The number of started NN processes exceeds the limit.
Something wrong with queue_schedule.
Wait for 10s after the last training process exited and try again.