Skip to content

Harden rollout serving and catch-up reliability#69

Open
jvmncs wants to merge 3 commits into
mainfrom
rollout-reliability
Open

Harden rollout serving and catch-up reliability#69
jvmncs wants to merge 3 commits into
mainfrom
rollout-reliability

Conversation

@jvmncs

@jvmncs jvmncs commented Jul 24, 2026

Copy link
Copy Markdown
Contributor

Summary

  • let cookbook trainers bake source trees before later image build steps
  • treat engine transport failures as retryable service failures while preserving abort and admission handling
  • coalesce an advancing checkpoint head before issuing a single GPU reload

Testing

  • UV_CACHE_DIR=/tmp/uv-cache uv run pytest -q (60 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.

1 participant