NB-IoT board: messages lost (Sodaq_NB-iot_All_Sensors sketch)

According to the debug output seen in the serial monitor, messages are all sent successfully, but often there are long periods of time (30 minutes - few hours) of not receiving any messages on the queue at all.

After that long period of no messages received, the last message will fail ([rdResp]: ERROR). Then a reconnect is attempted (adaptation of the sketch to do the reconnect was mine). After the reconnect has been successful, messages will be received on the queue, usually for a long time (hours).

This is with the board stationary indoors, tested on two locations, the behavior is the same. The reported signal quality (from AT+CSQ) is ~20 in both cases.

Is there some idea as to what causes this and if/when it will be fixed? I’m aware that it’s UDP, but missing all messages for hours makes it impossible to conduct the tests I have planned.

Hi,
I assume, you use the original ‘send message’ routine, correct? I don’t use this routine. Instead of this I’m creating a UDP socket by myself. And, doesn’t matter how bad the CSQ is, as long as connecting is possible, the message is going out.

@awk

Are the messages completely lost?
Or do all the messages come in at the same time?

Hopefully @EricBart from T-Mobile can help.
Maybe the T-Mobile backend can give us an indication what the problem is.

Best regards,
Jan

Thanks for your reaction Jan, the messages are lost. I sent sequence number in them to check. It is always a large number of consecutive messages that do not arrive (not just one or a few).

The network is still in a testing stage.
Both network and module are updated frequently.

Please let us and T-Mobile know about all the issues you have.

Together we can improve the nbiot experience :slight_smile:

Jan

Thanks Jan.
For the module updates we can check Git I guess, and is it possible to see somewhere when network updates are planned/completed and what was done? That would be a big help!

Today I received the first messages at 11:09 after having started two NB-IoT shields around 10:30. There were no message sending errors on the boards, so presumably the CoAP messages have been received OK. Probably the messages had not been forwarded from the T-Mobile network to the MQTT broker, because none had an earlier timestamp than 11:09.

Is there news on a fix for this issue?

Another question; Which qualityOfService settings are supported by the Allthinkstalk MQTT broker?