ESP32-C6 stop responding

[moferreira]

On my main.py I got a "while True" where I read RS485 data, send it via MQTT and write them to a file every 15 minutes. I Also print some of the information on the REPL. I do one RS485 read per second. Now, after running for a while it will stop printing data but will keep sending MQTT. After one or two hours it will also stop sending MQTT messages. However, in my code I have:

def main():
    while True:
        try:
            ...
            ...
        except KeyboardInterrupt:
            abort = input(f"Stop application? (y/n): ").lower()
            if abort == 'y':
                return

While it is not printing/sending MQTT message, if I press just "ctrl + c" it starts working again as if it "wake-up" from sleeping process.

I am running: release='1.24.0-preview', version='v1.24.0-preview.449.g1b89c503d on a ESP32-C6.

Anyone got a suggestion on what the problem could be?

[bidrohini]

If your MQTT connection is not handled properly (e.g., no keep-alive mechanism), the connection might drop after some time without sending heartbeats.

[sosi-deadeye]

Use the async approach. You can have two different tasks in a single thread. One async worker for MQTT and one async worker for UART. Dealing with StreamReader and StreamWriter is similar to CPython. To exchange data between the two workers could use this implementation of a ThreadSafeQueue: https://github.com/peterhinch/micropython-async/blob/master/v3/docs/THREADING.md#22-threadsafequeue

[peterhinch]

There are a few reasons why an MQTT link can die. The handling of MQTT PING packets is one which can cause the broke to disconnect (as per @bidrohini). Another cause is that WiFi links can suffer brief outages - the official MQTT clients do not recover after an outage. This asynchronous client does.

[moferreira]

Sorry for this late response... I was facing WiFi network problems. Once it got the WiFi fixed, then the problem disappeared. It is running for over 15 hours now and, in the last 85,000 reads from the serial and sent via MQTT, I didn't have any problem so far. So, is is more like that the MQTT client does not deal well with network problems. My MQTT send message function, that is called from main() loop, was:

def mqtt_send_msg(message="This is a MQTT test."):
    import ssl, machine
    from umqtt.simple import MQTTClient
    mqtt_config   = load_config(mqtt_config_path)
    if mqtt_config['mqtt_ssl'] == 'enabled':
        context = ssl.SSLContext(ssl.PROTOCOL_TLS_CLIENT)
        context.verify_mode = ssl.CERT_NONE
        client = MQTTClient(client_id=cpu_id,
                            server=mqtt_config['mqtt_server'],
                            port=mqtt_config['mqtt_port'],
                            user=mqtt_config['mqtt_user'],
                            password=mqtt_config['mqtt_password'],
                            ssl=context)
    else:
        client = MQTTClient(client_id=cpu_id,
                            server=mqtt_config['mqtt_server'],
                            port=mqtt_config['mqtt_port'],
                            user=mqtt_config['mqtt_user'],
                            password=mqtt_config['mqtt_password'])
    try:
        client.connect()
        client.publish(mqtt_config['mqtt_topic'], message)
        client.disconnect()
    except Exception as e:
        print("Error:", e)
        return False

As it can be seen, I call it, send the message and close it for every single message. In my mind, this would prevent any "memory leak" the MQTT client could have. Anyway, even having a mistake on the code, it should print an error message if the MQTT client fails somehow. This brings the question is: if I have an try/except on the function, why it does catch when something goes wrong?