Initial database setup in Railway

[alex-alecu]

What's the best practice for setting up initially the database users in Railway?

I've tried creating a service that runs only migrations, but running ... triggers the following:

ERROR: Graphile Starter must run inside of a git versioned folder.

Which makes me believe that CONFIRM_DROP=true yarn setup:db is not intended ever for the prod db

[benjie]

What is Railway?

No, you should not drop in production. There are instructions on deploying to Heroku here:

https://github.com/graphile/starter#deploying-to-heroku

Deploying to other environments would have similarities.

[alex-alecu]

The documentation just says to use Amazon RDS, nothing about setting up the initial users, structure.

In an automated, continuous delivery environment, what's the recommendation for setting the db structure?

[benjie]

After following the setup instructions following the link above, on every release you should run:

yarn db migrate

as done in our Procfile for Heroku:

starter/Procfile

Line 3 in 0b68eab

release: yarn db wipe-if-demo && yarn db migrate