Losing cloud connection, Thread CoAP Cloud example (SDK 0.11.0)

[Reto]

Dear Thread-experts,

I have troubles with your Thread CoAP Cloud example. I am using:

• Thread SDK V0.11.0
• Border Router V0.11.0, ethernet (not Wi-Fi)
• cloud_coap_client example, adapted that I send more data (buffer size =256bytes)
• Eclipse
• Preview DVK, V0.9.0

I observe that after a while (sometimes day, sometimes minutes), no data are sent to the cloud. I debugged and my FW is running find and I don't get any errors (function otCoapSendRequest() is executed without error). However, the data is not visible in the cloud. The only way to recover it is: erasing the FW on the Border Router Network-Co-Processor-Thread-Module (Preview DVK). After erasing and reprogramming the DVK with the NCP, I can send data again for a certain time, until it brakes down again.

What I have observed is: when I send only 64bytes (insted of 256bytes): it tends to run more stable (some samples are missing, but the system is recovering itself).

I send the 256bytes every 6 minutes.

How can I debug that issue in more detail?

Thank you, Reto

[Lukasz Duda]

Hello RetoKeller,

Thanks for using nRF5 SDK for Thread.

Do you have a Thread sniffer to ensure that CoAP cloud messages are sent over the air?

One way to try debug this problem is to build OpenThread libraries with the log support. In order to do that you need to change OPENTHREAD_CONFIG_LOG_LEVEL define to OT_LOG_LEVEL_DEBG in this file github.com/.../openthread-core-default-config.h . Then you should see all logs on the RTT interface.

Do you use NCP connected over JLink (UART) or native USB? Also you said that reflashing NCP helps. In this situation i guess you rebooted the Border Router, right? Can you try only to reboot the Border Router if this situation happens (without reflashing NCP)? If that helps, we could narrow down the issue to the Border Router itself.

We will try to reproduce this issue locally soon, and we will back to you.