Not sure of the exact cause, but I have a RIO with public tags configured as an IO unit on an EPIC PR1 controller (3.5.1-b.85).
When the RIO is unreachable, I call EnableCommunicationToIoUnit to attempt to bring it online. I believe this is causing the data service on the EPIC to crash and restart - this is happening every 10 seconds or so, which is the interval I am retrying the EnableCommunicationToIoUnit call.
When I remove the logic that calls EnableCommunicationToIoUnit, the data service doesn’t crash.