Troubleshooting Jupyter Notebooks

The Jupyter Notebook doesn't launch

While the scheduler is attempting to launch your Jupyter Notebook job, you may see it count through a number of retries before you see the "Open Notebook" link for accessing your notebook.

If you do not see the "Open Notebook" link after 10 retries, you might see one of the following errors instead:

In these situations, try the following steps and if you still can't launch a notebook, send email to supercloud@mit.edu for assistance.

1. Check to see if the job is running: Enter the LLstat command in an ssh session to the SuperCloud and look for a job with the name JupyterNotebook
2. If the job is running (you will see "R" in the "ST" column), refresh the browser page and you should see the "Open Notebook" link.
3. If the job is pending (you will see "PD" in the "ST" column), there aren't resources to launch your job. This can occur if the system is heavily loaded, or if you have exhausted your allocated number of cores.
4. If you have not exhausted your core allocation, check to see if there are system resources available: Enter the LLfree command in the ssh session

Other things to try if you are unable to launch a Jupyter Notebook:

• Refresh the browser page and relaunch the notebook
• Close your browser, open a new browser, and relaunch the notebook
• Delete the file $HOME/.jupyter/llsc_notebook_bashrc and relaunch the notebook

Other Jupyter Notebook problems

If you run into problems after your notebook has launched, you should check the log file for error messages. You can find the log file in your SuperCloud home directory:

$HOME/slurm-xxxxx.out

where xxxxx is the job number of your Jupyter Notebook.

Send email to supercloud@mit.edu with the job number and any error messages that appear in the log file.