Hey guys, I have what might be the first “change of behaviour” between 0.12.X and 0.13 🙂 I was mid update of the EC2 instance we host Prefect Server on, and looks like the prefect install pulled 0.13.0, which is great, apart from one minor flaw - my deployment script doesn’t work anymore. Previously I was able to get the Prefect Server instance to talk to itself over graphql by setting the following env variables before running

prefect backend server

&& `prefect server start`:

export PREFECT__CLOUD__API=http://$(curl <http://169.254.169.254/latest/meta-data/public-ipv4):4200/graphql>
export PREFECT__SERVER__UI__GRAPHQL_URL=http://$(curl <http://169.254.169.254/latest/meta-data/public-ipv4):4200/graphql>

where the above curls evaluate to the public IP of the EC2 instance, and then previously the various containers would be able to talk to the graphql container, but it’s not working since the update to 0.13.0. Instead, in the UI I now get the following:

Couldn't connect

<http://localhost:4200/graphql>

Any suggestions? Different env vars needed, something else?

Hi @Darragh! I think these ENV vars have changed

Thanks Dylan!!

Hey @Darragh, take a look at: https://github.com/PrefectHQ/server/blob/master/src/prefect_server/config.toml

Thank Dylan, I'll try replace the prefix in a few minutes and let you know how it goes!

Hi @Darragh, I’d try to enact a change on your local machine using a config.toml and then try to produce the corresponding env var

Hi Dylan, don't worry for now, I've gone back to 0.12.6 for the moment 👍

Makes sense!