Harden offload safety and coverage-complete controls#3
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c5a1218754
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if [[ "$EXPECTED_UNIQUE_INPUT_FILES" -gt 0 ]]; then | ||
| unique_inputs="$(manifest_unique_input_count)" | ||
| if [[ "$unique_inputs" =~ ^[0-9]+$ ]] && [[ "$unique_inputs" -ge "$EXPECTED_UNIQUE_INPUT_FILES" ]]; then | ||
| last_progress_epoch="$now_epoch" | ||
| log "worker_hold reason=coverage_complete unique_inputs=$unique_inputs expected=$EXPECTED_UNIQUE_INPUT_FILES" |
There was a problem hiding this comment.
This new stall branch suppresses restarts solely based on manifest coverage, independent of how the worker was launched. If --worker-args disables coverage-complete mode (or uses a different threshold) but watchdog --expected-unique-input-files remains at the default 510, any real stall after that coverage point will be held forever because last_progress_epoch is reset and the restart path is skipped. This can leave a genuinely stuck stage loop unmanaged in production watchdog deployments that customize worker args.
Useful? React with 👍 / 👎.
Bumps [actions/setup-python](https://github.com/actions/setup-python) from 5 to 6. - [Release notes](https://github.com/actions/setup-python/releases) - [Commits](actions/setup-python@v5...v6) --- updated-dependencies: - dependency-name: actions/setup-python dependency-version: '6' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 6. - [Release notes](https://github.com/actions/checkout/releases) - [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md) - [Commits](actions/checkout@v4...v6) --- updated-dependencies: - dependency-name: actions/checkout dependency-version: '6' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
1 participant
Summary\n- add coverage-complete mode to stage/shard loop and coverage-aware watchdog hold\n- harden shard offload gating (trained-registry fallback list, missing-registry skip mode, min coverage gate, min active token floor)\n- include offloaded manifests in coverage reporting metrics\n- update systemd/user unit defaults and docs\n- keep CI/tests aligned with new options\n\n## Validation\n- make lint\n- make test\n- bash -n scripts/.sh\n- python3 -m py_compile scripts/.py\n- manual: offload timer/service run now exits cleanly with coverage gate\n