r2r 3.3.31 version | Streaming returns only <completion> , But without stream working fine

[NajiAboo]

Describe the bug
When i tried with RAG with agents in 3.3.31 version, When i stream it its not wokring. But When i ask same in non stream its giving me answers.

response = client.retrieval.agent(
message={"role": "user", "content": "questin here"},
search_settings=search_settings,
rag_generation_config={"stream": True},
conversation_id=r1["results"]["id"]
)

for chunk in response:
print(chunk)

Resulst is :

But same configuration with non stream ( I removed actual answer from it ) working fine.

'results': {'messages': [{'role': 'function',
'content': 'Vector Search Results:\nSource ',
'name': 'search',
'function_call': None,
'tool_calls': None,
'tool_call_id': None},
{'role': 'assistant',
'content': 'In.',
'name': None,
'function_call': None,
'tool_calls': None,
'tool_call_id': None}],
'conversation_id': '48b5afaf-bc84-4b78-9a06-348f98372eae'}}

So what need to be done to fix it. Last two days i am on it..

[emrgnt-cmplxty]

Can you try updating to 3.4.xx?

[NajiAboo]

Thanks for the support. Now i have upgraded. Getting below error
message": "peer closed connection without sending complete message body (incomplete chunked read)"

In between i am checking my code. I just want to give the update from my side. Please let me know if i need to change anything.

[NajiAboo]

r2r_support_team.zip
I am using the latest version of r2r ,which is 3.4.1 ( as of today ). I have attached may code sample. Its working fine with search, rag , but failed in agent. Also, I used the same code from the tutorial for reasoning agent, thats also failing.

Please help how to proceed. I have suggested the framework to the customer, convinced him about the framework. So i am in a bad situation as i touched the deadline