DRONECAN sensors disconnecting in Flight (MRO Location 1 sensor)

Has anyone seen any instances of DroneCAN sensors just disconnecting and disappearing from the CAN bus for a seemingly random period of time? I am having an issue where my MRO Location 1 gps modules seem to disconnect from the can bus at seemingly random intervals. We will have hours of flight with no issues, then at some point, I start getting messages relating to 1 of my 2 gps units disconnecting from the bus. All data from the device at that nodeID is gone. Once triggered, this effect seems to persist for a random period of time. No traffic from the missing nodeID, not visible via SLcan in dronecan GUI tool or mission planner slcan page. Then at some point it will just randomly come back. I am having a lot of trouble debugging this issue on Arducopter 4.0.5 due to lack of can debugging or can_uc_options, but am currently unable to upgrade at the moment.

My theory is that the device is being perceived as a duplicate CAN node for an arbitrary period of time. When the device finally comes back, it may or may not be assigned a new node ID. The result seems a bit random. I have experienced this issue on both of my sensors on the CAN 2 bus on pixhawk cube blue, but I only experience 1 unit going missing at a time. I have not seen both disconnect, so it does not appear to be an issue with the cables or can bus itself. If I were on a more modern firmware, I might try clearing the DNA database or ignoring conflicting node IDs, but am currently unable due to being on 4.0.5. Is something like that possible on 4.0?

Also if I change the driver instance for the CAN 2 port from the first driver (CAN_D1) to a second driver instance (CAN_D2) by changing CAN_P2_DRIVER for 1 to 2, the device seems to come back, but all sensors have a different HWID which affects compass calibration perception and I expect that after a period of time, I would experience the same issue.

I can not replicate this effect on my bench pixhawk setup with dual Location 1 beacons. Does anyone have any good strategies for debugging can issues on earlier arducopter builds? This issue seems very persistent, stubborn, and random enough to make it really hard to track down.

Any help or advice would be greatly appreciated.

Thanks,
-James Megariotis