daily-fix: pod relaunch shape fully detaches stdio
kind: infra#wf-fix#wf-fix-fp:4391004ffbc7#daily-auto-filed
track:Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Auto-filed by /daily 2026-07-29 (problem sweep; emitting source: miner H-P2 (#1768)).
Goal
A pod-side relaunch must not keep the orchestrator's ssh wrapper alive as an implicit signal — pid/log breadcrumbs are the signal; attached stdio hangs the wrapper.
Workflow gap
- Bug observed: unverified hypothesis — verify at plan time: the wrapper hung on attached stdio semantics (mechanism inferred from the command text at transcript L2258 + ssh stdio behavior; not reproduced). Observed facts: 2.5h hung wrapper, killed; session death; ~2.7h marker gap reconstructed by a successor.
- Why it is a workflow gap: the pid-file launch contract covers the pid rewrite but the recipe does not mandate full stdio detach on the ssh-remote relaunch shape.
- Confidence (emitter): medium
- verified-at-filing: n/a — reason: the failing command shape is transcript evidence; the rule-file absence is the gap (planner greps pod-side-reporting.md for the stdio-detach mandate).
Proposed change (refine in planning)
Amend the detached relaunch recipe per the proposed shape; add the #1768 incident as the worked example.
Scope / surfaces
- Primary target:
.claude/rules/pod-side-reporting.md - Grep the workflow surface for the pattern before editing and update every hit.
Constraints / invariants
- Workflow-surface only;
scripts/workflow_lint.py --check-askspasses; ruff clean on touched files.
Provenance
- workflow_fix_target: .claude/rules/pod-side-reporting.md
- fingerprint: 4391004ffbc7
Activity