在ML中使用导出数据模块时覆盖Blob文件 [英] Overwrite Blob File when using Export Data Module in ML

问题描述

我正在尝试使用具有存储帐户访问类型的ML Studio中的导出数据"模型导出数据.我提供了正确的存储密钥，blob路径(即containername/folder1/folder2/file1.csv)，并将覆盖文件设置为true.

I am trying to export data using "Export Data" Model in ML studio with storage account access type. I have provided correct storage key, blob path (i.e containername/folder1/folder2/file1.csv ) and have set overwrite file as true.

但是我得到以下错误.在这里，我的Blob存储为通用v1.当存储帐户为经典"帐户时，其工作原理相同.

However I get below error. Here my Blob storage is General Purpose v1. The same works when the storage account is – "classic".

这里的错误是关于URL的，但是，当认证类型为n时，没有要配置的URL. -帐户"，您知道如何解决吗?

Here Error talks about URL but, there is no URL to be configured when the authentication type is  - "Account" Any idea how to solve this?

错误–

requestId = cf9fef2526774b84b7fc051df5a3525c errorComponent = Module. taskStatusCode = 400. {"Exception":{"ErrorId":"CloudStorageWriteError"，"ErrorCode":"0151"，"ExceptionType":"ModuleException"，"Message":"Error" 0151:写入云存储时出错:远程服务器返回错误:(400) 错误的请求.请检查网址. } 错误:错误 0151:写入云存储时出错:远程服务器返回错误:(400)错误的请求.请检查URL.进程退出，错误代码为-2

requestId = cf9fef2526774b84b7fc051df5a3525c errorComponent=Module. taskStatusCode=400. {"Exception":{"ErrorId":"CloudStorageWriteError","ErrorCode":"0151","ExceptionType":"ModuleException","Message":"Error 0151: Error writing to cloud storage: The remote server returned an error: (400) Bad Request.. Please check the url."}}Error: Error 0151: Error writing to cloud storage: The remote server returned an error: (400) Bad Request.. Please check the url. Process exited with error code -2

因此，接下来我选择了SAS身份验证.有2个发现–

1. 当我使用Azure Storage Explorer在容器"级别创建具有权限的SAS令牌–读，写，删除，列表时，文件第一次从ML推送到Blob存储.当我再次运行导出模块"时，它说– File或Blob 已经存在.现在，在工作室中，没有选项可以使用SAS选项覆盖现有文件.我还设置了权限= rdwl(所有权限).但是文件仍然没有被覆盖.
2. 然后，我在存储帐户级别为Blob服务创建了SAS令牌.这次，导出模块可以成功但令人惊讶的是，文件未显示在目标Blob路径位置.

1. When I created the SAS token using Azure Storage Explorer at the "Container" level with permissions – read, write, delete,list , the file got pushed from ML to blob store for the first time. When I ran the "Export Module" again it says – File or Blob already exists. Now here, in the studio there is no option to overwrite existing file with SAS option. I had also had set permissions = rdwl (all permissions). But still file was not getting overwritten.
2. Then I created SAS token at the storage account level for Blob service. This time the export module can successfully but surprising the file is not showing up at the destination blob path location.

您能在这里进行指导吗?–我的目标是每天要在ML实验运行时将csv文件从ML推送到Azure Blob存储(v1).文件名将始终是相同的，因此文件将被覆盖.

Can you please guide here – My goal is that I want to push csv file from ML to Azure Blob storage (v1) everyday when the ML experiment runs. The filename will be always be same and hence file needs to be overwritten.

推荐答案

似乎使用具有部署模型资源管理器"的Blob存储；会导致这个问题，它的工作正常 经典"部署模型.这应该是一个很好的新功能.我会将其转发给产品团队.谢谢.

It seems that using a blob storage with deployment model "Resource Manager" will cause this problem, its working fine for "Classic" deployment model. This should be a good new feature. I will forward it to the product team. Thank you.

问候，

Yutong

这篇关于在ML中使用导出数据模块时覆盖Blob文件的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！