Using Jupyter Notebooks in YTsaurus

YTsaurus allows to launch Jupyter Notebooks on the computational resources of the cluster. Functionality can be used on clusters using CRI job environment.

Following actions are required to launch a notebook:

1. Set the environment variable JUPYT_CTL_ADDRESS to the address of JUPYT strawberry controller.

export JUPYT_CTL_ADDRESS=jupyt.test.yt.mycloud.net

1. Select docker-image of the jupyter-notebook. You can use jupyter stacks as an image or as a base of the image. Example of such an image is minimal-notebook.

2. Create a notebook. To create a notebook you should specify docker-image with the notebook and scheduler pool for the operation.

yt jupyt ctl create --speclet-options '{jupyter_docker_image="quay.io/jupyter/minimal-notebook"; pool=my-pool}' test-jupyt

1. Launch the notebook.

yt jupyt ctl start test-jupyt

1. Wait until notebook is started and get the link to it using CLI.

foo@bar:~$ yt jupyt ctl get-endpoint test-jupyt
{
    "address" = "http://some-node.yt.mycloud.net:27042";
    "operation_id" = "60b8ec86-6f123a7d-134403e8-ee6b88de";
    "job_id" = "16f76ac9-2a20ab96-13440384-14e";
}