issue: ClientPayloadError: Response payload is not completed: TransferEncodingError

[poonesh]

Check Existing Issues

• I have searched the existing issues and discussions.
• I am using the latest version of Open WebUI.

Installation Method

Pip Install

Open WebUI Version

v0.5.20

Ollama Version (if applicable)

No response

Operating System

ubuntu 22.04

Browser (if applicable)

No response

Confirmation

• I have read and followed all instructions in README.md.
• I am using the latest version of both Open WebUI and Ollama.
• I have included the browser console logs.
• I have included the Docker container logs.
• I have listed steps to reproduce the bug in detail.

Expected Behavior

I have set up openwebui on EKS alongside bedrock-access-gateway following the instructions here: https://docs.openwebui.com/tutorials/integrations/amazon-bedrock

I expect to receive responses from the bedrock models, but occasionally do not.

Actual Behavior

I am occasionally getting a hang when I try to talk to the model, where I don't get a response and instead just the grey loading boxes, with errors from the docker container.

Steps to Reproduce

No specific pattern, happens intermittently, but frequently. Usually after a period of time of inactivity, but not always.

Logs & Screenshots

2025-03-24 21:17:36.826 | ERROR    | asyncio.runners:run:120 - Task exception was never retrieved
future: <Task finished name='Task-6229' coro=<process_chat_response.<locals>.post_response_handler() done, defined at /venv/lib/python3.11/site-packages/open_webui/utils/middleware.py:1114> exception=ClientPayloadError("Response payload is not completed: <TransferEncodingError: 400, message='Not enough data for satisfy transfer length header.'>")> - {}
Traceback (most recent call last):

  File "/venv/lib/python3.11/site-packages/aiohttp/client_proto.py", line 92, in connection_lost
    uncompleted = self._parser.feed_eof()
                  │    └ None
                  └ <aiohttp.client_proto.ResponseHandler object at 0x7f74f58dfb60>
  File "aiohttp/_http_parser.pyx", line 508, in aiohttp._http_parser.HttpParser.feed_eof
    raise TransferEncodingError(
          └ <class 'aiohttp.http_exceptions.TransferEncodingError'>

aiohttp.http_exceptions.TransferEncodingError: 400, message:
  Not enough data for satisfy transfer length header.


The above exception was the direct cause of the following exception:


Traceback (most recent call last):

> File "/venv/lib/python3.11/site-packages/open_webui/utils/middleware.py", line 1719, in post_response_handler
    await stream_body_handler(response)
          │                   └ <starlette.responses.StreamingResponse object at 0x7f74f5963250>
          └ <function process_chat_response.<locals>.post_response_handler.<locals>.stream_body_handler at 0x7f74f530b880>
  File "/venv/lib/python3.11/site-packages/open_webui/utils/middleware.py", line 1471, in stream_body_handler
    async for line in response.body_iterator:
                      │        └ <StreamReader e=ClientPayloadError("Response payload is not completed: <TransferEncodingError: 400, message='Not enough data ...
                      └ <starlette.responses.StreamingResponse object at 0x7f74f5963250>
  File "/venv/lib/python3.11/site-packages/aiohttp/streams.py", line 52, in __anext__
    rv = await self.read_func()
               │    └ <member 'read_func' of 'AsyncStreamIterator' objects>
               └ <aiohttp.streams.AsyncStreamIterator object at 0x7f74f652ff10>
  File "/venv/lib/python3.11/site-packages/aiohttp/streams.py", line 352, in readline
    return await self.readuntil()
                 │    └ <function StreamReader.readuntil at 0x7f76140cf600>
                 └ <StreamReader e=ClientPayloadError("Response payload is not completed: <TransferEncodingError: 400, message='Not enough data ...
  File "/venv/lib/python3.11/site-packages/aiohttp/streams.py", line 386, in readuntil
    await self._wait("readuntil")
          │    └ <function StreamReader._wait at 0x7f76140cf4c0>
          └ <StreamReader e=ClientPayloadError("Response payload is not completed: <TransferEncodingError: 400, message='Not enough data ...
  File "/venv/lib/python3.11/site-packages/aiohttp/streams.py", line 347, in _wait
    await waiter
          └ <Future finished exception=ClientPayloadError("Response payload is not completed: <TransferEncodingError: 400, message='Not e...

aiohttp.client_exceptions.ClientPayloadError: Response payload is not completed: <TransferEncodingError: 400, message='Not enough data for satisfy transfer length header.'>
...
2025-03-24 21:21:05.626 | ERROR    | aiohttp.client:__del__:443 - Unclosed client session
client_session: <aiohttp.client.ClientSession object at 0x7f74f5975a50> - {}

Additional Information

No response

[rmiranda87]

Has there been any work on this? I'm running into it as well.

[R-Bower]

Same. @rmiranda87 @poonesh have either of you found a fix?

[steveepreston]

@tjbck Hey man, can you please look at this.
facing same issue

the problem is: open-webui backend can't handle this error of aiohttp and model response box stuck in receiving/loading/thinking skeleton state for infinite. This means the "retry" button will never show and you are forced to start a new chat
it also ignores AIOHTTP_CLIENT_TIMEOUT=10 because there is a unhandled crash there
Thanks

Update: "Stop" button also doesn't stop it

[steveepreston]

@Classic298 @rgaricano Sorry guys can you please help

This unhandled error in the backend preventing model answer box to have a successful or unsuccessful state!
therefore it will not have any buttons (like "retry") and it stuck in loading/generating mode for the infinite time

[rsshilli]

I had this error and solved it. In my case I had an old version of the bedrock-gateway and I needed to upgrade to the latest one so I could use the latest Claude 4.5 models. To do that, I had to sync down the latest code and recreate the docker container. So something like:

# Download the latest Bedrock Gateway
pushd ~/bedrock-access-gateway
git pull
pushd src/
docker build . -f Dockerfile_ecs -t bedrock-gateway
popd
popd

That created a new docker image that runs on port 8080 (my old one was on 8900, and I don't know why it changed) that I then could use with the latest Claude 4.5 Haiku and Claude 4.5 Sonnet.