Failed jobs

When a job exhausts its tries, the worker logs it through FailedJobProvider.

./grind queue:failed
./grind queue:retry <id>
./grind queue:retry --all
./grind queue:flush

queue:retry re-queues the payload (attempts reset) onto the original connection/queue, then forgets the failed row.

Durable failed storage

queue.failed.driver	Storage
memory (default)	Process heap
file	JSON at queue.failed.path (default storage/framework/failed_jobs.json)
database	failed_jobs table (queue.failed.table); needs Cadence Database

Use file or database whenever retries must survive a worker restart.