I am load testing the async websocket, with emitting only one event and i am not able to connect even 1000 client to my server when i am emitting an event to all clients every 10 seconds.

[neerajdlh74]

this is how my server code looks.....

I need to connect 5000 client with the server emitting messages to all every 10 seconds

Right now i am able to connect 200 connections, after that
clients are getting disconnected because of high cpu usage

I have 2 core machine with 3.7 gb ram

import uvicorn
import asyncio
import socketio

url='url'
channel='public_chat'
mgr = socketio.AsyncRedisManager(url=url, channel=channel, write_only=False)
sio = socketio.AsyncServer(async_mode='asgi',client_manager=mgr, async_handlers=True, logger=True, engineio_logger=True)

app = socketio.ASGIApp(sio)

@sio.on('connect')
async def test_connect(sid, environ):
      print('connected', sid)




@sio.event
async def message(sid, message):
      await sio.emit('client', message)



@sio.on('disconnect')
async def test_disconnect(sid):
    print('Client disconnected')


if __name__=='__main__':
    uvicorn.run(app, host='0.0.0.0', port=61431)

[miguelgrinberg]

I'm sorry but I don't really know what do you need from me. Have you profiled your application? What is consuming so much CPU? The code that you pasted above doesn't do anything. It is even inconsistent with your description that it sends an event every 10 seconds.

Also, please don't write issues on this package when all you have is questions. The issues board is for bugs only. Use this discussion board for questions. Thank you.

[miguelgrinberg]

First of all, I have no way to confirm you have done this profiling correctly (I don't know what profiler you used, or how you configured it). But if we are to believe this, how can you explain this:

..134 AsyncSocket._upgrade_websocket  3921   0.128287  18.61914  0.004749

This indicates that there were 3921 WebSocket upgrades in this run. With 350 clients? How can that be? I think your test must having some issues.

[neerajdlh74]

the clients were disconnecting and reconnecting, the reason why websocket upgrades are so high.

here we can see how much disconnect were being called:

..AsyncServer._handle_eio_disconnect  3883   0.126660  2.795861  0.000720
..479 AsyncServer._handle_disconnect  3882   0.059634  2.633516  0.000678
..py:97 AsyncRedisManager.disconnect  3862   0.577131  2.293476  0.000594

[miguelgrinberg]

So you are interested in profiling a high disconnection/reconnection case? How is this a useful test? I have to be honest, I'm a bit frustrated with this thread. Wouldn't it be more useful to solve the disconnections before assessing CPU usage?

[neerajdlh74]

my apologies if you are frustrated because of me. i was just looking for answers that why disconnections are happening and cpu load increases with simple code from docs. Even the profiler says more disconnections. There can be the case of more disconnections and connections can take the cpu load. what could be the reason behind disconnections ? Need your support here.

[miguelgrinberg]

i was just looking for answers that why disconnections are happening

This is the first time you ask me about why disconnections happen. Your previous questions were all about supporting X number of clients.

Have you looked at the logs of your server? Error messages are almost always self-explanatory there.