We have been trying to debug multicast dropouts. This has occurred on ALL versions of WICED and is currently being experienced on WICED 6.0.1
The devices join the multicast group immediately after going online, the same time they set up a UDP socket to receive CoAP messages.
In the system we are transmitting multicast messages to groups of devices to get responses on various states and to notify them of actions occurring on other devices in the group.
We have noticed at times that the messages are not received by some of the devices. They seem to loose the ability to recv multicast messages.
Apply the following filter to the wireshark capture file: coap&&((ip.addr==10.6.0.8)||(ip.addr==10.6.0.7))
Device .8 sends a notification of an OTA update. The devices do the update and later the devices are sent a request for status. Record 14188
Only two devices respond. The request is repeated by .8. on a couple of occasions. records 14656, 16492. A unitcast request for the same data is sent to one device from .7 and is responded to immediately record: 22982. the multicast request is sent out again record: 27186. The .7 device sends another unicast to another member of the group and it is also responded to immediately. 28268. Some time later ~9 seconds the .8 device sends another multicast request and all devices respond.....
To try to understand this I dug into the source to see what statistics I could pull from the nx drivers. I found the routine. nx_igmp_info_get. When I called this routine the values of all statistics are 0 regardless of the responsiveness of the multi cast action.
This problem only occurs occasionally but is enough to derail this products operation as the notifications that are sent periodically are critical for the operation of other members of the group.
All units have good RSSI levels and acceptable noise levels. Normal operation is correct. Its just this occasional issue we are having.
How to proceed with debugging this issue?
