search

Azure / MachineLearningNotebooks

Python notebooks with ML and deep learning examples with Azure Machine Learning Python SDK | Microsoft
https://docs.microsoft.com/azure/machine-learning/service/
MIT License
4.07k stars 2.52k forks source link

Azure ML mounting Storage Account #1589

Open SamAchten opened 3 years ago

SamAchten commented 3 years ago

Hello,

We are trying to mount an Azure Storage account in Azure ML. This works perfectly fine, until we start a child run. In the logs of the child run, we can see the following: Set Dataset input__c79bd306's target path to /mnt/batch/tasks/shared/LS_root/jobs/ml-studio-01/azureml/train_classification_model_20210909_fr_1631216080_e3eca838/wd/input__c79bd306_f7faa3c3-938e-4cfc-950b-c91c9827dfa4

But when we try to access the mount, we get the following error: '/mnt/batch/tasks/shared/LS_root/jobs/ml-studio-01/azureml/train_classification_model_20210909_fr_1631216080_e3eca838/wd/input__c79bd306_f7faa3c3-938e-4cfc-950b-c91c9827dfa4': No such file or directory

The code to start the child run can be found below. Thank you for your help.

child_config = ScriptRunConfig(source_directory='.', script='src/main_child.py', arguments=arguments, environment=environment, docker_runtime_config=DockerConfiguration(use_docker=True), compute_target=compute_target) run.submit_child(child_config)

jarandaf commented 3 years ago

Can you provide the full code sample? In principle, you should not do any file system mounting: you can specify input datasets as job arguments and AzureML will do all the mounting for you so that you can access them within your jobs.