Cron reliability and operations
The preview runs the scheduler as part of the application's process system; it is not a durable queue. It configures one worker. Increasing the worker count can register duplicate schedules in each worker. Separate installations do not coordinate with one another. Use external coordination/locking for single execution across a cluster if your application provides it; the component does not add this mechanism.
Do not block the event loop with long SQL/HTTP/CPU work in callbacks. Use an external queue or worker for long-running/retryable tasks and implement idempotency at task level. The source does not persist schedule/execution history as durable job state, so process restart does not restore execution history. Verify error/retry behavior against the exact Localzet Cron version and workload.

