将数据从ADLS Gen 2加载到Azure Synapse

时间:2020-05-13 12:18:11

标签: sql-server azure azure-data-lake azure-synapse polybase

我正在尝试使用polybase外部表功能将Parquet文件从ADLS Gen2加载到Synapse。

下面是代码,但是在运行create external table命令时,查询永远不会完成。取消查询执行后,我看到此错误-

外部文件访问由于内部错误而失败:'访问HDFS时发生错误:调用HdfsBridge_IsDirExist时引发Java异常。 Java异常消息: HdfsBridge :: isDirExist-检查目录是否存在意外错误:UnknownHostException:''.azuredatalakestore.dfs.core.windows.net'

SQL查询

CREATE MASTER KEY ENCRYPTION BY PASSWORD = '<password>';
GO

DROP CREDENTIAL ADLSCredential
CREATE DATABASE SCOPED CREDENTIAL ADLSCredential
WITH
    IDENTITY = 'user',
    SECRET = '<secret-key>'
;

CREATE EXTERNAL DATA SOURCE AzureDataLakeStorage
WITH (
    TYPE = HADOOP,
    LOCATION = 'abfss://<container>@<storage-account>.azuredatalakestore.dfs.core.windows.net',
    CREDENTIAL = ADLSCredential
);

-- Create an external file format for PARQUET files.  
CREATE EXTERNAL FILE FORMAT parquet  
WITH (  
    FORMAT_TYPE = PARQUET,  
    DATA_COMPRESSION = 'org.apache.hadoop.io.compress.SnappyCodec'  
); 

CREATE EXTERNAL FILE FORMAT uncompressedcsv
WITH (
    FORMAT_TYPE = DELIMITEDTEXT,
    FORMAT_OPTIONS (
        FIELD_TERMINATOR = ',',
        STRING_DELIMITER = '',
        DATE_FORMAT = '',
        USE_TYPE_DEFAULT = False
    )
);

CREATE EXTERNAL TABLE [dbo].[CashReceipts_external] (
    [AMOUNT_APPLIED] [float] NOT NULL,
    [TRX_NUMBER] [nvarchar](50) NULL,
    [SHORT_NAME] [nvarchar](50) NOT NULL,
    [NAME] [nvarchar](1) NULL,
    [CURRENT_RECORD_FLAG] [nvarchar](50) NULL,
    [CURRENCY_CODE] [nvarchar](50) NULL,
    [FUNC_CURRENCY_CODE] [nvarchar](50) NOT NULL,
    [CASH_RCPT_AMOUNT] [float] NULL,
    [CASH_HISTORY_AMOUNT] [float] NULL,
    [FUNC_AMT_HISTORY] [float] NULL,
    [STATUS] [nvarchar](50) NULL,
    [ANTICIPATED_CLEARING_DATE] [nvarchar](50) NULL,
    [CASH_HISTORY_EXCHANGE_RATE] [nvarchar](50) NULL,
    [GL_DATE] [datetime2](7) NULL,
    [GL_PERIOD] [datetime2](7) NOT NULL,
    [BATCH_GL_DATE] [nvarchar](1) NULL,
    [EXCHANGE_RATE] [nvarchar](50) NULL,
    [RECEIPT_NUMBER] [nvarchar](50) NULL,
    [DEPOSIT_DATE] [datetime2](7) NULL,
    [RECEIPT_DATE] [datetime2](7) NULL,
    [ISSUE_DATE] [nvarchar](1) NULL,
    [TYPE] [nvarchar](50) NULL,
    [GL_POSTED_DATE] [datetime2](7) NULL,
    [AMOUNT] [float] NULL
)
WITH
(
    LOCATION='parquetfiles'
,   DATA_SOURCE = AzureDataLakeStorage
,   FILE_FORMAT = parquet
,   REJECT_TYPE = VALUE
,   REJECT_VALUE = 0
)
;

2 个答案:

答案 0 :(得分:0)

根据错误消息,错误是由“ parquetfiles”位置引起的。

请尝试以下CREATE EXTERNAL DATA SOURCE命令:

CREATE EXTERNAL DATA SOURCE AzureDataLakeStorage
WITH
  ( LOCATION = 'wasbs://<container>@<storage_account>.blob.core.windows.net' ,
    CREDENTIAL = AzureStorageCredential ,
    TYPE = BLOB_STORAGE
  ) ;

CREATE EXTERNAL TABLE时,使用文件名或文件夹名:

CREATE EXTERNAL TABLE [dbo].[CashReceipts_external] (
    [AMOUNT_APPLIED] [float] NOT NULL,
    [TRX_NUMBER] [nvarchar](50) NULL,
    [SHORT_NAME] [nvarchar](50) NOT NULL,
    [NAME] [nvarchar](1) NULL,
    [CURRENT_RECORD_FLAG] [nvarchar](50) NULL,
    [CURRENCY_CODE] [nvarchar](50) NULL,
    [FUNC_CURRENCY_CODE] [nvarchar](50) NOT NULL,
    [CASH_RCPT_AMOUNT] [float] NULL,
    [CASH_HISTORY_AMOUNT] [float] NULL,
    [FUNC_AMT_HISTORY] [float] NULL,
    [STATUS] [nvarchar](50) NULL,
    [ANTICIPATED_CLEARING_DATE] [nvarchar](50) NULL,
    [CASH_HISTORY_EXCHANGE_RATE] [nvarchar](50) NULL,
    [GL_DATE] [datetime2](7) NULL,
    [GL_PERIOD] [datetime2](7) NOT NULL,
    [BATCH_GL_DATE] [nvarchar](1) NULL,
    [EXCHANGE_RATE] [nvarchar](50) NULL,
    [RECEIPT_NUMBER] [nvarchar](50) NULL,
    [DEPOSIT_DATE] [datetime2](7) NULL,
    [RECEIPT_DATE] [datetime2](7) NULL,
    [ISSUE_DATE] [nvarchar](1) NULL,
    [TYPE] [nvarchar](50) NULL,
    [GL_POSTED_DATE] [datetime2](7) NULL,
    [AMOUNT] [float] NULL
)
WITH
(
    LOCATION='[filename]'
,   DATA_SOURCE = AzureDataLakeStorage
,   FILE_FORMAT = parquet
,   REJECT_TYPE = VALUE
,   REJECT_VALUE = 0
)
;

参考:

  1. Create external data source to reference Azure blob storage
  2. Create external table: Arguments

LOCATION ='folder_or_filepath'指定Hadoop或Azure blob存储中实际数据的文件夹或文件路径以及文件名。

如果您将LOCATION指定为文件夹,则从外部表中选择的PolyBase查询将从该文件夹及其所有子文件夹中检索文件。就像Hadoop,PolyBase不会返回隐藏文件夹。它也不会返回文件名以下划线(_)或句点(。)开头的文件。

在此示例中,如果LOCATION ='/ webdata /',则PolyBase查询将返回mydata.txt和mydata2.txt中的行。它不会返回mydata3.txt,因为它是隐藏文件夹中的文件。而且它不会返回_hidden.txt,因为它是一个隐藏文件。

请注意:一个表一个文件,我们可以加载多个文件来创建外部表!

希望这会有所帮助。

答案 1 :(得分:0)

“创建外部数据源”命令具有属性位置

LOCATION = 'abfss://<container>@<storage-account>.azuredatalakestore.dfs.core.windows.net'

应该是

LOCATION = 'abfss://<container>@<storage-account>.dfs.core.windows.net'

我把它与ADLS Gen 1位置属性混在一起。不好意思,谢谢大家花时间研究这个问题,将其标记为已关闭。我最终使用了AAD应用程序注册令牌,而不是存储密钥。