Deployed in wrong region

This is a branch of this thread: Always exactly one machine - #16 by eipe

I have this command flyctl deploy --app <appname> --ha=false --remote-only --regions fra -c fly.pull-request.toml db

But I can now see that I have machines deployed in San Jose instead.

What I expect is one of either

  1. Machine(s) are deployed where you specify to deploy it, in my case Frankfurt
  2. Or, if the platform cannot do that for whatever reason: throw an error

What I do not expect is that I get machines in some random region.

PS. My organization is on the launch plan (I saw that fra was limited to launch plan and higher). Nevertheless, the basic principle should still hold: deploy in specified region or throw an error.

EDIT:

This is just getting weirder. I now ran:
flyctl deploy --app <app-name> --ha=false --remote-only --regions arn -c fly.pull-request.toml db

And as a result a machine was deployed in ewr!

EDIT2:
In now destroyed the app, added a primary_region in the toml file and removed the --regions flag in the deploy cli command. It now deployed to arn as excepted. We’ll see if subsequent deploys also do so. But I have to be honest, this feels unreliable and unpredictable.

I’m not sure what the intended behavior is, but when you deploy an app for the first time, the primary_region is stuck, ie if you fly deploy --regions another-region, another-region is ignored.
It will create up to 2 machines in the same region as your primary_region (I believe that’s why you’re seeing 4 machines)

Sounds like a flyctl bug, can you do a bug report via their support? I’m on the poor plan so all my messages get rejected.

Well yes, I of course expect the --regions argument to be respected. Anything else is counterintuitive. Switching to specifying region in the toml file seems to have fixed it, but if primary_region is a required fly toml key I would expect fly to give an error if it is omitted.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.