Consistent MMQTTException: Connection Refused - Unauthorized after 2-2.5 hours uptime

[BlitzCityDIY]

Hello,

Recently I've noticed an issue where after approximately 2-2.5 hours of uptime, a KeepAlive period elapses and the code can't get a PINGRESP from the server and errors out:

7566.375: DEBUG - KeepAlive period elapsed - requesting a PINGRESP from the server...
7566.377: DEBUG - Sending PINGREQConnection error, reconnecting
PINGRESP not returned from broker.
7690.494: INFO - - iot_mqtt :: reconnect ::
7690.496: DEBUG - Attempting to reconnect with MQTT broker
7690.498: DEBUG - Attempting to establish MQTT connection...
7690.502: INFO - Establishing a SECURE SSL connection to iotc-----.azure-devices.net:8883
7691.429: DEBUG - Sending CONNECT to broker...
7691.434: DEBUG - Fixed Header: bytearray(b'\x10\xa7\x02\x00') Variable Header: bytearray(b'\x04MQTT\x04\xc2\x00x')
7691.445: DEBUG - Receiving CONNACK packet from broker
Traceback (most recent call last):
File "code.py", line 232, in
File "adafruit_azureiot/iotcentral_device.py", line 193, in reconnect
File "adafruit_azureiot/iot_mqtt.py", line 439, in reconnect
File "adafruit_minimqtt/adafruit_minimqtt.py", line 770, in reconnect
File "adafruit_minimqtt/adafruit_minimqtt.py", line 523, in connect
MMQTTException: Connection Refused - Unauthorized

There is a try and except in the loop:

except (ValueError, RuntimeError) as e:
        print("Connection error, reconnecting\n", str(e))
        wifi.radio.enabled = False
        wifi.radio.enabled = True
        wifi.radio.connect(secrets["ssid"], secrets["password"])
        device.reconnect()
        continue

But it doesn't seem to trigger a reconnection. Have tried with CP 7.3 and CP 8 on an ESP32-S2 Feather TFT and a QT Py ESP32-S2 both with my own project code and the example library code for IoT Central.

[brentru]

@jimbobbennett This seems to be an issue where the PING/PINGRESP packet is dropped. I am unsure if it's at Azure's end, or CircuitPython. Could you possibly shed some light on what's happening for @BlitzCityDIY ?

[jimbobbennett]

Can you confirm how long till the connection drops? The token that is generated is valid for 6 hours by default, and after that you will need to re-generate the device client object to get a new valid token. You can also change this by setting a longer expiry when creating the device client.

[BlitzCityDIY]

hi @jimbobbennett - it was consistently dropping around the 2 hour mark.

[jimbobbennett]

Weird. What time zone are you in? If you are 4 hours from UTC then maybe something has gone wrong with how it computes the expiry time, so the 6 hours becomes 2 hours plus a UTC offset of 4 hours.

[BlitzCityDIY]

Oh interesting! I am -4 UTC. I'll try running it without the offset to see if the same issue occurs.

[jimbobbennett]

Hopefully this is the issue! I guess we need better documentation around expiry and how to reconnect though.

[BlitzCityDIY]

So I changed the tz_offset to 0 from -4 and it did stay up for 6 hours. However, it did not reconnect and threw the same MQTTException

Traceback (most recent call last):
File "code.py", line 232, in
File "adafruit_azureiot/iotcentral_device.py", line 193, in reconnect
File "adafruit_azureiot/iot_mqtt.py", line 439, in reconnect
File "adafruit_minimqtt/adafruit_minimqtt.py", line 770, in reconnect
File "adafruit_minimqtt/adafruit_minimqtt.py", line 523, in connect
MMQTTException: Connection Refused - Unauthorized

This is the except portion of the code. I added OSError and broad Exception to try and catch the MQTTException:

except (ValueError, RuntimeError, OSError, Exception) as e:
        print("Connection error, reconnecting\n", str(e))
        wifi.radio.enabled = False
        wifi.radio.enabled = True
        wifi.radio.connect(secrets["ssid"], secrets["password"])
        device.reconnect()
        continue

[jimbobbennett]

reconnect doesn't create a new SAS token with an updated expiry, you will need to create a whole new device connection object. Maybe something to be improved in the API.

[BlitzCityDIY]

Oh I see. To generate a new SAS token then, should we be having folks reload code.py? Or is there a better method you'd suggest?

[jimbobbennett]

Recreate the device object - its in the construction of that object that the SAS token is create.