Azure 机器学习工作室:无法从 Azure SQL 数据库创建数据存储

Azure Machine Learning Studio: cannot create Datastore from Azure SQL Database

我正在尝试从 Azure 机器学习工作室连接到 Azure SQL 数据库。基于https://docs.microsoft.com/en-us/python/api/azureml-core/azureml.core.datastore.datastore?view=azure-ml-py,似乎推荐的模式是使用Datastore.register_azure_sql_database方法创建Datastore,如下所示:

import os
from azureml.core import Workspace, Datastore

ws = Workspace.from_config() # asks for interactive authentication the first time

sql_datastore_name  = "datastore_test_01" # any name should be fine
server_name         = os.getenv("SQL_SERVERNAME"    , "{SQL_SERVERNAME}") # Name of the Azure SQL server
database_name       = os.getenv("SQL_DATABASENAME"  , "{SQL_DATABASENAME}") # Name of the Azure SQL database
username            = os.getenv("SQL_USER_NAME"     , "{SQL_USER_NAME}") # The username of the database user.
password            = os.getenv("SQL_USER_PASSWORD" , "{SQL_USER_PASSWORD}") # The password of the database user.

sql_datastore = Datastore.register_azure_sql_database(workspace      = ws,
                                                      datastore_name = sql_datastore_name,
                                                      server_name    = server_name,
                                                      database_name  = database_name,
                                                      username       = username,
                                                      password       = password)

我很确定我已经正确设置了所有参数,从我的 SQL 数据库资源的 ADO.NET 连接字符串中复制了它们 --> 设置 --> 连接字符串:

Server=tcp:{SQL_SERVERNAME}.database.windows.net,1433;Initial Catalog={SQL_DATABASENAME};Persist Security Info=False;User ID={SQL_USER_NAME};Password={SQL_USER_PASSWORD};MultipleActiveResultSets=False;Encrypt=True;TrustServerCertificate=False;Connection Timeout=30;

但是,我收到以下错误:

Registering datastore failed with a 400 error code and error message 'Azure SQL Database Error -2146232060: Please check the correctness of the datastore information.'

我错过了什么吗?例如,防火墙规则?我还尝试将 Azure ML 计算资源的 public IP 地址添加到我的 SQL 数据库资源中允许的 IP 地址列表中,但仍然没有成功。


更新:将skip_validation = True 添加到Datastore.register_azure_sql_database 解决了问题。然后我可以用

查询数据
from azureml.core import Dataset
from azureml.data.datapath import DataPath

query   = DataPath(sql_datastore, 'SELECT * FROM my_table')
tabular = Dataset.Tabular.from_sql_query(query, query_timeout = 10)
df = tabular.to_pandas_dataframe()

是 vnet 背后的数据存储?你运行上面的注册码在哪里?在同一个 vnet 后面的计算实例上? 这是描述连接到 vnet 背后的数据所需执行的操作的文档: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-enable-virtual-network#use-datastores-and-datasets