Open avo-sepp opened 1 year ago
Hi @avo-sepp :) This is a great idea - makes total sense.
I think we can start with a guide in the docs on how to do it per data-fetcher you create with Pydantic (you can fork a fetcher for a specific case and enforce it's schema in the process code).
We'll need to think on how to best make this generic- I'm guessing this can be another field in the SourceEntry.
Yeah, that's exactly what I'm going to do for now in the Cosmos fetcher to get it out the door now. But I think it should be generic eventually. Thanks @orweis!
Describe the solution you'd like A bad OPAL fetch can result in unexpected and hard to troubleshoot errors. Especially if the maintainers are not as familiar with the original system. (Interns, New Hires, etc.) I would like to be able to supply a JSON-Schema along with any fetcher configs. OPAL will validate the output of
process()
from whatever fetcher got the data against the JSON-Schema. If the validation fails, OPAL should report "unhealthy" on its health check endpoint. Preventing Kubernetes rollouts from continuing a rollout.