Friend/LPN time management

[Shahin97]

Hello, 

I have a Friend SoC with nRF52833 via SPI HCI communicating with my nodes that are nRF52833 MCUs. The nodes are LPN nodes sends data to the friend and awaits a reply. Unfortunately the node goes to sleep before receiving the reply from the friend, which I suspect to be due to the SPI HCI latency. I am using rather large payloads for mesh, 200-300 bytes. well within the max segments/ max payload size, Which doesn't help. I am doing a manual poll (bt_mesh_lpn_poll) to the friend to retrieve the missed reply, this seems to be working.

However, some payloads are time critical and so here are my questions:

1 - Is it possible for the lpn node to know how long a message has been queued for upon arrival to the lpn node? 

2 - is it possible for the friend node to remove a payload if it has not been sent within a time frame?

Thank you, 

Shahin

[Elfving]

Hello Shahin,

I just want to clear some things up first. 

The communication between the Friend and LPNs that you are talking about, happens over SPI/HCI? I would then say that this is an OOB communication, in addition to all communication that happens over Mesh. It is not considered a mesh message. Unless there is something I have misunderstood here.

Regards,

Elfving

[Shahin97]

I mean the controller is split. nRF52833 on the friend is running hci_spi (docs.zephyrproject.org/.../README.html)

[Elfving]

So the controller on the '33 Friend node is on a separate device you haven't mentioned this far, right? I guess that is fine. 

I would be surprised if the HCI communication takes too long for this to work. Have you measured how much of a delay the SPI/HCI communication gives you? Are you getting any logs from the Friend or LPN about this? If the LPN didn't get a response then the friendship would get terminated. 

I bet the logs of the devices would be very eye opening here, though you could also just try increasing the timeout values and see if it helps.

Regards,

Elfving