hdfs, yarn, impala, and kudu are currently in service. If an io error occurs on one disk in one server during operation, Is the service down? Should I just leave it until I replace the disk? ... View more