I’m automating deployments on Fly.io with a GitHub Workflow that runs flyctl deploy --now --remote-only. This command ALWAYS fails on the first deploy to an app that was created/launched previously with --no-deploy, claiming that there’s “No deployment available to monitor” (example workflow run, line highlighted). This can be confirmed by going through our list of “preview deployment” workflow runs on GitHub - the first deployment of every new branch to a new app fails consistently (skip the latest runs after the 24th, these are irregular due to another bug with fly, will report right after this).
To work around this, as seen in the linked log, I end the command with an || sleep 15, and check the status manually after the deployment. This works around the problem for the first deployment, but now creates an issue with subsequent ones - when their deployments really fail because the build is bad, the workflow will pass because the app is still up (running an older version).
Are you aware of this issue? Is there something I can do on my end, other than some more bash wizardry?