Shopify

Fly.io has built in support for Shopify Remix applications: simply scaffold, build, and launch.

– https://x.com/i/status/1892986447788966011

To see this in action, Scaffold an app and immediately proceed to fly launch:

shopify app init
cd app
fly launch

(substitute the name you gave to your app for app in the cd line above)

Configuration

fly launch parses the output of shopify app env show.

The value of SHOPIFY_API_SECRET will be set as a fly secret. The remainder will be placed into your fly.toml:

[env]
  PORT = '3000'
  SCOPES = 'write_products'
  SHOPIFY_API_KEY = '…'
  SHOPIFY_APP_URL = 'https://….fly.dev'

Scaling

By default, all Machines will automatically be configured to stop when not in use, and restart on the next request; this is fine for development purposes, but for production you will need to adjust this to meet Shopify performance requirements.

Depending on your requirements:

• suspend can generally reduce startup time from hundreds of milliseconds to a single digit number of milliseconds. Some applications may experience time skew issues, and there are other limits, so this is not the default.
• Setting min_machines_running to 1 in the http service section of your fly.toml can make sure that there always is a started machine to process requests.
• setting auto_stop_machines to off in the http service section will prevent your Machines from being stopped all together.

You can also Scale Machine CPU and RAM, and Scale the number of Machines.

Database

Shopify’s template starts you off with SQLite using Prisma. Of particular note:

• By virtue of only running on a single Machine, SQLlite3 apps will experience brief unavailabilty during deploys. For a typical application without any new migrations, this will be on the order of a few hundred milliseconds.
• For PostgreSQL applications without any volumes, rolling deploys are the default, avoiding any downtime.

Additional information is available on deploying Prisma apps on Fly.io.