Spark created multiple base directories and delta directories in hdfs. If you copy the hdfs directory for table. Then have to pick the latest base directory (base directories are numbers) and then all delta directories.