Failed to start kernel- Problem with Sagemaker Studio

Question

Hello,

I have been facing this weird error for 2 weeks without any reason. Appreciate any help. It is worth noting that the error is not related to the instance type or size.

> Failed to start kernel
Failed to check app status: Error when retrieving credentials from container-role: Error retrieving metadata: Received non 200 response (500) from ECS metadata: Credentials are not yet available (Context: RequestId: 3a2e0874-f355-4ab9-9b26-4139828b4dbf, TimeStamp: 1688562272.366086, Date: Wed Jul 5 13:04:32 2023)

Answer

The error is potentially related to the JupyterLab version that you are using in SageMaker Studio. We have noticed an increase in similar cases where customers using JupyterLab version 1 have experienced issues with kernel startup. AWS has recently made updates to SageMaker Studio, and it is possible that the version of JupyterLab you are using is now outdated. [1] I recommend that you update your JupyterLab to the latest version. [2]

Additionally, can you please shut down and update SageMaker Studio as described in [3]. I would also recommend you to shut down and update associated Studio Apps as described in [4].

Please only open a new Studio session and a Studio kernel after you complete the the steps (described in [3] and [4] respectively) to see the issue to resolution.

I would also like to communicate that restarting the studio periodically is good practice which actually prevents such intermittent issues.

References :

[1]https://docs.aws.amazon.com/sagemaker/latest/dg/studio-jl.html   
[2]https://docs.aws.amazon.com/sagemaker/latest/dg/studio-jl.html#studio-jl-view   
[3] https://docs.aws.amazon.com/sagemaker/latest/dg/studio-tasks-update-studio.html    
[4] https://docs.aws.amazon.com/sagemaker/latest/dg/studio-tasks-update-apps.html

Failed to start kernel- Problem with Sagemaker Studio

Relevant content