Repository navigation
Description
Task
T-041a· Milestone M3 — Oracles + runner + store · Size M · Priority P0 · Epic #3Implements FR-RUN-001 · FR-RUN-005 · FR-SEL-006 · FR-SEL-007
Branch
feat/t-041a-scheduler-campaignloop-step-propose-enqueue— one task = one branch = one PR (doc 32 §6)
Summary
Implement runner/scheduler.py (target_inflight = ceil(min(batch_size, workers) · inflight_factor)) and runner/loop.py CampaignLoop.step() per doc 05 §5.1: load history, strategy.propose(q=free), insert runs in one transaction with seed = derive_seed(campaign_seed, 'run', ordinal, repeat).
Context / motivation
The orchestrator side of the intelligent loop (doc 04 §6, doc 14 §2).
Scope
In scope
- Budget accounting (
runner.count_invalid_in_budget), repeats ≥ 1 with distinct derived seeds, pins applied. - History excludes INVALID for learning (FR-SEL-007) but counts them for stop rules.
- Proposals' rationale persisted into
runs.proposal/phase. strategy.proposedlog +roboprobe_selector_propose_secondsmetric; propose timeout → Sobol fallback +selector.fallback.
Out of scope
- Checkpointing/leader lock (T-041b)
- Reaper/stop rules (T-042a/b)
- Multi-campaign service (T-048)
Implementation guidance
- Ticks must be idempotent: re-running a tick after a crash never double-enqueues (unique
(campaign_id, ordinal, repeat_index), doc 34 §8). Use the in-memory fakes for unit tests and a RandomStrategy fake if T-051a isn't merged yet. - Standards: respect the layering contracts (doc 16 §2) and conventions in doc 34; no hard-coded tunables — thresholds/defaults come from typed settings (doc 27); no magic numbers (ruff
PLR2004); public APIs typed (mypy strict) and documented.
Configuration keys
From the configuration reference (doc 27). Keys marked (new) must be added to the typed settings models and appear in the generated reference (T-003c).
runner.tick_interval_srunner.inflight_factorrunner.max_batch_sizerunner.count_invalid_in_budgetselection.propose_timeout_s
Acceptance criteria
- Unit tests with fakes: correct number enqueued for various worker counts/budgets.
- Two identical campaigns (same seed) enqueue identical scenario sequences (FR-SEL-005).
- Repeated tick after simulated crash enqueues nothing new.
- Tests listed below added and green in CI; coverage of new code meets the gate in doc 18 §3.
- Docs, config reference and CHANGELOG
Unreleasedupdated where behaviour, keys or interfaces changed. - Definition of Done (doc 32 §3) satisfied.
Tests required
tests/unit/runner/test_scheduler.py,test_campaign_loop.py.
Dependencies
Blocked by: #61 (T-011a), #79 (T-041c), #84 (T-050a)
Blocks: #86 (T-041b), #87 (T-042b), #94 (T-046), #134 (T-074b)
Parallelization notes
Files / paths owned by this task — avoid editing other paths; shared files (pyproject.toml entry points, cli/main.py, api/routers/__init__.py, router.tsx, docker/compose.yaml, ci.yml) are append-only:
backend/src/roboprobe/runner/scheduler.pybackend/src/roboprobe/runner/loop.py
Parallel with worker tasks (different modules).
Definition of Done
Per doc 32 §3 and AGENTS.md §5: lint, format, mypy strict and import-linter green; tests added and passing; docs updated; PR template completed with Closes #<this issue>; Conventional Commit messages referencing T-041a (doc 16 §6); commits authored only by the repository owner identity (AGENTS.md §4); CI green; reviewer checklist (doc 32 §5.1) satisfied.
References
Design docs: doc 05 §5.1 · doc 14 §2 · doc 04 §6 · doc 21 §3 · doc 34 §8
Requirements: FR-RUN-001 · FR-RUN-005 · FR-SEL-006 · FR-SEL-007 (traceability: doc 28)
Metadata
Metadata
Assignees
Labels
Projects
- StatusShow more project fieldsBacklog
Activity
roboprobe.bootstrap) and plugin entry-point registries #94confirmruns and update mode #134