How to reduce launch time for SageMaker Notebooks?
I am using SageMaker Notebooks and have added a lifecycle configuration to shut it down after inactivity. However, after a period of inactivity, it takes a long time (20 minutes) to boot back up. What can I do to reduce the launch time?
I have considered creating a custom container image with pre-installed packages and dependencies but I am not sure if this would help. I also wonder if using a larger instance size would reduce launch time.
What are the strategies and best practices to reduce launch time for SageMaker Notebooks?
- Newest
- Most votes
- Most comments
Typically lifecycle configuration scripts will timeout > 5 minutes. So if you're not seeing this timeout it could be something else that is causing the notebook to take 20 minutes to start. Check out the logs to see if you can pinpoint the issue as mentioned by another user: https://docs.aws.amazon.com/sagemaker/latest/dg/jupyter-logs.html
Instead of building a custom container image this is our published best practice to improve lifecycle configuration script time: https://aws.amazon.com/premiumsupport/knowledge-center/sagemaker-lifecycle-script-timeout/
Relevant content
- Accepted Answerasked a year ago
- Accepted Answerasked a year ago
AWS OFFICIALUpdated a year ago- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated 2 years ago
Can you tell whether it's taking 20 mins to start an instance or to run your lifecycle configuration?
Please check Jupyter notebook logs in cloudwatch at the duration of booting of your notebook instance. It will give the right indicators, what's happening and cause of delay. https://docs.aws.amazon.com/sagemaker/latest/dg/jupyter-logs.html Check out Lifecycle configuration best practices => https://docs.aws.amazon.com/sagemaker/latest/dg/nbi-lifecycle-config-install.html