296a4ed8-9ce2-4313-8827-59fc680abefb

I thought by doing this https://github.com/dClimate/jupyter-notebooks/blob/main/railway.json#L31 it would work but it doesn't appear to

This is what it looks like when it works

Essentially I want to programtically from the template map and expose the appropriate port for the public domain instead of the default railway one https://docs.railway.com/guides/public-networking

set the PORT variable to 8888

Hmm not sure what you mean? where?

in the service variables

in the railway.json?

I guess I'm confused, how does this help with creating a template that automatically sets the generated domain to forward to the appropriate port?

setting the PORT variable will tell the domain to point to 8888

So where I'm confused is where/how is that set automatically within the template?

would it be this within railway.json?

...
    "variables": {
        "JUPYTER_TOKEN": {
            "description": "Token for Jupyter authentication",
            "required": true,
            "sensitive": true
        },
       "PORT": {
           "value": 8888
        }
    }
...

is that a railway reference variable? if it sees the word port it will forward to 8888 instead of 8080?

it's just a regular variable

Again so my question is how will railway know to point to that value? I have 4 services running on different ports but the publicly generated domain should only point to 8888, by setting "PORT" in railway.json as shown above will this solve the problem? This should happen in a way that anyone who also deploys the template will have it work automatically for them without having to edit or do anything on their end

the template is here if you'd like to try fwiw https://railway.app/template/6Iv9BG?referralCode=1CR-cB

I'm not sure why railway.json keeps being used, I had previously mentioned service variables.

how will railway know? because that's just how it's been designed to work, don't know what to tell you here.

I think something is getting lost in translation

have you tried it?

I mentioned railway.json because this is how the project is being deployed from the template correct?

no

how can I create a repeatable programmatic file that when another user deploys from a template

you set PORT on the template service

they can automatically have the ports adequately setup

So like this?

yes, no need to over think this

do I need to update the template

now that I edited the original service?

save the changes, yes

save changes on the service that's already deployed or update the template? for the former I saved it but for the latter I see no option to have changes propagate

You'd need to redeploy the template

I just did that and there seems to be no difference

nvm

works now

thanks @Brody (was a slow refresh)

No problem!

hmm this wasn't obvious at all from the documentation, PORT seeming like a "magic" variable

just gotta trust me, I've been doing this for a long time

is there any way you guys can make this clear on the docs as I ctrl'd f'd maybe I missed it

yeah I do I just wasn't clear on exactly what needed to be done

as it was strange to think that things had to be done in a UI vs railway.json(the latter being only for deploying) whereas template is using the UI for the former

Can you guys add a sentence here

that this is the port that will be exposed on the domain as the default instead of the provided one?

will make a note of that

basically the way I write docs now is assuming an LLM will parse it

so the more information the better, even in an appendix to not overwhelm users

but this way more info is better

haha I'm not going that, but I will add the extra information

one more question while we're here, I see when I go to deploy the template now there is an option for the user to input details, is there a way to pre-fill it?

What I mean by that is I see this

to at least pre-fill port with 8888

set the variable to 8888 in the template composer then

hmm where is that?

this?

ah it took another second to update

okay perfect thanks again brody!

no problem