Improve llm global rate limit test

[ashera96]

Purpose

• Resolves: [Bug]: advanced-ratelimit: in-memory bucket for an apiname key is non-deterministically split per-route instead of shared across an API's operations #2317

Changes introduced via this PR

• Reverts the test which was commented out via [1] due to an intermittent test failure.
• Intermittent failure was noticed due to an issue in the advanced-ratelimit policy which is fixed via [2].

1. Fix intermittent ratelimit test failure #2306
2. fix(ratelimit): key stale-limiter cleanup by attachment level to stop cross-route bucket eviction gateway-controllers#216

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 8c8f29c1-c93a-45f1-91e4-7acf755edecb

📥 Commits

Reviewing files that changed from the base of the PR and between 737bef5 and ebbdcc1.

📒 Files selected for processing (1)

• gateway/it/features/llm-provider-global-ratelimit.feature

---

📝 Walkthrough

Walkthrough

Adds two Gherkin scenarios (9 and 10) to an existing feature file, testing that a global apiname-keyed advanced-ratelimit bucket is shared across API operations for both provider and proxy contexts, even when an operation-level rate-limit policy rejects requests on one endpoint first.

Changes

Mixed global/operation rate-limit scenarios

Layer / File(s)	Summary
Provider mixed rate-limit scenario gateway/it/features/llm-provider-global-ratelimit.feature	Scenario 9 combines a global apiname-keyed advanced-ratelimit policy with an operation-level basic-ratelimit on /chat/completions for a provider, asserting /embeddings also returns 429 once the shared global bucket is exhausted.
Proxy mixed rate-limit scenario gateway/it/features/llm-provider-global-ratelimit.feature	Scenario 10 applies the same mixed policy configuration to an LLM proxy context, asserting identical 429 exhaustion behavior on /embeddings after /chat/completions is rejected by the operation cap.

Estimated code review effort: 1 (Trivial) | ~5 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description only covers Purpose and a brief change note; most required sections from the template are missing.	Add the missing Goals, Approach, User stories, Documentation, Automation tests, Security checks, Samples, Related PRs, and Test environment sections.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise and matches the main change: improving the LLM global rate-limit test.
Linked Issues check	✅ Passed	The PR re-enables the flaky mixed global/operation rate-limit scenarios for issue #2317 using apiname-keyed advanced-ratelimit plus operation policy.
Out of Scope Changes check	✅ Passed	The changes stay within the linked issue scope by adding mixed global/operation rate-limit test scenarios and no unrelated code.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}