This counter counts lock requests that failed outright with an error rather than being queued to wait, typically because the requester asked not to wait, as with a conditional lock request. This differs from ordinary enqueue waits in that the requesting application chose to be told immediately rather than block, so a rising count here reflects applications designed to poll or retry rather than wait passively. For example, an application using conditional locking to check resource availability before proceeding would accumulate entries here each time the resource was already locked.