Reap stalled jobs from a Mac that died mid-render

FIXES → A job stuck in running/ after a crash, sleep, or unplugged Mac.

Farm operations: requeue stalled jobs, pause the queue and resume, plus the overnight autopilot policy — Captured from the real gateway on a demo farm

01 Reap from any Mac

Heartbeat-aware: a job is only requeued when its worker has genuinely gone quiet, so a slow render is never mistaken for a dead one. The app’s Ops view has the same button.

requeue jobs from a crashed Mac

./farm_status.sh --reap

02 Know the three local controls

Play lets a Mac claim work again. Pause drains safely — the current clip finishes, then no more claims. Stop ends the renderer within 30 seconds, removes partial output, and puts the interrupted job back in the queue. Overnight, autopilot runs the reap for you and logs every decision.

Done when Nothing sits in running/ without a live heartbeat, and the requeued job shows back on the board as queued.