Troubleshooting in Amazon SageMaker Unified Studio - Amazon SageMaker Unified Studio
EBS Volume Depletion with Local Notebook Execution Domain SAML Identity Provider Email Issue Project Creation Failure Data Explorer Visibility Issue Data Catalog Visibility Issue Connection to Amazon RDS MySQL in Existing VPC Visual ETL Flow Column Selection JupyterLab Configure Magic Error

Amazon SageMaker Unified Studio is in preview release and is subject to change.

Troubleshooting in Amazon SageMaker Unified Studio

EBS Volume Depletion with Local Notebook Execution

Question

“Enhance the StartExecution API response when throttling occurs due to low disk space, instructing the user to delete files from the jobs folder."

Answer

  1. Navigate to JupyterLab

  2. In the Jobs folder, select the folder and files

  3. Select delete

Domain

Question

From IAM SSO access portal URL, SAmazon SageMaker Unified Studio is not listed. When I click on Amazon DataZone, Amazon DataZone portal is shown. I clicked on SIGN IN WITH SSO, it failed due to Invalid redirectUri provided.

Answer

Visit DAmazon DataZone console, choose your domain, then click the Amazon SageMaker Unified Studio URL.

SAML Identity Provider Email Issue

Question

When using 3rd party SAML identity providers, the domain creation flow does not identify my email address.

Answer

This happens because during the user provisioning step, the email field was not populated in your local SSO instance. When sync-ing with 3rd party SAML identity providers, modify the default mapping to ensure it includes the "email" field and re-do the sync.

Project Creation Failure

Question

When I configure a project profile with resources pointing to another region/account, and try to create a new project using the project profile, it failed due to the error Project creation failed because one or more resources could not be provisioned.

Answer

Make sure that you complete following configurations:

  1. Domain owner account: In the Domains menu, choose your domain. Under the Account associations tab, verify that domain is associated with the target account, and the status is Associated.

  2. Target account: In the Associated domains menu, choose the associated domain. Choose your blueprint. Under the Regions tab, verify that the target region is added. Under the Authorization tab, verify that the target domain unit is shown.

  3. Domain owner account: In the Domain details, under the Project profiles tab, choose your project profile. Under the Blueprint deployment settings tab, choose Name of your blueprint, under Deployment order, verify that Account ID and Region are configured correctly. Under the Authorized users and groups, verify that your SSO user is added.

Data Explorer Visibility Issue

Question

On the data explorer, I cannot see my existing databases and tables on Glue Data Catalog. How can I query them?

Answer

Amazon SageMaker Unified Studio configures AWS IAM permissions and permission boundaries. You can optionally remove the permission boundaries to allow access to the existing databases and tables.

Data Catalog Visibility Issue

Question

On the data catalog, I cannot see my existing databases and tables on Glue Data Catalog. How can I view them?

Answer

  1. On your project page, choose Data sources.

  2. Choose CREATE DATA SOURCE to add the existing databases and tables as a data source.

  3. For Data source type, choose AWS Glue, and choose NEXT.

  4. Configure how to select your databases and tables here.

  5. Once everything is filled, choose NEXT and go ahead to register the data sources.

Connection to Amazon RDS MySQL in Existing VPC

Question

I want to connect to my Amazon RDS MySQL database instance that exists in my existing VPC. When I add a connection, I do not see any settings about VPC. How can I configure the reachability?

Answer

Amazon SageMaker Unified Studio uses the VPC and subnets that ar are specified in the domain creation. If you have the data source in a separate VPC, you can configure network reachability betweeneen the domain VPC and the data source VPC using VPC peering or Transit Gateway, or alternatively you can create a new domain using the data source VPC.

Visual ETL Flow Column Selection

Question

I created a data source, and now I am adding a new transform on top of it. But I cannot choose the columns for the transform.

Answer

When you start authoring a visual ETL flow, the data preview is also started. Once the preview is completed, then schema is automatically collected and available for further transforms.

JupyterLab Configure Magic Error

Question

When I ran %%configure magic, it returned the error Connection name cannot be empty.

Answer

The magic syntax is different from Glue Interactive Session's existing kernel. Instead, run the magic with following syntax: 


%%configure --name (compute) (-f)
{
"key": "value"
}

For example, if you want to change the default Spark SQL catalog name for project default Spark connection, run following magic: 


%%configure --name project.spark --f
{
"--conf": "spark.sql.defaultCatalog=glue_catalog"
}