The worker watchdog treated wall-clock time spent in host sleep as evidence that the worker was unresponsive. Count consecutive observed heartbeat misses instead so sleep does not destroy live terminal sessions while genuinely stalled workers still restart after the same active observation window.