CODEWITHSUNDEEP
azureazure-machine-learning-service
kotogo
kotogo

Reputation: 11

Azure Machine Learning Experiment can't mount blob storage to compute cluster

I'm using Azure ML workspace. I'm having an issue with running ML experiment for segmentation. When I submit a run it will always end with this error:

UserError: AzureMLCompute job failed. BFSMountError: Unable to mount blob fuse file system Info: Could not mount Azure Blob Container azureml-blobstore-{UID} at workspaceblobstore: Unauthorized. Cannot access the storage account with the given account key. Please verify that the account key is valid. Info: Failed to prepare an environment for the job execution: Job environment preparation failed on 10.0.0.5 with err exit status 1.

I believe that documentation says that this blob is created automatically for storage of model files.

Storage account in which this blob resides is integrated into VNET. Datastore has Use workspace managed identity for data preview and profiling in Azure Machine Learning studio set to yes, access key has been copied from storage account portal page, number of times. I have no idea why is this happening.

Upvotes: 1

Views: 2364

Answers (1)

Qianqian Nie
Qianqian Nie

Reputation: 56

Thanks for reaching out to us.

Please check out https://learn.microsoft.com/en-us/azure/machine-learning/how-to-secure-workspace-vnet for Azure Machine Learning resources vNet settings.

For this specific error, can you please help to confirm the following settings in your workspace:

  1. whether storage account(workspaceblobstore) is within the same vNet with the compute?
  2. whether allow access to microsoft service is checked in the storage account(workspaceblobstore)

enter image description here

Upvotes: 1

Related Questions