-
-
Notifications
You must be signed in to change notification settings - Fork 454
Description
Body:
Hello, I am trying to use the official OpenAI API with my self-hosted AppFlowy Cloud instance, but I'm unable to get a response in the client app.
Current Behavior:
When I use an AI command in the app, it shows the message "operation completed successfully" but no AI-generated text is displayed.
My Environment:
Self-hosted using the latest Docker Compose setup.
Images used: appflowyinc/appflowy_cloud:latest and appflowyinc/appflowy_ai:latest.
Exhaustive Troubleshooting Steps Taken:
I have worked extensively to debug this and can confirm the following:
My OpenAI API key is valid and has billing enabled. A direct curl test to https://api.openai.com/v1/models from inside the ai container returns a 200 OK with the full list of models.
The ai container has perfect network and DNS connectivity to the internet.
My .env file is correctly configured with OPENAI_API_KEY, OPENAI_API_BASE is commented out, and DEFAULT_AI_MODEL is set to gpt-4o to match the client's expectations.
I fixed an initial 502 Bad Gateway error by setting AI_SERVER_HOST=ai in the .env file, which successfully restored communication between the appflowy_cloud and ai services.
The ai service logs show a clean startup and no errors when the request from the app is made.
Conclusion:
All infrastructure, networking, and configuration appear to be correct. The fact that the app receives a "success" status but no content strongly suggests there is an application-level bug in how the ai or appflowy_cloud service parses, handles, or transmits the final payload received from the OpenAI API.