A session with 60 min on 1 core can be launched using: srun -ntasks 1 -t 60 jupyter-compute 15051 # please change port numberĪfter general output, JupyterLab prints a URL with a unique key and the network port number where the web-server is listening, this should look similar to. Therefore, a script is provided, taking care of port forwarding to the compute node and launching JupyterLab. On compute nodeĮspecially notebooks with computational and memory intensive tasks should run on compute nodes. Jupyter will present output as described in the next section including the URL and a unique key, which needs to be copied in your local browser. Where, -port 15051 specifies the above selected port number and -no-browser option prevents JupyterLab from trying to open a browser on the compute/login node side. On login nodes / virtual labsįor very small (computational cheap and small memory) the JupyterLab can be started on the login or virtual lab using: jupyter lab -port 15051 -no-browser Please launch compute or memory intensive tasks on a compute node. The JupyterLab server then can be started on the present node (login or virtual lab) or offloaded to a compute node. Or alternatively, and particularly if you are using a Māui ancillary node instead of Mahuika, you can use the Anaconda version of JupyterLab instead: module load Anaconda3 On the Mahuika login node, load the environment module which provides JupyterLab: module load JupyterLab opening a new terminal and start another ssh session: ssh mahuika
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |