[2/2] Expand OTLP metrics temporality preference coverage

[vpellan]

Motivation

APMAPI-2555: extend the existing three temporality cases to cover the recognized values, parsing, and defaults specified by the OTel metrics exporter specification.

Stack 2/2: depends on #7925, which moves the original code without changing test behavior. Review and merge #7925 first. This PR targets vpellan/move-otel-metrics-temporality-tests so its diff contains the added coverage and related declarations.

Changes

• Expand coverage from 3 to 16 parameterized cases: Cumulative, Delta, and LowMemory in lowercase, uppercase, and mixed case; explicit empty and invalid input; and separate Datadog-default and OTel-specification-default checks.
• Assert exported temporality for synchronous/asynchronous counters, up-down counters, and histograms. Keep the original detailed payload assertions in a separate regression method, including gauges, values, attributes, timestamps, and histogram buckets.
• Neutralize conflicting temporality settings, pin HTTP/protobuf, and disable unrelated metric/log sources. Leave OTEL_TRACES_EXPORTER unset because none disables the entire Node.js tracer, including metrics.
• Rename the class for the environment variable and associate it with Feature Health 645.
• Replace the carried-over activation declarations with registry-based support boundaries and narrower declarations for known limitations. PHP is incomplete_test_app because metric instrument endpoints are absent; C/C++ remain missing_feature.
• Split mixed-case Cumulative and Delta so Ruby can declare only its failing cases. Track .NET, Rust, Python, and Ruby gaps through language subtasks of APMAPI-2562; keep unaffected cases enabled.
• Fix an unrelated trace-filter race found in CI in a separate commit: wait for the expected trace instead of reading immediately. The C++ failure artifact shows the trace arriving after the empty read, and the same test passes on the base PR.

The registry descriptions are stale: LowMemory works in Node.js, Java, and Ruby (lowercase/uppercase), and PHP source supplies a Datadog Delta default despite the registry listing Cumulative. The local feature-map CSV lacks this newer variable; the exact ticket and live Feature Health entry independently confirm ID 645 and APMAPI-2555.

CI findings and tracking

The CI run on 6edd3f5c9 exposed 22 temporality failures across eight production/development jobs, plus the C++ trace-arrival race. Laravel 11 security build failures were excluded from this work.

Language	CI versions	Gap	Defect
.NET	3.54.0 / 3.55.0	All three LowMemory spellings export Delta for asynchronous counters instead of Cumulative	APMAPI-2563
Rust	0.5.2 / 0.5.3-dev+79746e6	Same LowMemory asynchronous-counter gap	APMAPI-2564
Python	4.15.4 / 4.16.0-rc1	Invalid input exports Cumulative instead of the unset Delta default	APMAPI-2565
Ruby	2.43.0 / 2.44.0-dev	Mixed-case Delta/LowMemory and empty/invalid input export Cumulative	APMAPI-2566

Nine method-level bug declarations reference these subtasks. Assertions still enforce the specification; documented Datadog-default differences retain their separate declarations.

Validation

• ./format.sh: passed after the final manifest edits, including typing, import policy, formatting, and manifest validation. The unrelated empty, untracked utils/build/docker/internal_server/app.sh directory was temporarily moved for shellcheck and restored.
• ./run.sh TEST_THE_TEST: 635 passed, 1 xfailed after the test split and trace-wait fix. Re-ran manifest/convention checks after the final declarations: 33 passed.
• Python runner: 3.12.4. Targeted PARAMETRIC runs used existing images with --skip-parametric-build.
• Ruby 2.42.0-dev with the final declarations: 8 passed, 8 expected failures, no ordinary failures. The four new defect cases carry their dedicated bug reference; passing Cumulative/stable/default cases stay enabled.
• Node.js 7.0.0-pre: 16 passed, 1 expected failure, covering all 16 temporality cases plus the trace-filter test.
• Earlier targeted checks: Java 1.66.0+a099fffb31 had 15 passed / 1 expected failure; Go had 12 passed / 4 expected failures.
• .NET, Rust, and Python declarations are based on the linked CI evidence; those language suites were not rerun locally. No C++ image was available locally, so the bounded-wait fix was exercised with Node.js and requires C++ CI confirmation.

No live Feature Health evidence was changed.

[github-actions]

CODEOWNERS have been resolved as:

manifests/c.yml                                                         @DataDog/system-tests-reviewers
manifests/cpp.yml                                                       @DataDog/system-tests-reviewers
manifests/dotnet.yml                                                    @DataDog/system-tests-reviewers
manifests/golang.yml                                                    @DataDog/system-tests-reviewers
manifests/java.yml                                                      @DataDog/system-tests-reviewers
manifests/nodejs.yml                                                    @DataDog/system-tests-reviewers
manifests/php.yml                                                       @DataDog/system-tests-reviewers
manifests/python.yml                                                    @DataDog/system-tests-reviewers
manifests/ruby.yml                                                      @DataDog/system-tests-reviewers
manifests/rust.yml                                                      @DataDog/system-tests-reviewers
tests/parametric/otel_env_vars/test_otel_exporter_otlp_metrics_temporality_preference.py  @DataDog/system-tests-reviewers
tests/parametric/test_trace_filters.py                                  @DataDog/system-tests-reviewers
utils/_features.py                                                      @DataDog/system-tests-core

[datadog-prod-us1-3]

 

❌ Errors

Your PR has failed checks. Please review the issues below and take necessary action before merging.

🚦 110 Pipeline jobs failed

Testing the test | System Tests (cpp_httpd, dev) / End-to-end #1 / httpd 1

View more details · View in GitHub Actions

Testing the test | System Tests (cpp_httpd, prod) / End-to-end #1 / httpd 1

View more details · View in GitHub Actions

Testing the test | System Tests (cpp_kong, dev) / End-to-end #1 / kong 1

View more details · View in GitHub Actions

View all 110 failed jobs.

ℹ️ Info

No other issues found (see more)

🧪 All tests passed
❄️ No new flaky tests detected

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 8b6e312 | Docs | View more details | Give us feedback!}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-02T14:14:56.579105Z	25f5c05	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 25f5c0585e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Declare the known Ruby failures in the manifest

In the Ruby 2.42.0-dev validation recorded for this change, DeLtA, mixed-case LowMemory, empty, and invalid inputs all fail, but this class-level activation leaves those cases enabled; only the specification-default and payload methods are deactivated below. Consequently the Ruby parametric job gets four ordinary failures instead of expected failures. Split the mixed-case method if necessary and add narrow manifest declarations for the affected cases before activating the expanded suite.

AGENTS.md reference: AGENTS.md:L33-L35

Useful? React with 👍 / 👎.

[MilanGarnier]

nit: I'd recommend to also keep the @features.otel_metrics_api decorator and add the env var one next to it

LGTM for what's being tested