Decouple flag eval metrics from CoreMetricCollector

[sarahchen6]

What Does This Do

This PR moves flag evaluation recording out of CoreMetricCollector and into product-owned FlagEvaluationMetrics using :products:metrics:metrics-api’s Accumulator for fixed counters by removing CoreMetricCollector.count() and instead adding a periodic telemetry adapter, preserving metric names, tags, namespace, and type.

Motivation

#11639 has the flag evaluation writer push metrics directly into CoreMetricCollector; however, this breaks the design of CoreMetricCollector which is to drain metrics for delivery (not capture / push metrics). It also coupled product-specific metric recording to a shared internal telemetry component - which we want to avoid.

The changes in this PR restore the ownership pattern for flag evaluation based on existing usages such as by SpanMetricsImpl and BaggageMetrics. Now feature flagging owns its counters and telemetry drains them for delivery.

Additional Notes

With these changes, truncation counts are collectible without waiting for an EVP writer flush, and collection timestamps follow telemetry’s schedule. Queue saturation keeps pending snapshots for later collection, which is similar to StatsDCountReporter. Besides this telemetry collection timing and retention, flag evaluation results and event payloads are unchanged.

Contributor Checklist

• Format the title according to the contribution guidelines
• Assign the type: and (comp: or inst:) labels in addition to any other useful labels
• Avoid using close, fix, or any linking keywords when referencing an issue
  Use solves instead, and assign the PR milestone to the issue
• Update the CODEOWNERS file on source file addition, migration, or deletion
• Update public documentation with any new configuration flags or behaviors
• Once approved, use merge queue to merge the PR

Jira ticket: [PROJ-IDENT]

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	14.09 s	14.00 s	[-0.0%; +1.4%] (no difference)
startup:insecure-bank:tracing:Agent	12.95 s	12.96 s	[-0.7%; +0.7%] (no difference)
startup:petclinic:appsec:Agent	17.20 s	17.04 s	[-0.0%; +1.8%] (no difference)
startup:petclinic:iast:Agent	16.99 s	17.16 s	[-1.8%; -0.1%] (maybe better)
startup:petclinic:profiling:Agent	16.68 s	16.61 s	[-0.9%; +1.8%] (no difference)
startup:petclinic:sca:Agent	17.13 s	16.95 s	[+0.1%; +2.1%] (maybe worse)
startup:petclinic:tracing:Agent	16.25 s	16.22 s	[-0.7%; +1.0%] (no difference)

Commit: 4a854c33 · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.

[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 2026-10-01T15:05:36.600308Z	236d499	Draft marked ready
🔒 Security Review	✅ Completed 2026-10-01T15:04:41.421280Z	236d499	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.

[datadog-official]

More details

Producer-owned flag-evaluation counters preserve existing metric wire metadata and retain pending snapshots when the telemetry queue is full.

Was this helpful? React 👍 or 👎

_{Open Bits AI session}

_{🤖 Bits Code Review · Commit 236d499 · @DataDog review to ask questions}

[PerfectSlayer]

💭 thought: ‏There might be a dependency issue. The FFE bootstrap module now has dependency to the metrics api. (Which might relocated it into the bootstrap? 🤔 )
Same with the telemetry module which gains a dependency to a product (FFE).

I would stick to the original design having a collector in internal-api module (not great but that's the current design pattern for debugger, iast, llmobs, WAF, etc...) to avoid adding such dependencies. WDYT?

[sarahchen6]

I would stick to the original design having a collector in internal-api module (not great but that's the current design pattern for debugger, iast, llmobs, WAF, etc...) to avoid adding such dependencies. WDYT?

Makes sense! I think this should follow other product collector formats as well as minimize shared module dependencies on products... I'll refactor