FAST API memory leak issue

[zhangyukun230]

First Check

• I added a very descriptive title here.
• I used the GitHub search to find a similar question and didn't find it.
• I searched the FastAPI documentation, with the integrated search.
• I already searched in Google "How to X in FastAPI" and didn't find any information.
• I already read and followed all the tutorial in the docs and didn't find an answer.
• I already checked if it is not related to FastAPI but to Pydantic.
• I already checked if it is not related to FastAPI but to Swagger UI.
• I already checked if it is not related to FastAPI but to ReDoc.

Commit to Help

• I commit to help with one of those options 👆

Example Code

@app.websocket("/ws/{client_id}")
async def websocket_endpoint(websocket: WebSocket, client_id: str):
    await websocket.accept()
    user_agent,user_memory = init_user_sever_agent()
    request_word_dir = local_directory + '/' + client_id
    append_files = []
    history:List= []
    try:
        while True:
            try:
                data = await websocket.receive_text()
            except WebSocketDisconnect:
                logging.info(f"Client {client_id} disconnected.")
                break
            message = json.loads(data)
            content = ''
            # 只有文本内容传输
            if message['type'] == 'text':
                content = message['content']

            # 文本内容和文件一起传输。 文件分块上传。
            elif message['type'] == 'file_chunk':
                chunkMessage = message['chunkMessage']
                if chunkMessage['content'] == '<END>':
                    content = message['content']
                else:
                    file_name = chunkMessage['fileName']
                    chunk_index = chunkMessage['chunkIndex']
                    total_chunks = chunkMessage['totalChunks']
                    chunk_content = base64.b64decode(chunkMessage['content'])

                    file_dir = os.path.join(request_word_dir, client_id + '_' + file_name)
                    os.makedirs(os.path.dirname(file_dir), exist_ok=True)
                    chunk_file_path = f"{file_dir}.part{chunk_index}"
                    with open(chunk_file_path, "wb") as f:
                        f.write(chunk_content)

                    # 检查是否所有分片都已接收完毕
                    if all(os.path.exists(f"{file_dir}.part{idx}") for idx in range(total_chunks)):
                        with open(file_dir, "wb") as final_file:
                            for idx in range(total_chunks):
                                chunk_file_path = f"{file_dir}.part{idx}"
                                with open(chunk_file_path, "rb") as chunk_file:
                                    shutil.copyfileobj(chunk_file, final_file)
                                os.remove(chunk_file_path)
                        print("文件的目录是: " + file_dir)
                        append_files.append(file_dir)

            if content == '':
                continue
            content_result = ''
            async for response in getLLMRequest(user_agent, content, request_word_dir, append_files, history):
                content_result += response
                await websocket.send_text(response)
                await asyncio.sleep(0)
            print(content_result)
            await websocket.send_text("<END>")
            append_files = []

    except Exception as e:
        logging.error(f"Error for client {client_id}: {e}")
    finally:
        logging.info(f"Client {client_id} disconnected. Agent removed.")
        del user_memory
        del user_agent
        gc.collect()


if __name__ == "__main__":
    import uvicorn

    logging.basicConfig(level=logging.INFO)
    uvicorn.run(app, host="0.0.0.0", port=80)

Description

Centos machine, Python 3.10
Using Fastapis to build websocket services,. Open some web pages, VIRT and RES increase a lot first, and after closing, RES memory only decreases slightly. But after opening a lot of web pages and then closing them, RES memory will decrease significantly, while VIRT will still increase significantly
This issue does not exist on Mac.
Why is this?
linux

mac

Operating System

Linux

Operating System Details

Centos7.2 machine, Python 3.10 fast api 0.111.0

FastAPI Version

0.111.0

Pydantic Version

2.7.1

Python Version

3.10

Additional Context

No response

[CrasCris]

Did you solve it ? , i have a api rest api, i just make what to endpoints does in a websocket, but even when i close the websocket i got a lot of memory

[zhangyukun230]

@CrasCris sorry, not solve it.