Azure ML Notebook Timeout.

Question

I was running a notebook which has a large task. My compute instance is set to shut down after 2 hours of idleness. My problem is, when the task is still running, the notebook said timeout. How can I fix that? I was trying to set a job to run the script, but i don't know what is my environment. My kernel is 'python 3( ipykernel) which i can't find it in the curated environment.

Question1: How can i prevent my notebook from timeout?

Question2: How can i find out what is my evnvironment?

Answer 1

Hello Emily Qian,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that your Azure ML Notebook always Timeout.

Regarding your questions:

Question 1: How can I prevent my notebook from timing out?

Azure ML considers a compute instance “idle” when there are no active Jupyter kernels (and a few other signals). If a cell is actually executing, it shouldn’t count as idle. So a “timeout” message in the browser is often a web/Notebook UI disconnection (e.g., websocket/proxy idle), while the kernel may still be working. - https://learn.microsoft.com/azure/machine-learning/how-to-manage-compute-instance

Question 2: How can I find out what is my environment?

“Python 3 (ipykernel)” ≠ curated environment. That kernel is just a Jupyter kernel running in whatever interpreter it points to. Curated environments are Azure ML environment assets (versioned Docker+Conda specs) you can select for jobs; they don’t automatically map to your local notebook kernel. - https://learn.microsoft.com/azure/machine-learning/how-to-use-environments

Jobs need an explicit environment, when you submit a job, you must reference an environment (curated or custom). If you want the job to use “what’s in my notebook now,” you need to capture that environment (requirements/conda YAML) and register or reference it. - https://learn.microsoft.com/en-us/azure/machine-learning/tutorial-azure-ml-in-a-day

Once you switch to jobs:

• First run of a custom environment may need to build a container, which takes time; curated envs are cached but may not match your current notebook packages exactly. - https://learn.microsoft.com/azure/machine-learning/how-to-use-environments
• For long runs, prefer a compute cluster over a personal compute instance to avoid notebook/interactive lifecycle surprises. - https://learn.microsoft.com/azure/machine-learning/how-to-create-attach-compute-cluster
• For data paths & authentication for jobs run in containers; make sure data is accessible via Azure ML Data assets/URIs (mount or download). - https://learn.microsoft.com/azure/machine-learning/how-to-read-write-data-v2

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.