feat(scheduler): register drift metric calculators with MetricsDirectory

[SudipSinha]

Summary

All 8 Prometheus metric tests timeout because drift metric calculators are never registered with MetricsDirectory. The scheduler finds scheduled requests but logs "No calculator found for metric CompareMeans" every 5 seconds indefinitely.

Root cause

The MetricsDirectory requires each metric to register a calculator function at module load time. Only fairness metrics (SPD, DIR) and Batch Mean registered calculators. None of the drift metrics did.

The fix

Add async calculator functions and module-level registration for three drift metrics:

• CompareMeans — calculate_compare_means_metric using CompareMeans.ttest_ind
• KSTest — calculate_kstest_metric using KolmogorovSmirnov.kstest
• JensenShannon — calculate_jensenshannon_metric using JensenShannon.jensenshannon

Each calculator:

1. Receives batch (current data from scheduler) and request (with reference_tag, fit_columns)
2. Fetches reference data via data_source.get_dataframe_by_tag(model_id, reference_tag)
3. Computes per-feature metric values
4. Returns MetricValueCarrier(named_values) for Prometheus publishing

Since drift calculators need async get_dataframe_by_tag, _calculate_metric in the scheduler is now async. It detects whether a calculator returns a coroutine and awaits it if so — sync calculators (SPD, DIR) still work unchanged.

Files changed

File	Change
src/endpoints/metrics/drift/compare_means.py	Add calculator + registration
src/endpoints/metrics/drift/kolmogorov_smirnov.py	Add calculator + registration
src/endpoints/metrics/drift/jensen_shannon.py	Add calculator + registration
src/service/prometheus/prometheus_scheduler.py	Make _calculate_metric async, add asyncio import
tests/endpoints/metrics/drift/test_drift_calculators.py	4 tests: per-feature values for all 3 metrics + column derivation from batch

Test plan

• All 446 tests pass (4 new calculator tests)
• Calculator tests verify MetricValueCarrier returns per-feature named values
• test_derives_columns_from_batch verifies empty fit_columns falls back to batch columns
• ruff check clean (no noqa needed)
• ruff format clean
• pyrefly check clean
• Deploy and verify GET /q/metrics shows drift metric gauge values after scheduled computation

Note

The streaming KS test (KSTestStreaming) and MMD calculators need the same treatment on their respective feature branches (PRs #99 and #149).

🤖 Generated with Claude Code

[coderabbitai]

Warning

Review limit reached

@SudipSinha, we couldn't start this review because you've reached your PR review rate limit.

More reviews will be available in 33 minutes and 26 seconds. Learn how PR review limits work.

Your organization has used up its prepaid credits, and credit purchases are no longer available. Enable the review add-on in the billing tab to keep reviews running — you're only billed for reviews past your plan's rate limits ($0.25/file).

⌛ How to resolve this issue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based credits.

🚦 How do rate limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan refill rate.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, the refill rate gradually slows as usage increases. The highest same-day bursts are limited more strictly.

Please see our Fair Usage Limits Policy for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: f79f1929-8a19-4fcc-a50b-367c2bb3a934

📥 Commits

Reviewing files that changed from the base of the PR and between 1e3e6d2 and 585a908.

📒 Files selected for processing (7)

• src/endpoints/metrics/drift/compare_means.py
• src/endpoints/metrics/drift/jensen_shannon.py
• src/endpoints/metrics/drift/kolmogorov_smirnov.py
• src/service/prometheus/prometheus_scheduler.py
• tests/endpoints/metrics/drift/test_compare_means.py
• tests/endpoints/metrics/drift/test_drift_calculators.py
• tests/endpoints/metrics/drift/test_jensen_shannon.py

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/register-drift-metric-calculators

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

PR image build and manifest generation completed successfully!

📦 PR image: quay.io/trustyai/trustyai-service-python-ci:585a9089fc8538ab6a3c6484634812b29aeb8c6c

🗂️ CI manifests

devFlags:
  manifests:
    - contextDir: config
      sourcePath: ''
      uri: https://api.github.com/repos/trustyai-explainability/trustyai-service-operator-ci/tarball/service-python-585a9089fc8538ab6a3c6484634812b29aeb8c6c