Accessing a Remote Jupyter Notebook Server

Suhas Somnath
Advanced Data and Workflows Group
National Center for Computational Sciences
Oak Ridge National Laboratory

10/9/2017

The guides below walk you through setting up the port-tunneling to access a Jupyter Notebook server running on a remote machine at port 8889. For the sake of simplicity, we will assume that the Jupyter server is running on a virtual machine on CADES Cloud.

The same instructions can be used for connecting to other remote severs such as a TensorBoard server.

Windows:

Win Setup

This guide assumes that you have already followed CADES' instructions to set up basic SSH access using PuTTY and have saved the configuration (such as JupyterVM, in this example).

1. Open PuTTY from the Start menu.

2. Under Load, save, or delete a stored session, select your saved configuration and then click Load. You should see the value inside the Host Name (or IP address) box at the top change and reflect the address for your remote machine.

   

3. On the left-hand pane (Category), Go to SSH → Tunnels

4. Set the following:

   1. `Source port`: `8889`
   2. `Destination`: `127.0.0.1:8889`
   

   

   3. Click on the `Add` button. You should see the following under `Forwarded ports`:
   

   

5. Go back to Session on the left-hand pane. Highlight the name of the configuration (JupyterVM in this case) under Saved Sessions. and click on the Save button.

   

6. You can verify that the tunneling settings have been saved by:

   • Closing PuTTY
   • Reopening PuTTY
   • Loading the configuration JupyterVM
   • Navigating to SSH → Tunnels
   • The line containing the port forwarding should still be there.
   • If it is not, re-do the appropriate steps above.

Win Access

Once you start the Jupyter server correctly on your remote machine: 1. Open PuTTy 2. Select the correct configuration (JupyterVM in this case) 3. Click Load. You should see the IP address update in the box at the top. 4. Click on the Open button in the bottom. You will be presented with a new SSH connection to the VM. You can close this if you do not need it. 5. Open a browser and go to: http://localhost:8889/

MacOS

Mac Setup

1. Open the Terminal application and go to Shell → New Remote Connection
2. Click on the + icon under the right-hand column (Server).

3. Type: -N -L localhost:8889:localhost:8889 cades@172.22.3.50. Replace the IP address with that of your machine. Click OK.

   

Mac Access

Once you start the Jupyter server correctly on your remote machine:

1. Open the Terminal application and go to Shell → New Remote Connection
2. Click on the correct configuration on the right-hand pane
3. Click on the Connect button shown above to establish the connection with your Jupyter server.
   You will be presented with a new Terminal window once you click on Connect. Closing this terminal window will cause the connection to your Jupyter server to terminate. Therefore, close this terminal only when you want to close the port forwarding.
4. To reestablish the connection to your Jupyter server, just follow steps 1-3.