4838d5a0-e5bd-482c-bbf8-35740edea841

is your frontend part of the shared repo? as in its importing stuff from the backend?

@Brody backend and frontend app do not depend on each other. I made sure of that.

isolated monorepo?

Backend imports some stuff from auth, db

Frontend imports some stuff from ui

gotcha

I assume you already have two services in your railway project, one for the backend and one for the frontend?

yes

perfect

90% there

damn it 😛

set the location to the nixpacks.toml file with the NIXPACKS_CONFIG_FILE service variable, and then change your start command to start Caddy, you will likely need to adjust the file path for the Caddyfile in both the start command and fmt command within the nixpacks.toml

Ok, before I do that I have a follow up question:
Is it the best and recommended way to run a vite react app with Caddy ? I saw in previous threads that some serve it with docker/node. Others were running with vite, but that was not recommended due to (very) high costs?

the command vite runs a development server and that's going to be unstable and yes cost far more than a production grade web server such as Caddy

a lot of people also use NGINX but I personally think Caddy is far more user friendly

Thanks

Trying your suggestion

let me know if you run into any problems

Ye

NIXPACKSCONFIGFILE

probably these

probably need to change the full path to the Caddyfile, given you have it in a subdirectory, you can edit that in the nixpacks.toml

and then you'll want to remove the start command in the UI that you set so that the start command in the nixpacks.toml file will be used

trying this now

mmm, it deployed successfully, but FE is not reachable/timing out.

Do i need to change something in the Caddyfile itself?

no you shouldn't, does it result in a 502?

Can't see if it's a 502

Wondering if this should be changed in the Caddyfile ?

Tried it, but didn't solve it …

nope, please do not change that

your target port is likely set incorrectly

well, don't change it unless you have some custom config that puts the built static files anywhere other than the default location

No, default target is the dist folder

Reverted the change 🙂

Probably this lol

yep!

Railway magic is magic 😄

all working now?

had a 404 when going to the URL, so did a re-deploy.

Still 404

you still have your build command set right?

build command yes, start command is removed.

Wondering if I need to set $PORT in variables ?

maybe try updating the root to use the full path? I can't remember needing to do that before though

which 'root' do you mean ?

the root directory in the Caddyfile

Will investigate further tomorrow

Already a big thanks for the help

no problem!

I was testing with caddy locally.

I discovered the following:

When I run caddy locally from the root of the project/monorepo and run the command
caddy run --config apps/admin-dashboard/Caddyfile, then it's not accessible on PORT 3000 . When i go to 8080 I receive a 404 error.

When I navigate into the FE folder (admin-dashboard), and run the same command caddy run --config Caddyfile, it's the same issue (404).

But when i just run caddy run in the FE folder, than it works locally (without the --config)

Which leads me to believe there is something still misconfigured in the Caddyfile ?

Ok, changing the root back to apps/admin-dashboard/Caddyfile fixed it. It's now up

Now I am figuring out how to do a reverse proxy so that api calls to /api/products go to the backend service.

I was able to figure out how to have the caddy reverse proxy working locally:

# global options
{
    admin off # theres no need for the admin api in railway's environment
    persist_config off # storage isn't persistent anyway
    auto_https off # railway handles https for us, this would cause issues if left enabled
    # runtime logs
    log {
        format json # set runtime log format to json mode
    }
    # server options
    servers {
        trusted_proxies static private_ranges 100.0.0.0/8 # trust railway's proxy
    }
}

# site block, listens on the $PORT environment variable, automatically assigned by railway
:{$PORT:3000} {
    # access logs
    log {
        format json # set access log format to json mode
    }

    # health check for railway
    rewrite /health /*

    handle /api/* {
        reverse_proxy localhost:3001
    }

    # Serve static files and handle client-side routing
    handle {
        # Serve files from the 'dist' folder
        root * apps/admin-dashboard/dist
        file_server

        # Rewrite all other requests to 'index.html' for client-side routing
        try_files {path} /index.html
    }

    # enable gzipping responses
    encode gzip
}

I had to do those two handle blocks, because try files was interupting my reverse proxy.
So it was always either my reverse proxy working and not my client side routing or the other way around.
By working with the two blocks, that got resolved.

I tried deploying it to Railway. Client side routing works, but the call to the backend via the reverse proxy takes a long time when i change this line

Before

   handle /api/* {
        reverse_proxy localhost:3001
    }

After

   handle /api/* {
        reverse_proxy {$BACKEND_BASE_URL}
    }

I have set the BACKEND_BASE_URL variable in my FE, but it doesn't work …

may I ask why the reverse proxy is needed? simply have your frontend call the public domain of the backend, the reverse proxy is very likely needless complications

Would've liked to hide my backend domain. But it's more of an optimization actually.

I personally don't see that as an optimization, but as long as it's working!

!s