Hello! I'm running prefect server on GKE using the helm chart. I have an ingress for the UI secured with oauth and a load balancer for Apollo. I'm able to use the graphql endpoint from Apollo and the CLI. However, when it comes time to connect to the graphql endpoint in the UI I'm getting an error message:

Oops! It looks like something went wrong when trying to connect; make sure Prefect Server is running at the URL above and try again.

Any advice on how to debug this? GKE already creates a firewall rule to allow ingress over tcp:4200. Maybe something else weird with the firewall? I previously had Apollo set up as in ingress, but I switched back to a load balancer in case that would simplify things.

Not sure what is happening here. When you say use the CLI to hit Apollo, that is from another pod/machine?

I was using my local machine. I could hit the health endpoint and run queries.

I think to expose the UI, you also may need to open port 8080

After doing some more digging, it looks like a CORS issue as i'm running the UI and Apollo in different subdomains.