Turbocharge your Heroku Apps with New Turboku

Summary

New Turboku clones a Heroku app on Fly.io without requiring data migration, while the resulting app should be able to use services such as Heroku Postgres or Redis through transferred environment variables. The rebuilt workflow integrates flyctl with Fly.io’s remote builders instead of AWS CodeBuild: it queries Heroku, selects a region, converts config variables to secrets, and generates fly.toml, an ephemeral Dockerfile, and a reconstituted Procfile from the slug and process map. A Phoenix LiveView web launcher wraps the deployment by starting a Fly.io Machine running flyctl, while the CLI is available through flyctl turboku. The launcher also configures webhooks and watches environment-variable changes for automatic updates, whereas the CLI lacks webhook integration; the source reports about one millisecond of added database latency for a US-based app.

Takeaways

1. 1

   flyctl turboku accepts a Heroku app name and token, uses an explicit --region when provided or otherwise selects the region nearest to the Heroku app, and generates deployment files from the Heroku slug.

2. 2

   The web launcher starts a Fly.io Machine running flyctl instead of duplicating the deployment implementation in Elixir, allowing users to authenticate both services and select a Heroku app through a GUI.

3. 3

   The web launcher supports Heroku webhooks and watches changes to variables such as DATABASE_URL; the CLI lacks webhook integration. The source measured about one millisecond of added database latency for a US-based app.