We try to do some automated checks to understand what kind of app is being deployed so we can make some best practice choices for the configuration. I wonder if that detection went wrong with your repo. What’s the architecture of your app? Language? Framework?
It’s also possible this was a transient failure for the deploy machine. Did you try it again after this initial failure?
I’ve not used the GitHub deploy method. I assume that the “automated checks” Alex refers to is the manner in which a Dockerfile is selected, or compiled, to match the needs of the application. I think this is very clever, and also a recipe for deployment instability.
I’d recommend building your own Dockerfile(s) from scratch, and then deploying them with the console (and without Depot) in your own repeatable CI process.
In my case, it was a Ruby project, and it doesn’t really have anything launchable. I was curious about what would happen or what Fly.io even means by Launching. I was hoping that investigating the inevitable failure would be informative, but it was just more confusing.