You would want to remove that config as code setting as Hobby workloads are now deploying to metal exclusively.

You would want to remove that config as code setting as Hobby workloads are now deploying to metal exclusively.

Where was this change communicated and why is this taking place?

In all honesty, this change should be transparent, there shouldn't have been any errors thrown, the region and runtime was supposed to be switched behind the scenes automatically.

As for why, because we are moving off of GCP onto the metal region.

Gotcha, well I can't blame ya for wanting to move from GCP onto bare metal. Probably the best way to save $$$ in the cloud world.

What I'm seeing that is odd, is that my config tells railway a few things. The region mismatch is causing deployment failures but also I tell railway to use Docker but in the deployment, Nix is being used:

{ "$schema": "https://railway.com/railway.schema.json", "build": { "builder": "NIXPACKS" }, "deploy": { "runtime": "V2", "numReplicas": 1, "sleepApplication": false, "multiRegionConfig": { "us-west2": { "numReplicas": 1 } }, "restartPolicyType": "ON_FAILURE", "restartPolicyMaxRetries": 10 } }

vs: https://github.com/the-hideout/stash/blob/main/railway.json

I'll try to force set the region to us-west2 and see what happens

Nix is going to be used if a Dockerfile is not found.

I do have a Dockerfile but I'm guessing that since the deployment error related to regions kills the deployment before railway even starts scanning for what files live where. Could also explain why the settings seem to mismatch

Yep Dockerfile or Nixpacks don't come into play if you are trying to deploy to the wrong region and getting an error.

Okay cool well I got deployments working again now! I'm on the V2 metal beta and I know I have been a little snarky and complaining about this. I'll stick on the V2 metal beta this time around and report back if I see any large cost increases like I did last time just so that others can be aware if this is the case

Thanks for your quick replies on a Saturday Brody

To be completely transparent, the increase you see on the V2 runtime will not be addressed.

The legacy runtime was under counting CPU usage for some workloads.