Running jupyter notebook in background on remote server

remote-jupyter.md

Running Jupyter notebook in background mode on a remote server

⚙️ Setting up the remote connection

• Login into the server

ssh user@ip

• Run this command to start Jupyter notebook at port 8888 with no-browser mode

nohup jupyter-notebook --no-browser --port=8888 &

The logs of Jupyter Notebook will be stored in a file nohup.out, which is stored in the same folder in which you executed the above command.

• Open a new terminal in your local pc and connect to port 8888 on the remote server using this command

ssh -NL 8888:localhost:8888 remote_user@remote_ip

• Open the browser and type this lochalhost:8888. You can now access the Jupyter Notebook running on your remote server.

If you disconnect from the current Jupyter session and then reconnect, you won't be able to see the output of the current running cell.

🎁 Bonus advice

If you are training a deep learning model, use CSVLogger in callbacks, and use this command to see the output from the CSV file.

cat model.csv | column -t -s, | less -S

---

🔁 Reconnecting after disconnecting

• Open a new terminal in your local pc and connect to port 8888 on the remote server using this command

ssh -NL 8888:localhost:8888 remote_user@remote_ip

• Open the browser and type this lochalhost:8888. You can now access the Jupyter Notebook running on your remote server.

---

🛑 Stopping the running Jupyter notebook

• jupyter notebook list to get jupyter used port-number.
• lsof -n -i4TCP:[port-number] to get PID, The PID is the second field in the output.
• kill -9 [PID] to kill this process.