Skip to content

[BugFix] Stabilize cross-format checkpoint recovery - #4252

Merged
vmoens merged 1 commit into
mainfrom
codex/fix-checkpoint-cross-format-recovery
Sep 6, 2026
Merged

[BugFix] Stabilize cross-format checkpoint recovery#4252
vmoens merged 1 commit into
mainfrom
codex/fix-checkpoint-cross-format-recovery

Conversation

@vmoens

@vmoens vmoens commented Sep 6, 2026

Copy link
Copy Markdown
Collaborator

Fixes cross-format checkpoint recovery when two manifests have identical creation timestamps, as observed on Windows. Rotation now uses filesystem modification nanoseconds before filename ordering, so interrupted replacements keep the newly written checkpoint regardless of container format.

The existing interrupted-replacement test now forces equal creation times and distinct write order.

Validation:

  • test/test_checkpoint.py: 60 passed, 1 skipped
  • pinned ufmt check
  • pinned flake8 check

Observed in: https://github.com/pytorch/rl/actions/runs/34016957895/job/101452018472

@pytorch-bot

pytorch-bot Bot commented Sep 6, 2026

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/rl/4252

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 7f83d31 with merge base 5380357 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@meta-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Sep 6, 2026
@github-actions github-actions Bot added the BugFix label Sep 6, 2026
@vmoens
vmoens merged commit 098d464 into main Sep 6, 2026
159 of 162 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

BugFix CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant