(Click to open topic with navigation)
Nitro has the ability to detect workers that have become unresponsive due to hardware, network, or software failure. During normal operation, workers periodically send an update to the coordinator. If the coordinator doesn't receive a status update after 45 seconds, the worker is deemed to be unresponsive, and any outstanding assignments will be revoked and reassigned to responsive workers. However, if the worker reports back to the coordinator before the assignment has been assigned to another worker, the assignment will be recovered and completed by that originally-assigned worker.