Reclaim tasks whose worker died mid-flight

Resets claimed tasks that have been held longer than timeout seconds back to pending (clearing owner + claimed_at) so another worker can pick them up. This is the crash-recovery half of the coordinator: a worker that dies after claiming a task would otherwise block its dependents forever. Called from the worker loop's idle branch, so recovery happens without a separate lead.

Usage

board_reclaim_stale(db_path, timeout = 300)

Arguments

db_path

Character. Board path.

timeout

Numeric. Seconds a claimed task may be held before it is considered stale (default 300).

Value

Integer. Number of tasks reclaimed.