atomic-save-daemon-convergence

Metadata

Description

Implement the `atomic-save-daemon-convergence` work item from `docs/plans/atomic-graph-work-save-dag.json` and `docs/design-atomic-graph-work-save.md`.
Start only after the shared kernel/schema gate is durably complete. Own only the files/modules listed below so sibling adapters remain safely parallelizable; coordinate any unavoidable overlap before editing.

Owned files/modules:
- `src/worker_control.rs`
- `src/worker_cli.rs`
- `src/service/convergence.rs`
- `src/commands/service/ipc.rs`
- `src/commands/service/mod.rs`
- `src/commands/service/coordinator.rs`
- `src/commands/service/triage.rs`
- `src/commands/sweep.rs`

Non-goals:
- No semantic verdict invention by the daemon.
- No provider planner formalization.
- No broad TUI work.

## Validation
- `cargo test service::convergence`
- `cargo test commands::service::ipc`
- `cargo test commands::service::triage`
- `cargo test --test atomic_save_faults dead_worker`
- `cargo fmt --check`

Depends on

• done atomic-save-kernel-schema
• done .assign-atomic-save-daemon-convergence

Required by

• open atomic-save-synthesis-canary
• done fix-nonsemantic-failed-dependency-convergence

Log

• 2026-08-03T14:55:05.303105986+00:00 Task paused
• 2026-08-03T14:55:21.159703811+00:00 Task published
• 2026-08-03T15:39:21.047353907+00:00 LIVE INCIDENT MUST BE FIXED IN THIS TASK: blocked downstream tasks accrued Dispatch falloff while atomic-save-kernel-schema was still in progress. When the kernel completed, seven genuinely ready tasks already had failures_without_progress=8 and future wake deadlines; dispatch exhaustiveness then incorrectly entered fail-closed PAUSED even though durable scheduled convergence deadlines existed. Replay capture also failed to reproduce its own reported no-forward violation. Preserve .wg/service/replay/violation-a2837d40651826c67c5cb18177ac8b996814b5185fe8cfbc17003c13a97be9d4.json as the incident fixture. Required: blocked work is breaker/falloff-neutral; a future durable wake counts as forward action; monitor and planner agree byte-for-byte; a ready task dispatches once its deadline arrives without operator resume loops.
• 2026-08-03T15:43:52.705429994+00:00 Spawned by coordinator --executor pi --model openai-codex:gpt-5.6-sol --isolation required-worktree
• 2026-08-03T15:44:31.769434387+00:00 Starting implementation; inspecting design, kernel, and daemon convergence adapters.
• 2026-08-03T16:18:16.066699735+00:00 Exact owner exited without a finish transaction; fenced once and scheduled same-session/worktree continuation (session=019fc84b-b45c-7060-9896-299ba8401cfe)
• 2026-08-03T16:18:17.940737542+00:00 Spawned by coordinator --executor pi --model openai-codex:gpt-5.6-sol --isolation required-worktree
• 2026-08-03T16:18:47.397331344+00:00 Starting implementation; resuming prior WIP and inspecting design/kernel APIs.
• 2026-08-03T16:21:44.332962958+00:00 FANOUT_DECISION: stay inline — broker, convergence scheduler, and dead-owner guards share one SaveTransaction journal and require sequential integration; prior WIP is on-track but incomplete.
• 2026-08-03T16:38:34.795979860+00:00 Implemented broker WAL/replay and dead-owner Prepared→Quiescing mechanics; exported replay_dead_owner_save_transactions for adversarial smoke/synthesis consumers. Cross-task message was correctly refused by worker capability scope.
• 2026-08-03T17:03:03.741725592+00:00 Validation progress: convergence, IPC, and triage scoped suites pass; fmt clean. The required atomic_save_faults target is not yet present in this worktree (owned by atomic-save-adversarial-smokes), so cargo reports no such test target; daemon-side dead-owner coverage passes in service::convergence and triage unit tests.
• 2026-08-03T17:06:53.401316503+00:00 Committed: ac1dcc6b — pushed to remote. Validated scoped convergence/IPC/triage suites, cargo build, cargo clippy, and fmt; atomic_save_faults target remains downstream-owned and absent.
• 2026-08-03T17:08:10.298491506+00:00 Bounded evaluator infrastructure state error[WG-EVAL-INSUFFICIENT-EVIDENCE]: bounded evidence insufficient [candidate-manifest:candidate-manifest:truncated,candidate-source:candidate-source:truncated,declared-artifact:declared-artifact-000:truncated,declared-artifact:declared-artifact-001:truncated,declared-artifact:declared-artifact-002:truncated,declared-artifact:declared-artifact-003:truncated,declared-artifact:declared-artifact-004:truncated,declared-artifact:declared-artifact-005:truncated,declared-artifact:declared-artifact-006:truncated,declared-artifact:declared-artifact-007:truncated] without semantic rejection or cross-executor fallback; source remains in-progress
• 2026-08-03T17:08:40.647048637+00:00 Consumed deep-readonly FLIP report deep-report-2930059e3256899dd5a37351d9e50c16566b237d279c96863a55c9fa0473f7c9; observations=12 findings=4 route=b3:07cbfd6e71528de4a03ba3051159b1a811230a2a0a97c83e9fcb743edd0c2466 usage=56428in/1180out
• 2026-08-03T17:08:45.748437930+00:00 Bounded evaluator infrastructure state error[WG-EVAL-EVIDENCE-UNAVAILABLE]: automatic immutable evidence assembly failed without semantic rejection or cross-executor fallback; source remains in-progress
• 2026-08-03T17:09:20.714645406+00:00 Bounded evaluator infrastructure state error[WG-EVAL-EVIDENCE-UNAVAILABLE]: automatic immutable evidence assembly failed without semantic rejection or cross-executor fallback; source remains in-progress
• 2026-08-03T17:09:50.552461291+00:00 Post-finalization merge commit: e2f000a0 — pushed; contains my ac1dcc6b implementation plus the task-owned merge of concurrent upstream legacy-migration work.
• 2026-08-03T17:25:38.598861780+00:00 late-process-diagnostic:1:attempt-1-2:3:agent-exit-nonzero observed after durable landed receipt wgcid:v1:blake3:0bb791277fe0f6728a02589df8a6abfe14144e4d47137489312ccc949de65fc2; lifecycle authority suppressed: Provider failure detected after streaming: provider-unavailable