fix: restore dcgm_fakers after lifespan tests

[debermudez]

The lifespan tests drive async with lifespan(fastapi_app) directly, which unconditionally appends two DCGMFaker entries to the module-level aiperf_mock_server.app.dcgm_fakers list. Those appends leak across files: tests/unit/server/test_app.py::test_dcgm_metrics_invalid_instance asks for /dcgm3/metrics expecting 404, but with 4 fakers in the list it gets 200.

Add an autouse fixture that snapshots dcgm_fakers on entry and restores it in finally, isolating the leak regardless of whether the lifespan body errors out before yield.

Summary by CodeRabbit

• Tests
  • Improved test reliability by fixing test isolation issues to prevent interference between mock server tests.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[github-actions]

Try out this PR

Quick install:

pip install --upgrade --force-reinstall git+https://github.com/ai-dynamo/aiperf.git@191aef99ba5531ae38d92df6a42a3ebf6e89dead

Recommended with virtual environment (using uv):

uv venv --python 3.12 && source .venv/bin/activate
uv pip install --upgrade --force-reinstall git+https://github.com/ai-dynamo/aiperf.git@191aef99ba5531ae38d92df6a42a3ebf6e89dead

---

Last updated for commit: 191aef9 • Browse code

[coderabbitai]

Walkthrough

The PR adds a pytest fixture that manages test isolation for lifespan tests by capturing the module-level dcgm_fakers list state before each test and restoring it afterward, preventing cross-test state leakage from lifespan's unconditional list appends.

Changes

Lifespan test state isolation

Layer / File(s)	Summary
State isolation fixture for lifespan tests tests/unit/aiperf_mock_server/test_app_lifespan.py	An autouse fixture snapshots aiperf_mock_server.app.dcgm_fakers before each test and restores the original list in a finally block to isolate tests from lifespan's unconditional appends.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Poem

🐰 A fixture most fine, with snapshot care,
Restores the list to pristine air,
No leaks betwixt tests shall remain,
State isolation breaks the chain! ✨

🚥 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 PR title 'fix: restore dcgm_fakers after lifespan tests' is clear, concise, and directly describes the main change—fixing test isolation by restoring module state after lifespan tests.
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.}

---

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.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tests/unit/aiperf_mock_server/test_app_lifespan.py`:
- Around line 15-16: The fixture function _restore_dcgm_fakers is missing a
return type annotation; update its signature to annotate the return type as
Generator[None, None, None] (and add "from typing import Generator" if not
already imported) since this pytest fixture yields control, ensuring all
function return types follow the project's type-hinting guideline.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: e37e9142-ed8e-4189-ba40-3a1293fb55ba

📥 Commits

Reviewing files that changed from the base of the PR and between 03167d0 and 191aef9.

📒 Files selected for processing (1)

• tests/unit/aiperf_mock_server/test_app_lifespan.py

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!