本文主要是介绍Azure databricks 还原备份的操作，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
起因：databricks 面临从HK迁移到国内的情况，目前只是迁移单个库就好，不需要全迁移，问了下azure的技术支持，把数据通过azure客户端azure copy 到指定源–》目标 目录就可
这一块参考：
https://docs.microsoft.com/zh-cn/azure/storage/common/storage-use-azcopy-blobs-copy
那拷贝过来后如何读取在storage目录中的数据呢？主要用的是python脚本来读取
#1、先要做好集群与storage的验证这一块没有搞好就需要重新来
# 目标表的databasesdatabaseName_target = "ods_xxx_xxx"# Create tables including both of partition and non-partitionsqlQueryCreateExternalTable = "CREATE EXTERNAL TABLE IF NOT EXISTS {0}.{1} USING {2} LOCATION '{3}'"
sqlQueryRecoverPartitions = "ALTER TABLE {0}.{1} RECOVER PARTITIONS"
path ="/demo/warehouse/库名/"for row in tableNames:try:flag = 0partitionName = ""tableBaseDirectory = spark.createDataFrame(dbutils.fs.ls((row["path"].replace('dbfs:', '')))).collect()for rowDirectory in tableBaseDirectory:
# print(rowDirectory)if rowDirectory["size"] != 0:flag = -1else:partitionName = rowDirectory["name"].split('=')[0]if flag == -1:spark.sql(sqlQueryCreateExternalTable.format(databaseName_target, row["name"], dataSource, row["path"]))print("INFO: {0} completed.".format(sqlQueryCreateExternalTable.format(databaseName_target, row["name"], dataSource, row["path"])))else:spark.sql(sqlQueryCreateExternalTable.format(databaseName_target, row["name"], dataSource, row["path"]))print("WARN: Table {0}.{1} has PARTITIONED BY {2}.".format(databaseName_target, row["name"], partitionName))spark.sql(sqlQueryRecoverPartitions.format(databaseName_target, row["name"]))print("INFO: Table {0}.{1} has been RECOVER PARTITIONED BY {2} completed.".format(databaseName_target, row["name"], partitionName))except Exception as e:print(e)print("ERROR: Create table {0} failed.".format(row["name"]))
这篇关于Azure databricks 还原备份的操作的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
