At the moment, the RAG API does not provide proper feedback on the upload of the file to an embedding endpoint/backend via an OpenAI API compliant server. This results in the user thinking a file was uploaded successfully when it could have failed silently in the background. The RAG API also doesn't check and gracefully handle the upstream health, or lack thereof, of the Open AI API compliant server. These issues result in the need for 2 succinct, but highly connected mini-features:
[ ] Have an upstream healthcheck prior to all requests or at start-up
[ ] Block response back to the client until file is uploaded successfully or upstream fails to embed and upload the file to document store
At the moment, the RAG API does not provide proper feedback on the upload of the file to an embedding endpoint/backend via an OpenAI API compliant server. This results in the user thinking a file was uploaded successfully when it could have failed silently in the background. The RAG API also doesn't check and gracefully handle the upstream health, or lack thereof, of the Open AI API compliant server. These issues result in the need for 2 succinct, but highly connected mini-features: