Skip to content

⚡ find tests that can be unforked#856

Merged
joerunde merged 3 commits intotorch-spyre:mainfrom
joerunde:fork-markers
Mar 30, 2026
Merged

⚡ find tests that can be unforked#856
joerunde merged 3 commits intotorch-spyre:mainfrom
joerunde:fork-markers

Conversation

@joerunde
Copy link
Copy Markdown
Collaborator

Description

This is a little sidequest inspired by @toddllm to make our tests run faster in CI. This PR marks tests by the type of cache that they use, and whether or not forking is strictly required for them.

This allows us to run tests that use the same type of vllm model cache without --forked, which saves a huge amount of time skipping spyre warmup for tests that hit the model cache.

Related Issues

Test Plan

As-is this doesn't change any behavior as the marking is strictly additive and none of these marks are used by our CI jobs currently.

We'll need to sync this with changes to the CI jobs to ensure that removing --forked for these cases actually works. I've manually verified that I can run both:

pytest tests -m "uses_llm and spyre and not fork_required"
pytest tests -m "uses_server and spyre and not fork_required"

on this branch without failure.

Checklist

  • I have read the contributing guidelines
  • My code follows the project's code style (run bash format.sh)
  • I have added tests for my changes (if applicable)
  • I have updated the documentation (if applicable)
  • My commits include a Signed-off-by: line (DCO compliance)

Signed-off-by: Joe Runde <joe@joerun.de>
@github-actions
Copy link
Copy Markdown

👋 Hi! Thank you for contributing to vLLM support on Spyre.
Just a reminder: Make sure that your code passes all the linting checks, otherwise your PR won't be able to be merged. To do so, run ./format.sh.
Now you are good to go 🚀.

We also recommend installing prek and configuring it to check your code before every local commit.

Comment thread tests/conftest.py Outdated

This is useful for running subsets of tests without requiring `--forked`.
Generally, we can't mix and match different vllm usage patterns within the same pytest process
without running into problems with relelasing spyre cards or crashing the compiler.
Copy link
Copy Markdown
Collaborator

@maxdebayser maxdebayser Mar 21, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

typo: relelasing -> releasing

Comment thread tests/conftest.py Outdated
joerunde and others added 2 commits March 30, 2026 13:12
Co-authored-by: Sophie du Couédic <sop@zurich.ibm.com>
Signed-off-by: Joe Runde <joe@joerun.de>
Signed-off-by: Joe Runde <joe@joerun.de>
@joerunde joerunde marked this pull request as ready for review March 30, 2026 19:21
Copy link
Copy Markdown
Collaborator

@maxdebayser maxdebayser left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@joerunde joerunde merged commit e91fa2b into torch-spyre:main Mar 30, 2026
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants