asyncio, multiple servers and [Errno 104] ECONNRESET

[davefes]

I am trying to run 3 servers each listening to a different port on one ESP32 (1.24 SPIRAM and ESP32-GENERIC) and sometimes get [Errno 104] ECONNRESET

The relevant part of the code:

def main():
    print('awaiting client connection ...')

    asyncio.create_task(heartbeat(1000))

    servers = []

    for i in range(3):
        print("Starting server {0}".format(i))
        server = await asyncio.start_server(handle_msg, config.server, 8089+i)
        servers.append(server)

  # serve forever
    async with servers[0], servers[1], servers[2]:
        while True:
            await asyncio.sleep(3600)

# connect to hotspot
connect()

asyncio.run(main())

I was getting this error when running only one server and hoped running three servers would sort it. As far as I understand the 3 servers are associated with the 3 different ports. The error seems to occur when two clients are sending messages at roughly the same time.

Any suggestions are where to look for the problem?

Thank you.

Edit:
maybe relevant:
#13533
micropython/micropython-lib#788

[davefes]

I thought it might be a problem with all 3 servers using the same callback, so made 3 servers calling their specific callback function. Still get the error. Seems to be something fundamentally wrong in my attempts.

def main():
    print('awaiting client connection ...')

    asyncio.create_task(heartbeat(1000))

    server1 = await asyncio.start_server(handle_ping, config.server, 8089)
    server2 = await asyncio.start_server(handle_gate_alarm, config.server, 8091)
    server3 = await asyncio.start_server(handle_camera_alarm, config.server, 8090)

  # serve forever
    async with server1, server2, server3:
        while True:
            await asyncio.sleep(3600)

# connect to hotspot
connect()

asyncio.run(main())

[davefes]

Well, found a work-around ... whether or not it is a bug is the question.

For example in my:

def handle_ping(reader, writer):
    peer = writer.get_extra_info('peername')

    try:
        while (data := await reader.read(100)):
            message = data.decode()
            print(f'Received {message!r} from {peer!r}')

            if message == 'ping' and config.watchdog is True:
                wdt_feed()
                wdt.feed()

            print(f'Send: {message!r}')
            writer.write(message.format(peer).encode('utf-8'))
            print('got this far in ping')
            await writer.drain()

if I add the print('got this far in ping') then my ECONNRESET errors disappear.

If someone can tell me if I am doing something wrong, before I take the time to learn how to file a bug report, I would appreciate that.

[davefes]

Another work-around as per:
https://github.com/orgs/micropython/discussions/11901
await asyncio.sleep(0.1) after await writer.drain() seems to fix it.

[davefes]

Apparently, one server can handle multiple clients on a single port ... so I am closing this and going back to a single server.