debugging NGINX configuration for Jupyter

docker-compose.yml

jupyter:
  image: jupyter/datascience-notebook
  environment:
    - PASSWORD=${PASSWORD}


nginx:  
  image: nginx
  links: 
    - jupyter
  volumes:
    - ./nginx.conf:/etc/nginx/nginx.conf
    - ./letsencrypt.crt:/data/cert.crt
    - /letsencrypt.key:/data/key.key
  ports:
    - 80:80
    - 443:443

nginx.conf

## Based on: https://github.com/calpolydatascience/jupyterhub-deploy-data301/blob/master/roles/nginx/templates/nginx.conf.j2
user  nginx;
worker_processes  1;

error_log  /var/log/nginx/error.log warn;
pid        /var/run/nginx.pid;


events {
    worker_connections  1024;
}


http {

  include /etc/nginx/mime.types;
  default_type application/octet-stream;

  map $http_upgrade $connection_upgrade {
        default upgrade;
        ''      close;
    }

  upstream jupyter {
      server jupyter:8888 fail_timeout=0;
  }


    server {
        listen 80;
        server_name xsede.carlboettiger.info;
        rewrite        ^ https://$host$request_uri? permanent;
    }

    server {
        listen 443;

        client_max_body_size 50M;
        server_name xsede.carlboettiger.info;

        ssl on;
        ssl_certificate /data/cert.crt;
        ssl_certificate_key /data/key.key;

        ssl_ciphers "AES128+EECDH:AES128+EDH";
        ssl_protocols TLSv1 TLSv1.1 TLSv1.2;
        ssl_prefer_server_ciphers on;
        ssl_session_cache shared:SSL:10m;
        add_header Strict-Transport-Security "max-age=63072000; includeSubDomains";
        add_header X-Content-Type-Options nosniff;
        ssl_stapling on; # Requires nginx >= 1.3.7
        ssl_stapling_verify on; # Requires nginx => 1.3.7
        resolver_timeout 5s;


        # Expose logs to "docker logs".
        # See https://github.com/nginxinc/docker-nginx/blob/master/Dockerfile#L12-L14
        access_log /var/log/nginx/access.log;
        error_log /var/log/nginx/error.log;

        location / {
            proxy_pass http://jupyter;

            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header Host $host;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        }
     location ~* /(api/kernels/[^/]+/(channels|iopub|shell|stdin)|terminals/websocket)/? {
            proxy_pass http://jupyter;

            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header Host $host;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            # WebSocket support
            proxy_http_version 1.1;
            proxy_set_header Upgrade $http_upgrade;
            proxy_set_header Connection $connection_upgrade;

        }
    }
}

Thanks this works!

what about just serving staring from / instead of ~?

where are the static files actually served from?

Thanks. It helped

Thank you, it works like a charm.

Thanks, It works for me

Thanks, works with nginx mainline and JH 0.8.1

Thanks for this - that worked for me, as well.

Hi, does anyone has been trying transfer the NGINX user uid header to jupyternotebook session? I mean something like to identify the authenticated user on the notebook session.

Thanks,

Dude you saved our ass here just now. Thanks :)

Thank you!. You've saved my time :)))))))))

Thank you!

Still works as of 2023, however note that on Safari, the kernel will seem like it's restarting continuously, while still generating the right cell output(s). If Firefox or Chrome is installed, better use that.

I've got many errors of the form below with the container, as well as with the official container. I turn of SSL:

WebSocket connection to 'ws://localhost/api/events/subscribe?token=9fe2a91c775e50ddc0535cf0cd55c58b38919586caf08470' failed: 

Does anybody have the same problem, and perhaps a solution?

@dmikushin have you found a solution?

@alesito85 , kind of. First of all, not all websocket errors are an indication of issue. For example, if connection to the Jupyter website has been established, but then lost due to the internet or network interruption, then these errors are presented naturally, as for any socket connection that becomes broken. If Docker container has been restarted, this is also a legitimate cause of websocket errors. So these errors may be attributed to Nginx configuration only given stable networking conditions. Be sure to refresh the page first, to clear out the old errors.

Instead of carrying out an own Nginx configuration, please first take a look at the official one, as it might have changed in the newer versions of Jupyter.