431 Request Header Fields Too Large

[trattaa]

DocsGPT is running in a Docker environment v24.0.2 on the Ubuntu 22.04.2, along with Traefik v2.10. Traefik has been properly configured for both the Frontend and Backend, with CORS enabled for the Backend. However, when making a GET request to the streams endpoint with a large payload, it is unable to handle such high volumes, resulting in a "431 Request Header Fields Too Large" error. This limitation prevents me from engaging in long dialogues due to the accumulated history or sending large text inputs.

However, when using POST requests to the Answer endpoint, I don't encounter such issues.

Is it possible to switch the streams endpoint to use POST requests instead of GET? Alternatively, could you provide any other solutions? Here'is the response from the Traefik developers regarding this issue, which I found.

[dartpain]

I think the only way to solve it especially for streaming is probably to store user session in the backend in mongodb.

unless you can try a more elegant or simple solution, im happy to hear

[edjoey]

| | Joey | | @.*** |

If the streaming media is too long, can you consider using GPT or other methods to simplify it, try to shorten the length as much as possible to ensure that the token is small and accurate, of course I am just an idea.

---- Replied Message ---- | From | @.> | | Date | 06/16/2023 07:24 | | To | @.> | | Cc | @.> , @.> | | Subject | Re: [arc53/DocsGPT] 431 Request Header Fields Too Large (Issue #261) |

I think the only way to solve it especially for streaming is probably to store user session in the backend in mongodb.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>