frontend-host options

[Jackiexiao]

I need to debug via a remote machine, but how can I change the port for a frontend webpage? I am currently able to access the API documentation via http://ip:8000/docs, but I cannot access the website through http://ip:3000."

I couldn't find a frontend-host option when run reflex run

Usage: reflex run [OPTIONS]

  Run the app in the current directory.

Options:
  --env [dev|prod]                The environment to run the app in.
                                  [default: dev]
  --frontend-only                 Execute only frontend.
  --backend-only                  Execute only backend.
  --frontend-port TEXT            Specify a different frontend port.
                                  [default: 3000]
  --backend-port TEXT             Specify a different backend port.  [default:
                                  8000]
  --backend-host TEXT             Specify the backend host.  [default:
                                  0.0.0.0]
  --loglevel [debug|info|warning|error|critical]
                                  The log level to use.  [default: info]
  --help                          Show this message and exit.

[picklelo]

You can set the --frontend-port flag to change it (or set the FRONTEND_PORT environment variable).

[Jackiexiao]

I want to change frontend-host to 0.0.0.0 instead of localhost, not port

[Jackiexiao]

according to docs: https://reflex.dev/docs/api-reference/config/

change rxconfig.py

config = rx.Config(
    app_name="blank",
    frontend_port=8001,
    deploy_url="http://0.0.0.0:8001/",
)

But add a option frontend-host like backend-host to reflex run will be better @picklelo @masenf

[masenf]

I think i'm still a bit confused by this discussion... the deploy_url only affects the hosting service and the sitemap generation for export. From double checking the code, deploy_url seems to change nothing about how reflex run serves the site.

By default, the frontend should be available from other hosts without further configuration.

[masenf]

@jakeayres with render, only a single port can be exposed, so my example here uses a reverse proxy (caddy) to route the traffic to the frontend or backend by url while exposing a single port, 10000

[jakeayres]

I'm running my app across two render services (backend and frontend). The backend appears to be working perfectly well as I can interface with it from a locally running frontend (--frontend-only). Despite my best efforts, I can't convince render to see my remote frontend and expose it to the public at <myappname>.onrender.com. The logs indicate that the frontend is running at localhost:10000 and that render can't find any ports on 0.0.0.0 even with deploy_url="http://0.0.0.0:10000", in my config.

[masenf]

the deploy_url probably isn't doing what you want.

on the frontend try setting api_url="https://mybackend.onrender.com"

[jakeayres]

My full config is:

config = rx.Config(
    app_name="mybackend",
    api_url="https://mybackend.onrender.com",
    frontend_port=10000,
    deploy_url="http://0.0.0.0:10000",
)

My backend service is run remote on render using reflex run --backend-only --backend-port 10000

If I run a local front end using reflex run --frontend-only all is well. Indeed, if I suspend the backend service on render, I am notified that my local front end is no longer able to connect to the backend.

If i run a remote front end on render with start command reflex run --frontend-only --frontend-port 10000. The logs return:

App running at: http://localhost:10000/ (Frontend-only mode)
==> No open HTTP ports detected on 0.0.0.0, continuing to scan...

The service running the front end also exceeds its memory limit (2GB) if left alone for 10mins. It isn't clear to me whether this is related or not.

[jakeayres]

For what it is worth, I never managed to find a fix for running the app across two render services.

Dockerizing my app and using Caddy and running on one services (as you suggested) worked well.