Hadoop, spark when using the start - all the orders, if one node failure need to wait for a long time (300 s) can I continue to start, how to set up can let a timeout waiting time shorter?