Hey there! We’re having issues using the auto-discovery feature for GCP that was recently introduced on our self-hosted GrowthBook instance. We have created a Service Account in our GCP project ‘wanted-weasel’ but want to read from BigQuery tables in a GCP project called ‘learning-honeybee’. Afair, service accounts can be used across GCP projects so I’m wondering why we’re getting this error. Why does it still ask for ‘wanted-weasel’ when I’ve configured ‘learning-honeybee’ as default project?

Hi Julian

this is a self-hosted version 🙂 (will add it to the question)

I think I know what the issue is. We are not specifying a project id when connecting with auto discovery. https://github.com/growthbook/growthbook/blob/main/packages/back-end/src/integrations/BigQuery.ts#L19

is there sth we can do from our side to help resolve this?

Using a JSON key file instead of auto discovery would work until we fix that bug. I need to do some more reading to make sure I fully understand BigQuery project ids and how they are used. We have both

projectId

and

defaultProjectId

currently, but I'm not 100% sure we need both.

Actually, the same problem occurs when using the JSON key file. I think the issue is the same for both auto-discovery and JSON key file.

It looks like the JSON key file was created for the

wanted-weasel

project. Is it possible to create one for the

learning-honeybee

project instead?

hi Jeremy, the idea here is that we’d like to make SA’s from project B doing things in project A

Ok, cool. I suspect the key file would also work if you manually edited the project id field

well, in our case it is not the best solution, as we create resources by declaring them in code and then if edit something manually, that can be overwritten at any moment by another terraform apply