ci: mitigate OOM kills on large-model export tests - #244
Merged
regisss merged 2 commits intoJul 22, 2026
Conversation
Large models (e.g. phi4 ~14.7B) exceed the 16GB RAM on standard ubuntu-22.04 runners and get OOM-killed during model load/quantization, surfacing as exit code 143 / 'The operation was canceled' rather than a test failure. Add an aggressive disk cleanup step and a 24GB swapfile so these jobs page to disk instead of being killed, plus a timeout-minutes so genuine hangs fail cleanly instead of dragging the run for hours. Applied to both the stable (test_models.yml) and nightly (test_models_nightly.yml) workflows.
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
The 'Free up disk space' step deleted $AGENT_TOOLSDIRECTORY (/opt/hostedtoolcache), which is where actions/setup-python installs Python -- removing it broke pip/pytest and caused 'pytest: command not found' on every job. Root already has ~87GB free and the swapfile lives on the separate /mnt disk, so the cleanup was unnecessary. Remove it and keep only the swapfile + timeout.
digantdesai
approved these changes
Jul 21, 2026
Collaborator
Author
|
I think efficientnet failure is due to API request overload, potentially unrelated or maybe uncovered by more of these tests passing and so we have more rpm Phi4 should be resolved by #245 |
regisss
approved these changes
Jul 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Large models (e.g. phi4 ~14.7B) exceed the 16GB RAM on standard ubuntu-22.04 runners and get OOM-killed during model load/quantization, surfacing as exit code 143 / 'The operation was canceled' rather than a test failure. Add an aggressive disk cleanup step and a 24GB swapfile so these jobs page to disk instead of being killed, plus a timeout-minutes so genuine hangs fail cleanly instead of dragging the run for hours.
Applied to both the stable (test_models.yml) and nightly (test_models_nightly.yml) workflows.
A larger runner can avoid the need for this swap, but this change keeps it within the current footprint and we can consider a larger runner in the future