fix(observability): demote transient OpenAI embeddings 429s to expected and reduce Sentry noise

[YellowSnnowmann]

Summary

• Updated OpenAI embeddings error reporting to route non-2xx failures through report_error_or_expected instead of always emitting hard error events.

• Standardized embedding HTTP error text to canonical format: Embedding API error (): .

• Ensured transient upstream statuses (especially 429 Too Many Requests) match observability classifiers and are demoted to warning breadcrumbs.

• Added a focused regression test for 429 formatting/classification to prevent future Sentry-noise regressions.

Problem

• OpenAI embeddings calls can return transient 429 rate-limit responses during normal load.

• These errors are retried by memory-tree/background flows, but each attempt was still producing Sentry error events.

• The previous message shape (Embedding API error : ...) did not consistently match the transient HTTP classifier expectations, increasing alert noise and masking actionable failures.

Solution

• Switched embeddings non-2xx reporting to report_error_or_expected(...), allowing transient failures to be classified as expected noise.

• Changed the emitted error string to canonical classifier-compatible shape: Embedding API error (): .

• Kept failure tags (model, status, failure=non_2xx) for diagnostics while reducing noisy escalation.

• Added embed_429_uses_canonical_transient_format test to assert:

  • message contains (429 Too Many Requests), and
  • is_transient_message_failure(...) returns true.

Submission Checklist

• If a section does not apply to this change, mark the item as N/A with a one-line reason. Do not delete items.

• Tests added or updated (happy path + at least one failure / edge case) per Testing Strategy

• Diff coverage ≥ 80% — changed lines (Vitest + cargo-llvm-cov merged via diff-cover) meet the gate enforced by .github/workflows/coverage.yml. Run pnpm test:coverage and pnpm test:rust locally; PRs below 80% on changed lines will not merge.

• Coverage matrix updated — added/removed/renamed feature rows in docs/TEST-COVERAGE-MATRIX.md reflect this change (or N/A: behaviour-only change)

• All affected feature IDs from the matrix are listed in the PR description under ## Related (N/A if no feature IDs are affected)

• No new external network dependencies introduced (mock backend used per Testing Strategy)

• Manual smoke checklist updated if this touches release-cut surfaces (docs/RELEASE-MANUAL-SMOKE.md) (N/A if not release-cut)

• Linked issue closed via Closes #NNN in the ## Related section (N/A until issue is linked)

Impact

• Runtime/platform impact: Rust core observability path only (desktop app behavior unchanged for users except reduced false-positive error noise).

• Performance: negligible runtime overhead change; likely reduced observability/event volume under transient rate limiting.

• Security/migration/compatibility: no migration or API compatibility changes; no new dependencies introduced.

Related

• Closes: OPENHUMAN-TAURI-RQ

Summary by CodeRabbit

• Bug Fixes

  • Standardized error message formatting for embedding API failures with clearer HTTP status code display
  • Enhanced error classification for transient upstream API failures

• Tests

  • Added test coverage for rate-limiting error handling to validate transient error classification

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 1aae62a9-ebb8-44a8-a223-20ec141d6051

📥 Commits

Reviewing files that changed from the base of the PR and between 401bfe6 and 6acfcd0.

📒 Files selected for processing (1)

• src/openhuman/embeddings/openai_tests.rs

---

📝 Walkthrough

Walkthrough

Updates OpenAI embedding non-2xx error formatting to include parentheses around HTTP status and clarifies transient reporting via report_error_or_expected; adds a Tokio test asserting HTTP 429 embedding failures produce the canonical transient-format message and are classified as transient.

Changes

Embedding API transient error handling

Layer / File(s)	Summary
Error message format and transient reporting docs src/openhuman/embeddings/openai.rs	Error message format changed to Embedding API error ({status}): {text} and comments updated to indicate report_error_or_expected demotes transient upstream HTTP failures into warning/breadcrumb logging.
HTTP 429 transient error classification test src/openhuman/embeddings/openai_tests.rs	New async test embed_429_uses_canonical_transient_format mocks a 429 OpenAI embeddings response and verifies the error string matches the canonical transient HTTP shape and is_transient_message_failure returns true.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Possibly related PRs

• tinyhumansai/openhuman#2190: Also updates OpenAI embedding non-2xx handling to use the transient reporting path and adds tests asserting error message shapes (400 case).
• tinyhumansai/openhuman#2216: Related edits routing OpenAI embedding errors through report_error_or_expected with tests for expected/transient classifications.

Suggested reviewers

• senamakel
• graycyrus

Poem

🐰 A parenthesis snug around a status line,
Four-twenty-nine tiptoes, labeled benign.
The test gives a nod with a breadcrumb cheer,
No alarms—just a warning, soft and clear.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and specifically summarizes the main change: demotion of transient OpenAI 429 errors to reduce Sentry noise through use of report_error_or_expected, which matches the core changes in both the embeddings handler and the regression test.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

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

[graycyrus]

Nice, focused fix — the parenthesized format is clearly needed for the is_transient_upstream_http_message classifier (line 240 in observability.rs: lower.contains(&format!("api error ({code}"))) and the test is a welcome addition.

One thing worth tightening up on the test — see inline comment.

[graycyrus]

Re-review — all prior changes addressed.

The new commit (6acfcd02) adds the api error (429 substring assertion I flagged in the prior review. The test now has three layers of defense:

1. Canonical (429 Too Many Requests) shape check
2. Classifier-arm substring guard (api error (429 )
3. Broad is_transient_message_failure classifier pass

This is a clean, focused fix — format string tweak + solid regression test. No new issues found.

File	Change
openai.rs	Format {status} → ({status}) in error string + updated comment
openai_tests.rs	New embed_429_uses_canonical_transient_format test with three assertions

[graycyrus]

Looks good, nice work!

[M3gA-Mind]

Looks good, nice work!