OTBR and FTD lost on-mesh prefix fd11:22:: address

[taoyinzhou]

I want to make a thread protocol based luminaire product, there will be 300+ luminaires and 10+ OTBRs, and then control them through the cloud platform.
Now I need to verify if so many devices can run stably, I got 100+ 833 modules to do the test and the THREAD are set to FTD.
Then I found some problems during the test.
When I used only one OTBR during the test, if the number of modules was around 40, the network would run relatively stable, even if the OTBR rebooted or 40 modules rebooted at the same time, the connection could be restored normally and seen online in the cloud platform.
However, when the number of modules exceeds 60, there is a high probability that the modules and OTBR will lose the fd11:22 address and cannot recover it, which will cause the modules to be unable to connect to the Internet. OTBR has a way to retrieve the fd11:22 address, the module can't retrieve the fd11:22 address even if it is rebooted. But as soon as I reduce the number of modules to about 40, the module will automatically restore the fd11:22 address.

Attached is my captured packet.

   47FTDOK.pcapng   100+FTD_FAIL.pcapng

Thanks in advance.

Best,

tao

[Edvin]

Hello,

I need to ask a few questions. Since an openthread network can exist just fine without an OTBR (of course not being connected to the internet), what do you use the OTBR for? I understand that you are connecting the devices to the cloud, but are you polling the status of the nodes very frequently? What happens if you slow down the poll rate? Does it allow for more nodes to remain attached? In general, is it a lot of traffic on your network?

Best regards,

Edvin