I think there’s a few sections of the documentations that really misses some key, subtle but important points.
JavaScript on Fly.io · Fly Docs or Run a Python App · Fly Docs
both of the code examples in these seem to run at a different port (3000 and 5000) which is different to the internal_port in fly.toml which gets generated. The other apps just co-incidentally (or not) listen on 8080. But nowhere in the docs does it specify that the internal_port should match the port that the app is running on. I think this is key especially under the Configure the App for Fly
It’s also not instantly clear on which tier of the Postgres offering is free.
The recent blog post introducing free postgres only touches on this.
Free Postgres would start with a single “Development” node, but if you scale it, it seamlessly becomes a leader-replica cluster for high availability.
I believe shared-cpu-1x and a 1GB volume is the ‘free tier’ option and you can run 3x of them?
And on Fly Postgres · Fly Docs
It is stated that
Note that for the app to be able to attach to the database it must have private_network enabled in
fly.toml. This enables it to look up.internaladdresses like the database URL. See Using Fly DNS for further details.
But no where on the page does it talk about private_network as a fly.toml config. I could not find this on the fly.toml reference either.