What uses port 8888
jupyter lab and jupyter notebook serve on http://localhost:8888 with a login token in the URL. A Jupyter server can run arbitrary code, so anyone who reaches it without a token or password controls the machine.
Should port 8888 be open to the internet?
Keep private. Never expose this port to the internet. Keep it on localhost, a private network or behind a VPN or SSH tunnel.
-
To use Jupyter on a remote server, keep it on localhost and tunnel:
ssh -L 8888:localhost:8888 user@server, then open the URL with the token locally.
Check what is listening on port 8888
sudo ss -ltnp 'sport = :8888'
sudo lsof -nP -iTCP:8888 -sTCP:LISTEN
Get-NetTCPConnection -LocalPort 8888 -State Listen
netstat -ano | findstr :8888
No output means nothing is listening. If a service should be, check that it is running and which address it binds to: 127.0.0.1 only accepts local connections, 0.0.0.0 or :: accepts them on every interface.
Fix “address already in use” on port 8888
Another process already holds the port. Find it with the commands above, then stop that service properly, or end the process:
sudo kill $(sudo lsof -t -iTCP:8888 -sTCP:LISTEN)
Stop-Process -Id (Get-NetTCPConnection -LocalPort 8888 -State Listen).OwningProcess
Allow port 8888 from trusted networks only
The examples allow 8888 only from 10.0.0.0/8; replace it with your own private network, office IP or VPN range.
sudo ufw allow from 10.0.0.0/8 to any port 8888 proto tcp
# Close it again:
sudo ufw delete allow from 10.0.0.0/8 to any port 8888 proto tcp
sudo firewall-cmd --permanent --add-rich-rule='rule family="ipv4" source address="10.0.0.0/8" port port="8888" protocol="tcp" accept'
sudo firewall-cmd --reload
New-NetFirewallRule -DisplayName "Jupyter 8888/TCP" -Direction Inbound -Protocol TCP -LocalPort 8888 -Action Allow -RemoteAddress 10.0.0.0/8
On AWS, Google Cloud and Azure, also allow the port in the security group or network firewall rules; host firewalls and cloud firewalls are separate.
Port 8888 with Docker
# Publish on localhost only. Docker's port rules bypass ufw and firewalld,
# so -p 8888:8888 would expose the port to the whole network.
docker run -d -p 127.0.0.1:8888:8888 quay.io/jupyter/base-notebook
The format is -p host_port:container_port. Containers on the same Docker network reach each other on
the container port directly, without publishing it. Convert a long docker run command into a Compose
file with docker run to compose.
Test port 8888 from another machine
nc -zv example.com 8888
Test-NetConnection example.com -Port 8888
Related ports
- Port 8000 (Development server (Django, Python)): a common port for development servers, including Django and Python’s http.server.
- Port 8080 (HTTP alternate): the most common alternate HTTP port, used by proxies, app servers and dev tools.
More development servers ports
3000 (Development server / Grafana) , 4200 (Angular dev server) , 5000 (Flask / development servers) , 5173 (Vite dev server) , 9000 (PHP-FPM / MinIO / SonarQube) , 9418 (Git protocol) . See all ports in the port lookup.
Sources
Spotted a mistake? Email hello@devtoolsmith.com and we’ll fix it.
Frequently asked questions
What is port 8888 used for?
Port 8888 is the default port of Jupyter Notebook and JupyterLab. jupyter lab and jupyter notebook serve on http://localhost:8888 with a login token in the URL. A Jupyter server can run arbitrary code, so anyone who reaches it without a token or password controls the machine.
Is it safe to open port 8888?
Never expose this port to the internet. Keep it on localhost, a private network or behind a VPN or SSH tunnel. Allow it only from the networks that need it, as in the firewall examples on this page.
How do I check if port 8888 is open?
On the server, list listening sockets with "sudo ss -ltnp 'sport = :8888'" on Linux or "sudo lsof -nP -i :8888" on macOS. From another machine, run "nc -zv host 8888" or, on Windows, "Test-NetConnection host -Port 8888". Connection refused means nothing is listening; a timeout means a firewall is dropping the traffic.