waiting for the deployment in the CI -> railway up --ci
generating a domain for it and extracting the url -> DOMAIN=$(railway domain)

I can't use any railway command in the CI is my issue. I get the error messages
Project Token not found
or
No linked project found. Run railway link to connect to a project, and a service.

I know the environment variable is there and set. I can run a check for the variable, I've regenerated it so many times and reset the github secrets, and I can use the latest token locally and get responses.

RAILWAY_API_TOKEN is not a project token though

It's an account token? https://docs.railway.com/guides/cli#account-tokens

yes

Do I not want to be using an account token? I am confused. I've gotten it to work locally. I even unlinked and reset railway CLI locally to make sure it wasn't using any pre-existing auth before I tried this flow locally.

Logs from github:

RAILWAY_API_TOKEN is set. Proceeding with Railway deployment.
RAILWAY_API_TOKEN is set. Proceeding with Railway deployment.
RAILWAY_TOKEN is not set. Proceeding with Railway deployment.

> Installing railway, please wait…
                   .
         /^\     .
    /\   "V"
   /__\   I      O  o             
  //..\\  I     .                             Poof!
  \].`[/  I
  /l\/j\  (]    .  O
 /. ~~ ,\/I          .               Railway is now installed
 \\L__j^\/I       o               Run `railway help` for commands
  \/--v}  I     o   .
  |    |  I   _________                    
  |    |  I c(`       ')o            
  |    l  I   \.     ,/
_/j  L l\_!  _//^---^\\_ 

Railway CLI installed successfully.
No linked project found. Run railway link to connect to a project, and a service.

Bash script:

if [ -z "${{ secrets.RAILWAY_API_TOKEN }}" ]; then
            echo "RAILWAY_API_TOKEN is not set. Skipping Railway deployment."
            exit 0
          else
            echo "RAILWAY_API_TOKEN is set. Proceeding with Railway deployment."
          fi

          if [ -z "$RAILWAY_API_TOKEN" ]; then
            echo "RAILWAY_API_TOKEN is not set. Skipping Railway deployment."
            exit 0
          else
            echo "RAILWAY_API_TOKEN is set. Proceeding with Railway deployment."
          fi

          if [ "$RAILWAY_TOKEN" ]; then
            echo "RAILWAY_TOKEN is set. Skipping Railway deployment."
            exit 0
          else
            echo "RAILWAY_TOKEN is not set. Proceeding with Railway deployment."
          fi

          curl -fsSL https://railway.app/install.sh | sh
          echo "Railway CLI installed successfully."

          POLL_INTERVAL=10
          TIMEOUT=600
          TIME_ELAPSED=0
          SERVICE_NAME="api"
          ENVIRONMENT=web-app-pr-${{ steps.pr-number.outputs.result }}

          railway environment $ENVIRONMENT

did you delete the config.json locally?

Where is that located at?

what even is your end goal here?

$HOME/.railway

I am using the preview environment to auto deploy. I then want to watch for my api service to be deployed, generate a domain name for it, then use that url in my vercel build.

may i ask what you are running on vercel that cant be ran on railway?

It's our nextjs app. I have slowly been convincing the team to move away from vercel, but it's a process.

However, I'd want the url from the deployment either way, as we run the playwright tests against the deployed preview environment to ensure that everything is good in a deployment environment before pushing to prod.

is it? one night we just decided to run railway.com on railway instead of vercel, and its been working great ever since

They are big fans of vercel and we are using nextjs, they want to keep the web apps on it for now, and see how the api goes first.

Getting this pipeline working, which I'd need even if we had it all on vercel because of how we run playwright tests, would be very beneficial in convincing the team 🙂

(railway.com is nextjs)

So are you trying to use the cli to create the PR env?

I am allowing railway to auto create the environment with the preview option turned on. I am just wanting to use the CLI to wait for it to be finished building (but using up --cimight be better if it reuses/closes environments on PRs for us), then generate the domain, and use it later in the github CI

so all you really want to do is get the domain for a service that exists within the pr env?

you mentioned generating a domain, but why wouldnt the service have a domain to begin with?

If it has a generated public domain already for previews, awesome! Ya, that is all I want 🙂

if the base service has a domain, so will the service in the pr env

I did delete my $HOME/.railway file, and the RAILWAY_API_TOKEN way I was doing it locally no longer works. Good call, so I am doing something wrong

Ah okay, so how can I get the domain for it? 🙂

honestly, youd likely want to write some javascript to call the api and get the domain for a given service in a given environment

Oh, :/

Well, I can do that. Kinda weird to have to do that imo.

I'll take a look at the graphql docs then for where I can ping for this info

1. project

2. environmentConfig

pretty sure those are the only two you need, the frontend uses the same api, so you can just look at the fetch requests it makes

Thanks for the help!

let me know if you are having any trouble with the api