_second=196.123k/s OtlpSpanFixture/MaxAttributes_cv 1.17 % 1.17 % 5 items_per_second=1.15% BM_PopulateRequest/span_count:1_mean 0.378 us 0.378 us 5 items_per_second=2.6486M/s BM_PopulateRequest/span_count:1_median 0.378 us 0.378 us 5 items_per_second=2.64307M/s BM_PopulateRequest/span_count:1_stddev 0.002 us 0.002 us 5 items_per_second=11.8787k/s BM_PopulateRequest/span_count:1_cv 0.45 % 0.45 % 5 items_per_second=0.45% BM_PopulateRequest/span_count:512_mean 25.5 us 25.5 us 5 items_per_second=20.0764M/s BM_PopulateRequest/span_count:512_median 25.4 us 25.4 us 5 items_per_second=20.1812M/s BM_PopulateRequest/span_count:512_stddev 0.248 us 0.248 us 5 items_per_second=194k/s BM_PopulateRequest/span_count:512_cv 0.97 % 0.97 % 5 items_per_second=0.97%

branch: dbarker:fix_otlp_attribute_spec_compliance (this PR #4166)

----------------------------------------------------------------------------------- Benchmark Time CPU Iterations ----------------------------------------------------------------------------------- OtlpSpanFixture/Minimal_mean 340 ns 343 ns 5 OtlpSpanFixture/Minimal_median 340 ns 344 ns 5 OtlpSpanFixture/Minimal_stddev 0.711 ns 0.795 ns 5 OtlpSpanFixture/Minimal_cv 0.21 % 0.23 % 5 OtlpSpanFixture/Nominal_mean 610 ns 614 ns 5 OtlpSpanFixture/Nominal_median 607 ns 611 ns 5 OtlpSpanFixture/Nominal_stddev 6.20 ns 6.58 ns 5 OtlpSpanFixture/Nominal_cv 1.02 % 1.07 % 5 OtlpSpanFixture/MaxAttributes_mean 17773 ns 17776 ns 5 items_per_second=7.20295M/s OtlpSpanFixture/MaxAttributes_median 17727 ns 17730 ns 5 items_per_second=7.21921M/s OtlpSpanFixture/MaxAttributes_stddev 337 ns 336 ns 5 items_per_second=134.931k/s OtlpSpanFixture/MaxAttributes_cv 1.89 % 1.89 % 5 items_per_second=1.87% BM_PopulateRequest/span_count:1_mean 0.349 us 0.349 us 5 items_per_second=2.86195M/s BM_PopulateRequest/span_count:1_median 0.349 us 0.349 us 5 items_per_second=2.86567M/s BM_PopulateRequest/span_count:1_stddev 0.001 us 0.001 us 5 items_per_second=10.7304k/s BM_PopulateRequest/span_count:1_cv 0.37 % 0.38 % 5 items_per_second=0.37% BM_PopulateRequest/span_count:512_mean 26.1 us 26.1 us 5 items_per_second=19.5965M/s BM_PopulateRequest/span_count:512_median 26.1 us 26.1 us 5 items_per_second=19.5943M/s BM_PopulateRequest/span_count:512_stddev 0.104 us 0.104 us 5 items_per_second=77.6371k/s BM_PopulateRequest/span_count:512_cv 0.40 % 0.40 % 5 items_per_second=0.40%

marcalff · 2026-06-17T21:01:29Z

Implementation MUST by default enforce that the exported attribute collections contain only unique keys.

My reading is as follows:

(a) Implementation MUST have code to enforce that the exported attribute collections contain only unique keys.
(b) Such code is executed by default, but there is a way to disable it.

For applications that trust the instrumentation to be correct, having the de duplication logic is overhead.

This hints at the existence of a configuration parameter, which is not defined in the spec.

Implementations MAY have an option to allow exporting attribute collections with duplicate keys (e.g. for better performance).

owent · 2026-06-18T07:48:52Z

-  OtlpPopulateAttributeUtils::PopulateAttribute(proto_record_.add_attributes(), key, value, true);
+  for (auto &attr : *proto_record_.mutable_attributes())
+  {
+    if (attr.key() == key)


Can we use a hash map (e.g., unordered_map) to track existing keys? The current linear scan on every SetAttribute call is O(n) per insertion—O(n²) total—and the overhead compounds quickly at scale.

It's not just the number of attributes: string comparison cost grows with key length. Each loop iteration does a full character-by-character comparison, whereas a hash map typically checks hash codes first, only falling back to full comparison on collisions.

88e47f6

This implements the hash map for attribute keys.

Compared to the linear search (1525b4c), the performance is better in the max attributes case by -5.5us or -31%, but worse for the nominal attribute case (6 attributes) by ~220 ns or +37%.

Benchmark main linear search (1525b4c) hash map (88e47f6)

Minimal 342 ns 340 ns 359 ns

Nominal 596 ns 610 ns 835 ns

MaxAttributes 7,517 ns 17,773 ns 12,256 ns

PopulateRequest/1 378 ns 349 ns 367 ns

PopulateRequest/512 25.5 µs 26.1 µs 26.0 µs

----------------------------------------------------------------------------------- Benchmark Time CPU Iterations ----------------------------------------------------------------------------------- OtlpSpanFixture/Minimal_mean 359 ns 361 ns 5 OtlpSpanFixture/Minimal_median 359 ns 361 ns 5 OtlpSpanFixture/Minimal_stddev 0.440 ns 0.477 ns 5 OtlpSpanFixture/Minimal_cv 0.12 % 0.13 % 5 OtlpSpanFixture/Nominal_mean 835 ns 839 ns 5 OtlpSpanFixture/Nominal_median 834 ns 839 ns 5 OtlpSpanFixture/Nominal_stddev 5.53 ns 5.61 ns 5 OtlpSpanFixture/Nominal_cv 0.66 % 0.67 % 5 OtlpSpanFixture/MaxAttributes_mean 12256 ns 12261 ns 5 items_per_second=10.4397M/s OtlpSpanFixture/MaxAttributes_median 12233 ns 12238 ns 5 items_per_second=10.4593M/s OtlpSpanFixture/MaxAttributes_stddev 62.0 ns 61.9 ns 5 items_per_second=52.5819k/s OtlpSpanFixture/MaxAttributes_cv 0.51 % 0.50 % 5 items_per_second=0.50% BM_PopulateRequest/span_count:1_mean 0.367 us 0.367 us 5 items_per_second=2.72565M/s BM_PopulateRequest/span_count:1_median 0.367 us 0.367 us 5 items_per_second=2.72547M/s BM_PopulateRequest/span_count:1_stddev 0.001 us 0.001 us 5 items_per_second=4.4662k/s BM_PopulateRequest/span_count:1_cv 0.16 % 0.16 % 5 items_per_second=0.16% BM_PopulateRequest/span_count:512_mean 26.0 us 26.0 us 5 items_per_second=19.6842M/s BM_PopulateRequest/span_count:512_median 26.0 us 26.0 us 5 items_per_second=19.7045M/s BM_PopulateRequest/span_count:512_stddev 0.212 us 0.212 us 5 items_per_second=160.233k/s BM_PopulateRequest/span_count:512_cv 0.82 % 0.82 % 5 items_per_second=0.81%

…/empty keys and add test.

ensure unique keys for otlp attributes, always allow bytes values in …

1525b4c

…otlp attribute AnyValue types. Use a variant vistor instead of sequential if/else conditions to set otlp attributes. Populate otlp attribute and AnyValue tests.

fix ci issues

de16c15

dbarker commented Jun 16, 2026

View reviewed changes

owent reviewed Jun 18, 2026

View reviewed changes

use unordered map to track otlp recordable attributes. check for null…

88e47f6

…/empty keys and add test.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[OTLP EXPORTER] fix OTLP recordable attribute spec compliance#4166

[OTLP EXPORTER] fix OTLP recordable attribute spec compliance#4166
dbarker wants to merge 3 commits into
open-telemetry:mainfrom
dbarker:fix_otlp_attribute_spec_compliance

dbarker commented Jun 16, 2026

Uh oh!

codecov Bot commented Jun 16, 2026 •

edited

Loading

Uh oh!

dbarker Jun 16, 2026 •

edited

Loading

Uh oh!

marcalff commented Jun 17, 2026 •

edited

Loading

Uh oh!

owent Jun 18, 2026

Uh oh!

dbarker Jun 18, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Benchmark	main	linear search (`1525b4c`)	hash map (`88e47f6`)
Minimal	342 ns	340 ns	359 ns
Nominal	596 ns	610 ns	835 ns
MaxAttributes	7,517 ns	17,773 ns	12,256 ns
PopulateRequest/1	378 ns	349 ns	367 ns
PopulateRequest/512	25.5 µs	26.1 µs	26.0 µs

Conversation

dbarker commented Jun 16, 2026

Changes

Uh oh!

codecov Bot commented Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

dbarker Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Benchmark comparison:

branch: dbarker:add_otlp_recordable_benchmark (#4165)

branch: dbarker:fix_otlp_attribute_spec_compliance (this PR #4166)

Uh oh!

marcalff commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

owent Jun 18, 2026

Choose a reason for hiding this comment

Uh oh!

dbarker Jun 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov Bot commented Jun 16, 2026 •

edited

Loading

dbarker Jun 16, 2026 •

edited

Loading

branch: `dbarker:add_otlp_recordable_benchmark` (#4165)

branch: `dbarker:fix_otlp_attribute_spec_compliance` (this PR #4166)

marcalff commented Jun 17, 2026 •

edited

Loading

dbarker Jun 18, 2026 •

edited

Loading