Jmix behind OpenShift Route → Nginx → Service: WebSocket closes abnormally and UI reloads

Support
#1

Hi,

I’m deploying a Jmix (Vaadin) application on OpenShift and I’m running into an issue with Vaadin Push when I put an Nginx layer in front of the Jmix service.

My topology:

  • Working path (no problem):

Route → Service → Jmix pods

Vaadin UI works fine, no reload loops, WebSocket stable.

  • Failing path:

Route (nginx-route) → Nginx (DMZ) → Service → Jmix pods

When I scale Jmix pods to more than 1, the Vaadin UI starts to reload and I see WebSocket errors. Nginx ConfigMap for the frontend (simplified):

upstream jmix_backend_officer {
    server dlcn-officer.backend-d01-ktdlcn.svc.cluster.local:8080;
}

server {
    listen 8080;

    location / {
        proxy_pass http://jmix_backend_officer;

        proxy_http_version 1.1;

        proxy_set_header Host              $host;
        proxy_set_header X-Real-IP         $remote_addr;
        proxy_set_header X-Forwarded-For   $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto https;

        proxy_set_header Upgrade           $http_upgrade;
        proxy_set_header Connection        "upgrade";

        proxy_read_timeout 3600;
        proxy_send_timeout 3600;
        proxy_buffering off;
    }

    location ^~ /VAADIN/push {
        proxy_pass http://jmix_backend_officer;

        proxy_http_version 1.1;
        proxy_set_header Upgrade           $http_upgrade;
        proxy_set_header Connection        "upgrade";

        proxy_set_header Host              $host;
        proxy_set_header X-Real-IP         $remote_addr;
        proxy_set_header X-Forwarded-For   $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto https;

        proxy_read_timeout 3600;
        proxy_send_timeout 3600;
        proxy_buffering off;
    }
}

Browser console (Vaadin push) shows:

Invoking executeWebSocket, using URL:
wss://nginx-route-kl3init-dev.apps.rm2.thpm.p1.openshiftapps.com/VAADIN/push?v-r=push&...

Atmosphere: websocket.onerror
Atmosphere: websocket.onclose
Websocket closed, reason: Connection was closed abnormally (that is, with no close frame being sent). - wasClean: false
Atmosphere: Firing onReconnect
...
Navigated to https://nginx-route-kl3init-dev.apps.rm2.thpm.p1.openshiftapps.com/login?error

Notes:

  • Jmix is configured with embedded Hazelcast for clustering.

  • When I bypass Nginx and point the Route directly to the Jmix Service, Vaadin Push works fine with multiple pods.

  • Only when I insert Nginx between the Route and the Service, the WebSocket to /VAADIN/push keeps closing abnormally and the UI reloads / redirects to /login?error.

Questions:

  1. Is there a recommended Nginx configuration for Jmix/Vaadin Push when running behind an OpenShift Route (TLS termination at Route, then HTTP from Route to Nginx, then HTTP to Jmix Service)?

  2. Are there any known issues with Vaadin Push when there are two proxy layers (Route and Nginx)? Do I need to adjust X-Forwarded-* or Host headers in a specific way?

  3. Do I need a special location for /VAADIN/push or additional settings (timeouts, proxy_read_timeout, proxy_send_timeout, proxy_buffering, etc.) to keep the WebSocket stable?

  4. Any examples of a working Nginx config in front of a Jmix/Vaadin app on OpenShift (Route → Nginx → Service) would be very helpful.

Thanks in advance for any guidance.